PRAISE

“I love this book—DominicaDeGrandis talks about the chronic problemswe all have in knowledgework andtechnologywork inawaythat isbreezy, familiar,andoften irreverent,butalsoshowsoffdecadesof learningsand concrete techniqueswe canquicklyadopt,bothatworkandathome.Alsowonderfully rewarding iswhenDeGrandisdescribesthetheoryofwhythesepracticeswork,inawaythatisaccessibleandenlightening.”

GENEKIM,researcher,founderofITRevolution,andcoauthorofTheDevOpsHandbookandThePhoenixProject

“Themostpracticalbook I’veseenonmakingprocesses lean.Dominica’sdeepexperiencecoachingcompanies isfullyondisplayasshewalksthereaderthroughaseriesofexercises to findwasteandeliminate it—or, inherterms,tocatchthosesneaky‘timethieves’intheact.ReadthisonaSundayandyou’llwanttostarttryingouttheexercisesonMonday!”

MARKSCHWARTZ, former CIO of US Citizenship and Immigration Services and author of The Art ofBusinessValueandASeatattheTable

“Itisabouttimesomeoneaddressestimetheft(akatheperfectcrime)headon.NotonlydoesDominicaprovidealotof thewhybehindthe forces thatcauseus tomakebaddecisionsaboutour time, shealsoprovides ideasofwhattodoaboutthem.IwishIhadthisbookwhenItookmyfirstmanagementjob!”

JULIAWESTER,LeanConsultantandBloggeratEverydayKanban.com

“Manyofuswearourbusynessasabadgeofhonor.InMakingWorkVisible,DominicaDeGrandisshowsushowwecanmakehiddenwork-in-processvisible,toclearlyseetheeffectithasonourabilitytogetthingsdone.Oncewe can see it, shedivesdeep into thehiddenaspects of ourWIP that steal our time, energy, andproductivity,alongwithstrategiesforcombatingeachofthem.MakingWorkVisiblehelpsustotakeastepbackfromallthatbusynessandreallysee.”

CHRISHEFLEY,ChiefRevenueOfficer,Retrium

MakingWorkVisible

EXPOSINGTIMETHEFTTO

OPTIMIZEWORK&FLOW

DOMINICADEGRANDISFOREWORDBYTONIANNEDEMARIA

ITREVOLUTIONPRESSPORTLAND,OR

25NW23rdPl,Suite6314Portland,OR97210

Copyright©2017byDominicaDeGrandis

Allrightsreserved,forinformationaboutpermissiontoreproduceselectionsfromthisbook,writetoPermissions,ITRevolutionPress,LLC,25NW23rdPl,Suite6314,Portland,OR97210

FirstEditionPrintedintheUnitedStatesofAmerica

10987654321

CoverandinteriorbookdesignbyBelindaBowling,JoyStauber,andRichardWeaver,StauberBrandStudioCoverandinteriorillustrationsbyDominicaDeGrandis

AuthorphotographbyLaurenceG.Cohen

LibraryofCongressControlNumber:2017948984

ISBN:978-1942788157ePubISBN:978-1942788164KindleISBN:978-1942788188

WebPDFISBN:978-1942788171

Publisher’snotetoreaders:Manyoftheideas,quotations,andparaphrasesattributedtodifferentthinkersandindustryleadershereinareexcerptedfrominformalconversations,correspondence,interviews,conferenceround-tables,andotherformsoforalcommunicationthattookplaceoverthedevelopmentandwritingofthisbook.Althoughtheauthorandpublisherhavemadeeveryefforttoensurethattheinformationinthisbookwascorrectatpresstime,theauthorandpublisherdonotassumeand

herebydisclaimanyliabilitytoanypartyforanyloss,damage,ordisruptioncausedbyerrorsoromissions,whethersucherrorsoromissionsresultfromnegligence,accident,oranyothercause.

Forinformationaboutspecialdiscountsforbulkpurchasesorforinformationonbookingauthorsforanevent,pleasevisitourwebsiteatwww.ITRevolution.com.

MAKINGWORKVISIBLE

Idedicatethisbooktomygreatestinspirers:myfourbrilliantchildren,Rachel,Robert,Angelo,andAugustus.Youteachmemoreaboutlifeandjoythananycareeraccomplishmentpossiblycould.

CONTENTS

FOREWORD

INTRODUCTION:WORKANDFLOW

PART1THEFIVETHIEVESOFTIME

1.1TOOMUCHWORK-IN-PROGRESS(WIP)1.2UNKNOWNDEPENDENCIES1.3UNPLANNEDWORK1.4CONFLICTINGPRIORITIES1.5NEGLECTEDWORK

PART2HOWTOEXPOSETIMETHEFTTOOPTIMIZEWORKFLOW

2.1MAKEWORKVISIBLEEXERCISE:DEMANDANALYSIS

EXERCISE:IDENTIFYWORKITEMTYPES/CATEGORIES

EXERCISE:CARDDESIGN

EXERCISE:WORKFLOWMAPPING

2.2AMBUSHTHERINGLEADEREXERCISE:EXPLORETHEFIVEREASONSWHYWETAKEONMOREWIP

2.3EXPOSEDEPENDENCIESEXERCISE:THE“OH,BYTHEWAY”DEPENDENCYMATRIX

2.4COMMITTINGTHEPERFECTCRIME—UNPLANNEDWORKEXERCISE:THEINTERRUPTIONREDUCTIONEXPERIMENT

2.5PRIORITIZE,PRIORITIZE,PRIORITIZEEXERCISE:VISUALIZEPRIORITIES

2.6PREVENTINGNEGLIGENCEEXERCISE:CREATEANAGINGREPORT

2.7USEFULBOARDDESIGNEXAMPLES

PART3METRICS,FEEDBACK,ANDCIRCUMSTANCES

3.1YOURMETRICSORYOURMONEY3.2THETIMETHIEFO’GRAM3.3OPERATIONSREVIEW3.4THEARTOFTHEMEETING3.5BEASTLYPRACTICES

CONCLUSION:CALIBRATION

GLOSSARY

NOTES

ACKNOWLEDGMENTS

LISTOFFIGURES

INTRODUCTION:WORKANDFLOW

Figure1.BuildsDon’tTakeThatLong

PART1:THEFIVETHIEVESOFTIME

1.1TOOMUCHWORK-IN-PROGRESS(WIP)

Figure2.PrepImplementFeedbackBoard

1.2UNKNOWNDEPENDENCIES

Figure3.ThreeDependencyChart

PART2:HOWTOEXPOSETIMETHEFTTOOPTIMIZEWORKFLOW

2.1MAKEWORKVISIBLE

Figure4.VisibilityGrid

Figure5.TheToDo,Doing,DoneBoard

Figure6.BalancedWorkItemTypes

Figure7.WorkItemTypeExample

Figure8.ToDo,Doing,DoneBoardwithColors

Figure9.ExpandedDoingColumn

2.2AMBUSHTHERINGLEADER

Figure10.ExposeWIP

2.3EXPOSEDEPENDENCIES

Figure11.PhysicalDependencyMatrix

Figure12.Arts&CraftsDependencyBoard

Figure13.DependencySwimlaneBoard

Figure14.DependencyTagsonKanbanCards

Figure15.ShowDependenciesBetweenDifferentTeams

Figure16.ExerciseExample

2.4COMMITINGTHEPERFECTCRIME—UNPLANNEDWORK

Figure17.AStudyinInterruptions

Figure18.AStudyinPinkDots

Figure19.ExposeUnplannedWork

Figure20.MonthlyDeltaTrendforUnplannedWork

2.5PRIORITIZE,PRIORITIZE,PRIORITIZE

Figure21.AnExperimentinTaggingandPrioritizing

Figure22.A3Example

Figure23.ExposingConflictingPriorities

Figure24.InputsThatContributetoCostofDelay

Figure25.CostofDelay

Figure26.LineofCommitment

PREVENTINGNEGLIGENCE

2.6PREVENTINGNEGLIGENCE

Figure27.TheValidatePit

Figure28.ExposeNeglectedWork

2.7USEFULBOARDDESIGNEXAMPLES

Figure29.Multi-LevelBoardDesign

Figure30.Donevs.DoneDone

Figure31.Plan-Do-Check-ActBoardDesign

Figure32.HomeProjectBoard

Figure33.ManageYourMoveBoard

Figure34.RepetitiveTasks

Figure35.PurchaseOrderBoardDesign

Figure36.StudentBoard

PART3:METRICS,FEEDBACK,ANDCIRCUMSTANCES

Figure37.TeamsWithinTeamsBoard

3.1YOURMETRICSORYOURMONEY

Figure38.FlowTimeMetrics

Figure39.LeadTimeandCycleTime

Figure40.TheWIPReport

Figure41.QueuingTheory

Figure42.AgingReport

Figure43.FlowEfficiency

Figure44.OptimalBatchSize

3.2 THETIMETHIEFO’GRAM

Figure45.TheOriginalTimeThiefO’Gram

Figure46.CongregatedTimeThiefO’Gram

Figure47.BalancedScorecard

3.3OPERATIONSREVIEW

Figure48.CumulativeFlowDiagramforOpsReview

3.4THEARTOFTHEMEETING

Figure49.LeanCoffeeSetup

3.5BEASTLYPRACTICES

Figure50.IndividuallyNamedSwimlanes

Figure51.T-shapedSkills

Figure52.Specialization

CONCLUSION:CALIBRATION

Figure53.TheJCurve

FOREWORD

Day,n.Aperiodoftwenty-fourhours,mostlymisspent.—AmbroseBierce

Soaboutthatinternetmeme,theoneassuringourfrazzledselvesthateveryonehas the same twenty-four hours in their day as <insert entrepreneurial rockstarhere>.

I’dliketonipthatbitofcondescensioninthebudandofferanemphatic,Notquite.Whilemanyofourbusinessrolemodelsareinfactdrivenbyaseeminglysuperhumanworkethicsupportedby100+hourworkweeks,theyneverthelesshave an advantage over us mere mortals. While the number of minutesavailabletouseachdaymightbethesame,controloverwhatwedowiththosehoursdiffers significantly.WhenElonMusk is facedwith toomuchwork-in-progress(WIP),hehastheauthoritytodelegate,deprioritize,orsimplysayno.Whenvariationrearsitsheadandawell-thought-outstrategicplannolongeralignswiththeorganization’sneeds,SherylSandberghastheabilitytoswitchgears. And when Jeff Bezos is confronted with conflicting priorities, it islikewise doubtful he needs to seek direction via a convoluted bureaucracy togainclarityoverwhichcoursetofollow.

Whenthese thingshappentous (and let’s face it, theyoftendo),we’re facedwith a very different set of repercussions than those of our billionairecounterparts.

Sowhataboutus?Intheabsenceofunbridledagencyandanextensivesupportstaff, how do we do all that needs to get done, when it needs to get done,withoutsacrificingqualityoroursanityintheprocess?Inaculturethatexaltsproductivity and perpetuates the mythology of multitasking, how do wemaximize our time and our workflow to the point that our effort and our

energyyieldsthegreatestimpact?Mostimportantly,howdowedoallofthatandstillhavetimeforliving?

Time saved. Time spent. Time wasted. We frame conversations about timemuch in thewaywedomoney.Ostensibly “free”butnevertheless invaluable,timeisarguablyoneofthemostpreciousresourceswehave,yetoneweneverseemtohaveenoughofasindividuals,asteams,orasorganizations.

Anyone who has ever been faced with a deadline can certainly relate toParkinson’sLaw:workexpandstofillthetimeavailableforitscompletion.Let’sbehonest—when was the last time you completed quality work days or evenhoursaheadofdeadline?

You’renotalone.

It seemswe’re constantly doing, but doingwhat exactly?Why are our weeksfilledwith dayswherewe return home exhausted, only to lament howwe’vebarelymadeadentinourto-dos?Liketheelusivesockthatmysteriouslygoesmissinginthelaundry,wheredothoselosthoursgo?Who—orratherwhat—isresponsibleforstealingourtime,ourfocus,ourenergy?

The attempt to harness or “keep time” is in no way a modern or evenpremodern convention. Prehistoric humans tracked the phases of themoon.The Sumerians created the sexagesimal numeral system still in use today,employing sixty to divide the hour into minutes and then minutes intoseconds.TheEgyptiansusedobeliskstocalculatethelengthofshadowscastbythe sun. The shortcomings of solar-based measures became apparent themomentcloudsappearedor thenightskyarrivedandsowithclepsydras—orwater clocks—the Persians and Greeks offered an alternative, monitoringwaterflowinsteadtomeasurethepassageoftime.

Withtheseancienttime-trackingtoolscametheearliestformsofscheduling—whentoplantcropsandbringintheharvest,whentoconductcommerce,andwhentoperformdailyritualssuchaseatingorsleeping.

Fast forward to today.Despite all ourmodern “conveniences,” effective timemanagement has for many become an uphill battle, an all-consuming if notquixoticgoal.WhiletheInformationEconomyusheredin24/7connectivity,itlikewisebegotround-the-clockexpectations,andso,paradoxically,technologylike mobile phones and email and video conferencing—tools that wouldostensiblymake life easier—often enslave us.We allow the chaos ofmodernwork coupledwith an often paralyzing number of options at our disposal tooverloadus,todistractus,tostealthilystealourtimeandfocusandultimatelyimpedeoureffectiveness.

Wetendtofetishizethecomplex,butjustastheearliestsolutionsfortrackingtimewerebotheasilyimplementedandyieldedextraordinaryresults,sotooaretheideasdetailedinMakingWorkVisible:ExposingTimeThefttoOptimizeWork& Flow. Much in the way sky, sun, sticks, and sand provided ancient manactionable,visualfeedback,sotoodothesuggestionsDominicaoutlinesinthepagesthatfollow.

It should come as no surprise that we can better manage what we can see.Whenwecan’tseeourwork,ouroptionsareobscured.We’reblindtoourowncapacity and we certainly can’t communicate that capacity to others. Theresultant mental overload creates stress. Stress compounds the work wealready have, essentially contributing to WIP, compromising the ability tofocus, prioritize, make decisions, and complete work with quality, let alonecompleteworkatall.

The visualization andWIP-limiting strategies Dominica offers demystify ourcognitive load;normalizeexpectationsamong teammembers;promote focus;situateworkinitscontext,surfacingproblems(andallowingforsolutionstobe

made) in real time; and provide a clear path to completion with quality.Elegantly explained and deeply-insightful, the utility of the suggestionscontainedwithincannotbeoverstated.

To be sure, it is no small irony in that I’m writing about “time theft” whilespending aweek aboard a sailboat exploring an archipelago in the Salish Seabeholdenonlyto“islandtime.”It’sthefirstholidaywhereI’veintentionallyleftmy watch at home, instead choosing to be fully present to the wildlife andseascapesaroundme:baldeaglesandperegrinefalconssoarhighabovecraggybluffswherethepristinecoastlinemeetsold-growthforests.Ottersgracefullypropel themselves through the glassy surf, disrupting kelp and eelgrass insearch of their nextmeal. Along the rockier parts of the shore, scores of sealions loll inthesun,ashauled-outharborsealsnursetheirspottedpupsonasecludedbeachnearby.Acongregationofboatsidlinginthedistanceservesasafamiliarsign,andit’snotlongbeforeItooglimpseafamilyoforcasperformfortheirNikon-wieldingaudience(affectionatelyknownbylocalsasthe“pod-parazzi”)oncrystallinejadewatersthatseeminglyendataceruleanbluesky.

IftherewaseveraplacewhereI’vegivenlessthoughttowatchingtheclock,itisinthisPacificNorthwestjewelboxknownastheSanJuanIslands.

ThisispreciselywhyDominica’sbookissoimportant.Ourcultureofoverwork,our obsession with productivity versus effectiveness, our default mode ofexisting rather than living—these things aren’t simply unnatural andunhealthy, they’re unsustainable—for the individual, for the team, for theorganization’sbottomline.

The thoughtful observations and easily implementable suggestionsDominicaoffersarethefirststepinhelpingcreatehabitsthatleadtoavirtuouscycleofhealthy,sustainable,andimprovablework.Workinwhichweexperiencemoreclarity, less stress, increased focus, improved decision-making, manageable

workloads,and,byextension,amorefulfillingworkday,whichinturnaffordsustheslackneededtofullyliveourlivesratherthansimplychaseproductivity.

So while technically we have the same number of hours in a day as <insertentrepreneurial rockstar here>, it’s creating thoughtful work systems thatmakeuscognizantofwhatweusethosehoursforduringtheworkday.Anditiswhatthosehoursaffordustheopportunitytodowhenweleavetheofficethatmakesforafullyintegratedlife.

Indeed,timeissacred.Treatitassuch.Visualizeyourwork.Limittheamountofworkyoutakeon.Payattentiontoitsflow.Buildthoughtfulworksystemstoreflectwhatreallymatters.

Tobreathe.Tothink.Tolearn.Togrow.Toplay.Tolove.Tolive.

For it is in working well that we can live well. I am confident the wisdomDominica offers in the pages that follow is the first step to achieving suchexistencesothatyoutoocanbegintoexperiencelessstolentimeandinsteadplanformoreislandtime.

TonianneDeMariaOrcasIsland,Washington

INTRODUCTION:WORKANDFLOW

Donotsquandertimeforthatisthestufflifeismadeof.—BenjaminFranklin

Asabuildengineer,myfirst joboutofcollegewastomakebuildsvisible.Thismeanttrackingwhichversionof what file went on which computer and in whatenvironment. Three months into that job, I wasworking on a build—getting all the code from thesource code repository, compiling it into executables,and then installing the resulting new functionalityintoaplacewhereothers(analysts,developers,testers,and other interested people) could see it. The buildwasn’t compiling though, and I sat there

troubleshootingthebrokenbuild,intheoffice,alone,at2:00a.m.Tired,Iwasmaking mistakes, so I went home. I seriously questioned my career choice.Technologywork apparentlymeantworkingmany late nights. After a nap, Ireturned to the office to track down code dependencies between variousdevelopersandeventuallygotthebuildworking.

I’mnot sure exactly howmanyhours I’ve spent tracking downdependenciesduring all my years of merging, building, and releasing software, but I’mconvincedthatit’sbeenwaytoomany.IfIhadadollarforeveryminutespenttroubleshooting builds and broken environments, Iwould have a sweet littlenestegg.Delayedwork,whetheritismeasuredinhours,days,weeks,orevenmonths, carries with it a cost. Losing time due to avoidable problems isexpensiveanddispiriting.Lifeisshort.Wastedtimecanneverberegained.

Inthesci-fifilmInTime,timeisliterallymoney—peopleearn minutes, hours, and days to buy food, housing,transportation, and everything else imaginable. Streetthugs kill people by stealing all their minutes. Wastedtime is thekiss of death. Inonememorable scene,WillSalas,playedbyJustinTimberlake, saves the lifeof thewealthyHenryHamilton, played byMattBomer.WhenWillandHenrygettoasafeplace,HenrytellsWill thathe is 105 years old and tired of living.He asks twenty-eight-year-oldWillwhathewoulddowith100years.Willquips back, “I sure as hell wouldn’t waste it.” Later, asWillsleeps,HenrygivesWillhis100yearsandleaveshimanote, “Don’twastemy time,” beforehe runsoff to timeoutby allowinghis

ownclocktorundownwhilesittingontheledgeofatallbridge.1

Ascrawlednotefromadystopiansci-fifilmembodiesourreality.Timeislife—usetimewisely.

Weworkersaredrowninginnonstoprequestsforourtime.FromdeveloperstoIT operations people, it’s overwhelming to keep up with the ever increasingdemand.Inthisregard,thingshaven’tchangedmuchsincemyfirstjoboutof

college, where as a software configurationmanagement lead at Boeing, I didbuilds and deployments on IBM mainframes at Hickam Air Force Base inHawaii.

Alineofpeopleformedoutsidemycubewantingtoknowthestatusofabuild.Did everything compile okay?Whenwill thebuildbedeployed to thequalityassuranceenvironment?CanIgetonelastchangein?Iwantedtosay,“Pickanumber.I’mworkingasfastasIcan.Everyoneofyourinterruptionsdelaysthebuildbyanothertenminutes.”ThefactthatdevelopersandtestershadcometometoaskforstatusupdateswasasymptomofamuchlargerproblemthatIdidn’trecognizeatthetime.

My calendar was booked with meetings all daylong. Rarely did I get a chance to workuninterrupteduntiltheeveningortheweekend.Four months into the new job, I pulled an all-nighter in theoffice,workingas fastaspossibletocatchuponthemountainofwork.Whentheprogram manager arrived at 6:30 the nextmorning,hethoughtIhadjustarrivedearly.HewasnotpleasedtohearthatIwasheadedhometotakeanap.SleepdeprivationwasanotherredflagthatIdidn’tgiveenoughthought to at the time. Later on, after years spent working in technology, Irecognizedthatrelentlessheroism—stayinglatenightafternight,wearingtwohats, and consistently playing catch-up—is unsustainable. Quality doesn’thappenonfourhoursofsleep.

Weoverloadourselvesandweoverloadourteams—thisistheeverydayrealitywithintheinformationtechnologysector.And,becausewegetinterruptedallthetime,westopworkononetaskandstartworkonadifferenttask,fromoneproject to thenext,never focusingonone thing longenough todo it justice.

This context switching kills our ability to settle into work and concentratesufficiently.Asa result,weareunhappywith thequalityofourworkdespiteourdesireforittobegood.

Theproblemisthatweareworkingwithdysfunctionalprocesses—companieshaven’tadaptedtokeepupwithdemandinahealthy,sustainableway.Instead,weseethecontinueduseofantiquatedapproachesmeanttokeepworkersbusyallthetime.Theseprocessesarenotworking.Thisistheelephantintheoffice.Ifworkerswereabletogeteverythingdonerightandontime,therewouldn’tbe an issue. But that’s about as common as a black swan. The amount ofrequests (the demand) and the amount of time people have to handle therequests(theircapacity)isalmostalwaysunbalanced.Thisiswhyweneedapullsystem—inwhichpeoplecanfocusononethinglongenoughtofinishitbeforestartingsomethingnew—likekanban.Kanbanisavisualpullsystembasedonconstraintsthatallowworkerstopullworkwhentheyhaveavailabilityinsteadof work being pushed onto them regardless of their current workload. Sincedemandandcapacityarefrequentlyunbalanced,andit’salmost impossibletogeteverythingdoneontime,systemslikekanbanareforhelpingpeoplebalancealltheirworkdemand.

We’llgetintokanbanandwhereitfitsintotheprocessofmakingworkvisibleabit later, but for now, know that kanban is an approach to make work andproblemsvisibleandimproveworkflowefficiency.Kanbanhelpsyougetworkdoneefficientlywithoutburningthemidnightoilnightafternight.

Inthe2000s,IworkedatanimagelicensingcompanyinSeattleownedbyBillGatescalledCorbis.ImanagedtheBuildandConfigurationManagementteam.

We had a decent reputation among the Engineering department until 2005,when our two preproduction, seven-server environments quad-rupled intoeightpreprod,twenty-five-serverenvironments.Wehadseventeendatabases.Each configured manually within the tightly coupled, highly dependentarchitecture. On top of that, the business asked us to develop new majorsystems at the same time, and they wanted the ability to deploy either onebefore theother.Thedependenciesbetween theexisting systemand the twonewsystemsballooned.Myjobgrewfrombuildingoutandmanagingtwenty-fiveserverstobuildingoutandmanagingtwohundredservers.

To deal with the changes, we created and maintained additional long livingbranches in source control,which is theplacewheredevelopers check in theircodeforsafekeeping.Itwasaterriblesolution,butithelpedtheteamsavoidclobberingeachother’schanges.Thinkoflonglivingbranchesasaplacewherecodeisstoredinisolation,whereit’simpossibletoseetheimpactitmighthaveonthecodealreadyreleasedtoproduction. It’skindof likeadoptinganoldercat and praying that he and your current,much older cat will embrace eachotherwithopenpaws.Withmorethantwohundredserverstoconfigureandmaintain,configurationmanagementwaselevated.Ittook,atbest,twoweeks

torestoreproductiondatatopreproductionenvironments.Wescheduled“MIsforMerge”dayseverysixweeks,whichconsumedmanydevelopers’time.

Our reputation plunged. Developers complained that builds were taking toolong. This, of course, offended me, and I set off to prove them wrong bycollectingbuild-and-deploytimemetrics.

Figure1.BuildsDon’tTakeThatLong

I pointed out that the big ball-of-mud architectural design disaster on ourhands made deployment and maintenance of environments problematic. Ipointedout that themanual smoke tests (tests to see ifwebsite functionalitystill works) delayed the time that developers and testers could see the latestchanges and that lack of automated testing hurt our ability to quickly spotproblems.Manual smoke tests were the norm. Both of these problemsweredismissed fairly quickly as not real issues. The fact remained that developersand testerswere unhappy. Business peoplewere unhappy. And the bosswasunhappy. It’s no fun being on the team that “doesn’t deliver.” The barriersbetweenteamswerestrongerthantheconnections.Thisistheproblemwitha

badsystem.

My own experience with a bad system coincided with the CFO deciding toreplacetheenterpriseresourceplanning(ERP)systemwithanotherERPproductcalled SAP. An ERP system is a management information system whichintegrates things like planning, purchasing, inventory, sales, marketing,finance, andHR. SAP is its own ERP system, created by SAPAG, the fourthlargestsoftwarecompanyintheworld.

Mybossaskedme,“Hey,doyouwanttomanagetheSAPBasisteamaspartofmanaging thebuild and release team?”Likean idiot, I saidyes. Idon’tknowhowIcouldhavepossiblysetmyselfupformorefailure.IhadzeroexperiencewithSAP,andaddingSAPtomylistofresponsibilitiesspreadmethin—tothepointwhereImanagedtobeterribleatmanydifferentjobs.Multitaskingisagoodwaytoscrewupprogress,asI’msuremanyofyoureadingthisbookknowfromexperience.

Atthetime,Ididn’tknowthatallthesethingswereredflagsofabadsystem.AllIsawwasthatmyperformancewaslessthanexemplaryandthatIwasanunhappyemployeewhohadstartedtoconsiderotheroptions.

Iupdatedmyresume.

In2006,wespentagooddealoftimeanalyzingandcomparingdifferenttoolstomanageoursourcecode.OurteamchoseTeamFoundationServer(TFS).Wewere aMicrosoft shop, after all, and I ended up installing, configuring, andmaintaining TFS—while also learning SAP, interviewing new candidatesweekly, and helping to implement a new sustainment process. This processmade it possible for us to deliver improvements every two weeks instead ofeverysixmonths.

Auserinterface(UI)developernamedDwayneJohnsonrecognizedthevaluein

delivering small changes frequently and began socializing the idea ofmakingsmall improvementsonaconsistentschedule.DwaynestartedtheprocessbyfixingUI bugs on a regular bi-monthly cadence. At the time, it was just onemorethingtosupport,butitwasaveryimportantone.TheseincrementalanditerativeimprovementsdoneonaregularcadencewasourAgilealternativetotraditional Waterfall development. These Agile methods wandered into ourprocess,gettingusthinkingaboutabetterapproachtoourwork.

In April of ’06, a Scottish Fellow fromMicrosoft appeared at Corbis. DavidAndersonvisitedusmonthlytoteachushowtoapplytheTheoryofConstraints(TOC)toourworkinexchangeforpermissiontowriteastoryabouttheCorbisAgile transformation. TOC is a way to identify the most important limitingfactor (the constraint) that stands in the way of achieving a goal and thensystematically improving that constraint until it is no longer the limitingfactor.TherewasmuchexcitementwhilereadinghisbookAgileManagementforSoftwareEngineering:ApplyingtheTheoryofConstraintsforBusinessResultsaswethoughtwewoulddoFeatureDrivenDevelopment,atypeofAgiledevelopmentfocused on cross-functional, collaborative, and time-boxed activities to buildfeatures. AsDarrenDaviswrites in his blog “The SecretHistory of Kanban,”David’smethods“...eliminatedexplicitestimationfromtheprocess,andreliedon data to provide a probabilistic means of determining when software was

likelytobedone.”2Davidgotusgoingonoperationsreviewsandexplainedhowimportantitistomeasureprogress(orlackthereof).Learningwhattomeasurechangedmyworld.Rantingdidn’twork,butmeasuringcycle time (the time ittakes to do work) and presenting that data to leadership, did. I was able toinfluenceleadershipthenandgotbuy-intohireadditionalteammembers.

Sometimes the obvious gets lost in the crunch of the corporate world. Weintuitivelyknewwehadtoomanyprojectsinflight,butitwashardtoseeuntilwemeasured theactual time that it took togetworkdone,atwhichpoint itbecameobvious theworkspentmore time inwait states than inworkstates.

Wespenttimewaitingforapproval.Waitingforotherstofinishtheirpartsowecouldstart(orfinish)ourpart.Waitingforuninterruptedtimetofocusonfinishingthework.Waitingfortherighttimeofday/week/month.Andwhilewe waited, we started something new, because, you know, with resourceutilizationasagoal,youhavetostaybusyallthetime.

AsKateMurphywrites inherarticle“NoTimetoThink,”“Oneofthebiggestcomplaints in modern society is being overscheduled, overcommitted andoverextended. Ask people at a social gathering how they are and the stockanswer is ‘super busy,’ ‘crazy busy’ or ‘insanely busy.’ Nobody is just ‘fine’

anymore.”3 I seeevidenceof thiseveryday.Whenthere isastillmoment forreflective thought—say, while waiting for a meeting to begin—out comepeople’sphones.Busynesscanbeanaddictionforterminallywiredambitiouspeople. But busyness does not equate to growth or improvement or value.Busynessoftenmeansjustdoingsomanythingsatoncethattheyallturnoutcrappy.Sometimeswalkingintheparkandallowingourselvestimetothinkisthe bestway to seize the day. But horrors if an engineer sits idle for fifteenminutessimplythinking.

AtCorbis, lookingatthereasonswhyweworkedontoomanythingsatoncewasarevealingexercise.TheCFOwantedtoimplementanewfinancialsystem.The SVP of Global Marketing wanted to blah, blah, blah. The VP of MediaServicesalsowantedblah,blah,blah.TheheadofSaleswantedblah,blah,blah,blah. And they all wanted everything now. The resulting business prioritiesclashedall thewaydownthehierarchyandthatwas justthebusinesssideofthehouse.Ontheengineeringside,notonlydidweneedtoimplementallthebusiness requests, we also had our own internal improvements tomake andmaintenance work to do. Furthermore, we still had to be available to dropeverythingwhenproductionissuesoccurred—likeitornot,productioncomesfirst.Theclashingprioritiesbecameapparentwhilelookingatthemanylong-standingbranchedcodelines,butotherthanthat,therewasnoclearvisualof

theimpactofworkingontoomanythingsatonce.It’shardtomanageinvisiblework. With invisible work, we don’t notice the explicit reminders that ourmentalbudgetisalreadyfull.Thereisnotimetosimplythink.

After eight years at Corbis, I was one of forty-two people let go during theSeptember 2008 round of layoffs. At this point, I decided to try somethingdifferent. Igota jobwithAT&TMobileontheirprogrammanagementteam.But the regression from using the Lean kanban approach I helped create atCorbis to using a Waterfall approach (a traditional software developmentmethod where work waits until all the parts of the previous stage arecomplete), with estimations based off of time reports, was too much of athrowbackforme.InJuly2010,Ifiredmyself.

In January 2011, David Anderson offered me the opportunity to research,develop, and teach a new course for David J. Anderson & Associates calledKanbanforITOperations.Atthetime,EuropeledtheUnitedStatesinkanbanimplementations, somy research inFebruarybegan inEngland,Sweden, andGermany.InMarch,weranthefirstbetaworkshopinBoston,whereIattendedandspokeatDevOpsDaysBoston2011attheMicrosoftNewEnglandResearchandDevelopmentCenterinBoston.

Originally, Isetofftowriteareferenceforstudentstouseduringworkshopswhiledesigningtheirkanbanboards.Later,thispiecegrewintoatime-savingreference forme as well. It became a place to capture not only everything IlearnedaboutapplyingLean,kanban,andflowpracticestomyownwork,butalsoselectedequations,theories,andstatsfromthoughtleaders.Forexample,how to define Lean? For that, I prefer Niklas Modig and Pär Åhlström’sdefinition.IntheirfantasticbookThisIsLean:ResolvingtheEfficiencyParadox,theydefineLeanas,“astrategyofflowefficiencywithkeyprinciplesofjust-in-

timeandvisualmanagement.”4

So,whatdoweknow?Weknow thedemand fordeliveringbusinessvalue to

production, so that we can be competitive, is high. We know that manyorganizations are running deployment strategies that are slow andcumbersome. We also know that we are wired to do our best when we canclearly seewhatwe aredoing right aswell aswhatwe aredoingwrong.Thismightseemobvious,butit’sconsistentlyignored.

Thetechnologyworldshowsnosignsofslowingdown.Thepaceatwhichweneed to deliver new capabilities to win new customers and prevent existingcustomersfromwalkingaway(churn)seemslikewarpspeed.Manycompaniestodayare insurvivalmode, they justcan’tsee it.Thismeansthat there isnobettertimethanrightnowtoelevatehowwework.So,howdowelevelupourgame?

Theanswerisstraightforwardandaccessible.Itdoesn’tcostyoutonsofmoney,and it doesn’t take geniuses or specialists. All it takes is a shift fromhaphazardlysayingyestoeverythingtodeliberatelysayingyestoonlythemostimportantthingatthattime.Andtodoitvisually.

The solution is to design anduse aworkflow system that does the followingfivethings:

1. Makeworkvisible.2. Limitwork-in-progress(WIP).3. Measureandmanagetheflowofwork.4. Prioritizeeffectively(thisonemaybeachallenge,butstaywithme—I’llshowyouhow).5. Makeadjustmentsbasedonlearningsfromfeedbackandmetrics.

Whatwewillcoverinthisbook:

Howtospotthefivethievesthatstealyourtime.Howtoexposethetimethievesinordertomakeworkvisibleandoptimizeworkflow.Howtoknowhowyouarereallydoingusingmetricsandfeedback.Whatpracticescangetyouintotrouble.Howtoinfluenceleadershipdecisions.

Theexamplesdescribedthroughoutthepagesofthisbookareallbasedonmyownreal-lifeexperiencesandonthoseofotherswhohavestoodaswitnessesto

time-theft scenarios. Some prefer to avoid publicizing the crimes committedwithintheircompanies,soforthem,thenameshavebeenchangedtoprotectboth the innocent and the guilty. We will also be looking at systemicorganizationalissuesthatmustbeaddressedinorderforyoutobesuccessful.

AsEdwardsDemingsaid,“Abadsystemwillbeatagoodpersoneverytime.”5

This book is simultaneously an explanation, a how-to guide, and a businessjustification for using Lean, kanban, and flowmethods to increase the speedandeffectivenessofwork.

Everything in thisbookmaynotapply toyourspecific situation. Ithasan ITbent to it, with several non-IT examples thrown in for good measure. Takewhat does apply to you and use the rest to gain insight intowhat people inotherpartsofyourorganization,oryourcompetitors,mightbedealingwith.Each section inPart 2 includes exercises frommyworkshops,wherewe stepthroughaseriesofactivitiesdesignedtomakeworkvisible,improveworkflowefficiency, and surface problems.Theybuild upon eachother, so it is best toreadthesectionsinsequentialorder.

Explaining the concepts in this book to others should be a straightforwardprocess. Getting buy-in to implement the suggested approachesmay not be.Change is hard for humans. So, before we dive into workflow design, let’sinvestigateexactlywhatpreventsyoufromgettingyourworkdonequickly inthefirstplace.Oncewescrutinizethecrimescommittedagainstyourexistingworkload, we can proceed with the insight and awareness necessary to dosomethingaboutit.Let’sgetstarted.

Bewarethebarrennessofabusylife.—Socrates

1.1

TOOMUCHWORK-IN-PROGRESS(WIP)Ontheroofofabuilding,Saturday,9:00a.m.

Amanundertakesanitemonhishoney-do list (his to-do list, heavily influencedby his spouse), which includesdismantling the roof of an outbuilding.Over the years, thisman’s list of thingsto do has included repairing everythingfromappliancestosepticsystems.Hehasburiedpowerlines,felledninety-foot-tallcedar trees,andbuilt cabinsandgaragesstep-by-step from excavatingfoundations to installing flooring,heating, plumbing, electricity, androofing.

Recently, he seismically retrofitted anunreinforced, hollow-tile foundation. Iam this man’s assistant, which includes(but isbynomeans limitedto)actingastape measure holder, safety inspector,

anddemolition and cleanup crew.Oneday,whilehelpinghim teardown theraftersofanold,crumblingtwenty-four-by-thirty-sixfootoutbuilding(Iontheground,heontheroof),Icasuallysuggestedthatwebuildasixteen-by-twenty-four-footgreenhouseonthebackforty.Fromthetopofthetwenty-five-foot-tall rotting roof, my beloved husband looked at me incredulously and said,“Hon,can’tyouseeI’mbusyuphere?!”

Thetechsectordoesnothaveamonopolyontoomuchworktodo.Talentedpeopleeverywherereceive longto-do lists.Theproblemforaspousewhocanbuildorfixanythingisthattheotherspouseprovidesthemwithalonglistofthingstodo.Andit’shardforthemtosayno(unlesstheyareatopatwenty-five-foottallrottingroof).

Wehumanshaveahardtimesayingnoforavarietyofreasons.Reasonnumberone is becausewe like the personwho asked us. The same holds true at theoffice. Because network engineer Sean givesme a heads up onwork comingdownthepipethatimpactsme,Isayhe’sniceandamwillingtohelphimoutwhenheneedssomething.ButCarlos!Carlosknewaboutthisportchangetwoweeks ago and is only just now tellingme at 5 p.m. on Friday?!Mymentalnarrativesays,“Idon’treallywanttohelpyou.”

There are five other main reasons peoplegivewhenIaskthem,“Whydoyoutakeonmore work than you have the capacity todo?”

1. Weareteamplayers—“Idon’twanttobethepersonwholetstheteamdown.”

2. We fear humiliation—“I don’t want to be criticized orfired.”Yes iseasiertosaythanno—especiallytotheboss.Refusingamanager’srequestcanberiskyinsomecultures.

3. We like new and shiny—It’s much more fun than doing the grunt work it takes to finishsomethingcomplicatedandunglamorous.

4. Wedon’trealizehowbigtherequestisuntilwestartworkingonit—“Oh,noproblem.Icangetthatdoneinjustacoupleofhours,”butthetasktakesmuchlonger.

5. We like to please people.—“I say yes tomost requests in general because I want to be liked,admired,respected.“

Vanessa Bohns, social psychologist and professor ofmanagement sciences attheUniversityofWaterlooinOntariosays,“Itcomesdowntothisfundamentalmotivationwehavetostayconnectedtootherpeople.Wedon’twanttorejectpeople.Wedon’twantpeopletothinkpoorlyofus…sowearereallymanaging

theimpressionsotherpeoplehaveofus.”1Ontheflipside,werarelyrealizethepowerwehaveoverotherpeoplewhenweaskthemtodosomething,especiallyifothersworryaboutexplicitorperceivedpositionsofpower.

In textbook terminology, too much work-in-progress (WIP) is when thedemandontheteamexceedsthecapacityoftheteam—whichisaratherboringwaytosaythatourteamsaredrowninginwork,oftenbecausetheirscheduleiscompletelyfull.Everyminuteofthedayisfullyscheduled(orfullyallocatedto100% resource utilization). The most talented have the longest lists. Thisequates to people doing their full-time job on top of everything else that isexpected of them, such as troubleshooting environment issues (problemswiththe configuration of servers that prevent website functionality and otherthings fromworking right),hiringnewteammembers, andcompletingmeritreviews,tonamejustafew.Similartohowourdigestionsystemletsusknow

whenwe’vestuffedtoomuchfooddownit,ThiefTooMuchWIPattacksusifwecramtoomanymeetingsintoourday,leavingusunabletobegintheday’sto-dolistuntil6:00p.m.

WhyTooMuchWIPMatters

ToomuchWIPmattersforanumberofreasons.Itcanresult inmanyissues,including delayed delivery of value, increased costs, decreased quality,conflictingpriorities,and irritablestaff, tonamea few.Whenwestartanewtaskbeforefinishinganoldertask,ourWIPgoesupandthingstakelongertodo. So does the potential loss from things taking too long to complete andbeingunabletorealizethevalueofitsooner.Wemeasurethiswithcycletime.Cycle time is the amountof elapsed time that awork item spends aswork-in-progress.Inaddition,businessvaluethatcouldhavebeenrealizedsoonergetsdelayedbecauseoftoomuchWIP.Thisisknownascostofdelay.It’saconceptusedtocommunicatevalueandurgency—ameasureoftheimpactoftimeonthe outcomes we want, such as customers buying our product this monthinsteadofnextmonth.

When you delay the delivery of a new feature because another request gotbundled in with it, there’s a cost for the delay of that new feature. It could

mean late feedback, lessprofit, or amissed sales leadopportunity.Yournewfeaturegetshijackedonitswaytoyourcustomer,andthemorestuffyouadd,thelongerthecustomerwaits.Ifcustomerswaittoolong,theyshopelsewhere.Oncecustomersgiveupandmoveon,youlose.Maybeitwasworthit—butdoyoureallyknow?

Ingeneralandforthepurposesofthisbook,Idefinecustomersintwoflavors:

External customers: Peopleoutside yourorganizationwhobuyoruse yourproductorservice.Iftheymoveontogreenerpastures,youloserevenue—andyouriska less-than-favorablereviewonyourcompany’sFacebookorAmazonpage.

Internal customers: People inside your organization who ask you to dosomething or who consume your work. A Product Development team is acustomerofthesecurityengineerwhodetectsvulnerabilitiesintheproductortheplatformitsitson.Anemployeeisacustomerofthemanagerwhoprovidesfeedback.InternalcustomersimpactWIP.Forexample,HelpDeskWIPgrowswhentheaccountingadmingets lockedoutofhiscomputer.MarketingteamWIP growswhen the technical evangelist adds a new conference to her tour.PlatformOperationsWIPgrowswhen theVPofwhateverhires a third-partyvendortobuildanewintegration.

WIPisaleadingindicatorofcycletime.Themoreitemsthatareworkedonatthe same time, the more doors open up that allow dependencies andinterruptionstocreepin.Trailingorlaggingindicatorsarebackwardfocused—theymeasure performance data already captured.Mostmetricsmeasured intechnology and business, such as lead time (the elapsed time it takes tocomplete a request from the time it was first requested), cycle time, andthroughput(thenumberofthingscompletedoveraperiodoftime),aretrailingindicators.Thatis,wedon’tknowhowlongcertainthingswilltakesomethinguntilthosethingsarecompleted.

ThereisarelationshipbetweentheamountofWIPandcycletime—it’scalledLittle’sLaw,wheretheaveragecycletimeforfinishingtasksiscalculatedastheratiobetweenWIPandthroughput.WIPisaprimaryfactorintheequation.It’sobviouswhenyouthinkaboutit—assoonasyougetonacloggedfreewayyouknow that your commute is going to take longer. For this reason, Thief TooMuchWIPistheringleaderofalltheotherthieves.

YouknowThiefToo-Much-WIPisstealingtimefromyouwhen:

Contextswitchingiscommon:Whencomputerscontextswitch,thestateoftheprocesscurrentlybeingexecutedissavedsothatwhenitisrescheduled,thestatecanberestoredtoitscorrectspot.Becausecomputersperformhundredsof context switches per second, it’s easy to believe that multiple tasks areperformed in parallel, when in reality the central processing unit (CPU) isactually alternating or rotating between tasks at high speed. As ToddWattswrites in his blog post “Addressing the Detrimental Effects of ContextSwitchingwithDevOps,”theoverheadincurredbyacontextswitch,managingthe process of storing and restoring the state, negatively impacts operating

system (OS) and application performance.2 Because a context switch can

involve changing a large amount of data, it can be one of the most costly

operationsinanOS.3

Just likecomputers,humansincuroverheadwhencontextswitchingbetweendifferent tasks. Although with humans, the overhead is much higher. DatastructurescontainingalltheinformationregistersandOS-specificdata,alongwith the exact point of entry for where to resume the process, aren’tautomaticallyrescheduledinthebrainliketheyareinaCPU.Contextswitchingincomputershasaprogrammableflowtoit.

Thenotionof flow inhumansdoesn’t happenwhen context switching is thenorm. Flow is the concept of focused motivation. It’s charac-terized bycompleteabsorption inwhatonedoes (energized focus). It’sanoptimalstatethatresultsinhighlevelsofproductivityandsatisfaction.Toachieveflowistobeinthezone—thatspacewhereintrinsicmotivationandcreativityflourish.

Toachieveflow,afocusedconcentrationonthetaskathandisnecessary.Thisdoesn’t happen when distractions, whether in the form of email, food,coworkers,orsocialmedia,interruptus.Whenweareresponsibleformultiplethings, such asmaintainingproduction anddeliveringnew features,we shiftwhat we’re doing based on perceived priorities. By the time we get back toworkingonwhateverweweredoingbeforethe interruption,wehavetostartalloveragain.Flowrequiresa“donotdisturb”ethos.

Yourcustomerswaitforlongperiodsoftime:Flowalsorequiresalevelofefficiency.When it comes to flow efficiency, the lengthof timeyoukeepyourcustomerwaiting isofprimeconsideration. Ifnewprojectsarestartedbeforeexistingprojectsare finished,workpilesup, requiringmore resourcesand/ormorepeople.Itisinefficientfromacustomerperspectivetoprioritizestartingnewworkoverfinishingthethingsyouhavealreadybegun.IfI’mwritingablogaboutkanbanandthenextstepintheprocessistohaveiteditedbysomeoneon the Marketing team, then beginning a new blog about DevOps before

incorporatingMarketing’seditsforthekanbanblogmeansI’llhavetodealwithacontextswitchwhentheeditorgetsbacktome.

Quality suffers: Quality suffers from toomuchWIP.When Iwas at Corbisand took on the additional role ofmanaging thenewSAP basis team, I shotmyself in the foot. I had to learn a complicated mainframe product whilebuildinganewteamontopofcontinuingtodomyoriginaljob.Ihadn’tdoneanything with mainframes since my first job out of college seventeen yearsbefore,andIdidn’tknowanythingaboutSAP.Ididn’ttakethetimetolearnitverywellbecauseIhadalltheseotherthingsthathadtobedoneonmyplate.Theresultwasapredictableone inretrospect:Neither the teamnorSAPnormyotherresponsibilitiesgotadequateattention.Thisledtoapoorlymanagedteamandanirritatedme.

Irritatedstaff:Contextswitchingisirritating—you’rerarelyleftwithenoughtime to do a good job, nor with sufficient space tomaster the task or skill.HarryF.Harlow, anAmericanpsychologist, says inDanielPink’s bookDrive,“Thejoyisinthepursuitmorethantherealization.Intheend,masteryattracts

preciselybecausemasteryeludes.”4Masteryeludesbecausethereisinsufficienttime to pursue something long enough and deep enough before beinginterrupted.

Interruptionsthwartdeepthinking.SherlockHolmesthinksbestwhenhegoestohis “mindpalace” in theBBCadaptationofConanDoyle’s famous sleuth’s

escapades.5 Using a mental technique called the method of loci (Latin forlocation),hetravelstohismemorybank,asortofmentalmapwherememoriesare deposited, to withdrawmemories. But he needs an environment void ofdistractionsandinterruptions,andhegetsquitecrankyifothersinterferewithhis mind palace. And with good cause—it’s downright irritating to beinterruptedwhendeepinthought.Timethieveslovetheareaofdeepthoughtbecause, as David Rock relates in his book Your Brain at Work: Strategies for

OvercomingDistraction,RegainingFocus,andWorkingSmarterAllDayLong,itcantake up to twenty minutes to get back to that same thinking spot after an

interruption.6

Someoneasksyouifyouhavefiveminutesandyousayyes:Alltoooften,whensomeoneasksyou,“Doyouhavefiveminutes?”andyousayyes,youendup working late. It’s annoying and sometimes exhausting, and we do it toourselves. Even though I teach this stuff, I fall into this trap too. In our

defense, we do get more endorphins from saying yes7—enough that evengrouchypeoplewanttosayyes.However,it’snotproductivetosayyesallthetime, nor is it sustainable to consistently work evenings and weekends. Thepowerofkanbangivesyouthefreedomtofinishwork.

Kanban is Japanese for signal card—a card that, very simply, signals youravailabilitytodosomework.Whenyoupullacardfromthebacklogontothein-progressareaofyourkanbanboard,youcommittobeingavailabletodotheworkthatthecardrepresents.

Figure2.PrepImplementFeedbackBoard

Thenumberofcardsunderthein-progressarearevealstheamountofWIPonthekanbanboard.Theboard inFigure2showsaWIPoffour.WIP limitsarewhatmakekanbanapull system.Whenacard is finished, it signalsavailablecapacity and causes another card to be pulled into In Progress. Work flows

acrosstheboardbasedontheWIPlimitsandpullpolicies.IfWIPlimitsaresetappropriately, the system cannot become overloaded. TheWIP limit is whatallowsyoutosay, “No, there isnocapacitytotakeonmoreworkrightnow.”Thinkof reducingWIPnotas limitingbutas liberating.The rightamountofWIPiswhatenablesustomaintainahealthyamountofwork.

When you say to your pal, “Yes, I’ll do that,” you have just authorized andprioritized “that” request over all the other requests in the backlog. DanWeatbrook, Tableau WebOps Manager, calls this “born-in-doing”—a way of

cutting in line, if you will.8 It’s a thief stealing time away from previousrequests and is one reason why requests in the backlog take so long (andsometimesnevermakeit)totheInProgressstate.

TheseelementsoftoomuchWIPcropupacrossallthethieves.ThiefTooMuchWIP is the ringleader, and the other thieves steal ideas from this relentlesstroublemaker.Wewill get intomoredetailsonhowthe thieves interactwitheach other a bit later. For now, let’s take stock ofwhatwe’ve covered so farabout our thieving ringleader and move on to thief number two, ThiefUnknownDependencies.

KEYTAKEAWAYS

Wehaveatendencytosayyestoanyrequest,regardlessofhowbusyweare.ToomuchWIPpreventsusfromcompletingworkontime,causesqualitytosuffer,increases

costs,andirritatesstaff.Work-in-progressandcycle timehavea relationship.HighWIPmeans thatother itemssitidle,waitingforattentionlonger.Contextswitching,whichwastestime,isamajorconsequenceoftoomuchWIP.Wemustlearntosaynotoadditionalworkwhenourschedulesarefull.

Thedefinitionoffreedomisthatthereisnodependency.—DadaBhagwan

1.2

UNKNOWNDEPENDENCIESAfriendofmineworksforacompanywith$23billioninyearlyrevenue,whereProduct Team X deployed a component that broke Team Y’s product. Now,TeamY’scustomershavetoforkout$5millionforthenewYpart.Thisisontopofthe$10millionthattheyjustpaidtobuyanewXpartbecausetheoldXpartwasattheendof its lifeandwasno longersupported.Thecustomers inthisscenariousedpartscreatedbyboththeXandYteams.PartYneededpartXtofunctioncorrectly.TheonlywayTeamY’scustomerscouldgettheirneedsmetwastobuythenewXpart.

Nowthis companyhasamajorpublic relationsdisasteron theirhands.Theyarelosingsignificantmarketsharebecausethetwoproductteamsdidn’ttalktoeachother.TeamYhadzerovisibilityintoTeamX’sdecisiontoreleaseanewversionofsoftwarethatTeamY’sproductwasdependenton.Theblamegame,repletewith finger pointing, begins and now aVP’s head is on the choppingblock. It’s very expensive when teams are unaware of mutually criticalinformation.This is the typeof thing thathappenswhen thereareunknowndependencies.

Let’s define dependency. From my perspective, when we talk aboutdependencies,threetypesemerge:

1. Architecture(bothsoftwareandhardware)—wherechangeinoneareacanbreakanotherarea(i.e.,causeittostopfunctioning)

2. Expertise—wherecounseloraidfromapersonwithspecificknow-howisneededtodosomething3. Activity—Whereprogresscannotbemadeuntilanactivityiscomplete

Ifyourmanager is stuck inameetingand thusyoucan’tget thego-ahead toregister for the conference before the day is out, then you’ve got yourself adependency. Another example would be waiting on a test environment or adatabase restore from production before you are able to move your workforward.

Tightlycoupledsoftwarearchitecture isavictimof thebigbully that isThiefUnknownDependencies.When a decision to remove a table froma databasenegatively impacts another team, Thief Unknown Dependencies scores bigtime.Thisisanexampleofasoftwarecodedependency.

Specialized expert skill-sets are at risk of being hit by this big bully thief. Adeveloper wonders, “Are there unknown vulnerabilities in this code?” whilewaiting on feedback from a security expert. But the security expert is busydiscoveringhowsomeonehackedintotheirnowunsecuredatabase.Aquestionwaitsoninputfromadatabasearchitect.“Isthedatainthetestenvironmentwrong?Cantheypleasecheckitout?”Butthedatabasearchitectisbusyhelpingthesecurityexpert.Whenyouaretheonlyoneontheteamwithaspecialskillset,youcanbethebottleneckpulled inmanydirections.Expertskills inhighdemand are often unavailable when you need them. Thief UnknownDependencysnickersindelight.

A similar problem occurs for changes outside of your control in the form ofthird-partyvendors.Majorcloudproviders,likeAmazonEC2,MicrosoftAzure,and Google Compute Engine, provide service-level agreement policies thatguaranteetheircustomers99.95%uptime.Thisequatestotwenty-twominutesofallowabledowntimepermonth.Whenyourcloudproviderisdown,youaredown, and Thief Unknown Dependency laughs at you. Granted, your cloudprovider is a known dependency, but do you always know itwhen they firststrike?Howmuchtimedoestheteamspendtroubleshootingaproblembefore

they realize it was the cloud provider who did it in the datacenter with thecandlestick?Butyou still lose even if it’s theprovider’s faultbecauseyouareconstrained by contractual agreements. You may be compensated with timecredits, but if a failureoccurs,howmuch time iswipedout trying to recoverlostdata? Ifyoutotaledthehoursa teamtroubleshootsan incident likethis,howmuchtimeisreallystolen?

WhyDependenciesMatter

TroyMagennis gave an enlightening talk on dependencies at the Agile 2015ConferenceinWashington,D.C.Troyusesbasicbooleanlogic(whereallvaluesare either true or false) to show that there is only ever one possiblecombination of inputs that result in an on-time delivery. Every time youremove one dependency, half of the total possible delay combinations areremoved.Ifdeliveryrequireseverypiecebeingcomplete,everydependencyyou

removedoublesyourchancesofdeliveringontime.1

Take a look at this example. If there are two inputs needed to deliversomething, then there is only one chance in four of delivering on time.Onechance in 2^n is the formula that computes the total number of binarypermutations.

Comeonnow—mathisfun!Yougotthis.Abinarydigitcanonlybea0ora1.Apermutationisawayinwhichanumberofthingscanbearranged.Abinarypermutation,then,isanarrangementofbinarynumbers.2nis2tothepowerofn.Whenthenumberofinputsistwo,n=2,andwehave2×2,whichequals4,or22.

Let’swritethemalldowntoseehowitworks.Therearefourwaystohavetwoinputs.

Iftherearethreeinputsneededtoship,thenthereisonlyaoneineightchanceofon-timedelivery.

Justbyremovingthestrictdeliverydependency,youdoublethechances(fromone in eight to one in four) of on-time delivery. Therewill always only be asinglechancewhereeverythingisontime.

Imagine you have reservations for dinner with four people who travelindependentlytoafinediningrestaurantandyou’retoldyouwon’tbeseateduntilallfourpeoplearrive.Therearesixteenpossibleoutcomes.

Thatistosay,sixteenpossiblecombinationsforwhetherpeoplearriveontimeornot.Ifyouchartthatout,fifteenoutcomeshaveatleastonepersonarrivinglateandonlyoneoutcomewhereeveryonearrivesontime.Dependenciesareasymmetrical in their impact. With four dependencies, it’s not a 25%probabilityyouwon’tbeseated,it’sa93%(15/16th)probabilitythatyouwon’tbe seated. There’s a much greater chance that fifteen out of sixteen times,someonewillbelate.Timetocallitanightandhituptheburgerjoint.

Figure 3, a three dependency chart, helps to visualize this math for three

dependencies where the probability is 12.5% of being seated on time.Whenyouaddonemoredependency, theprobability is justone insixteen,or .06%probability.Unless,that is,theyworkinOperations—thenthey’llnever leaveworkearlyenoughtoarriveontime.

Figure3.ThreeDependencyChart

YouknowThiefUnknownDependencyisstealingtimefromyouwhen:

Coordinationneedsarehigh,andprojectmanagersrunaroundtryingtogeteveryonealigned.Peoplearen’tavailablewhenyouneedthem.Achangeinonepartofthecode/outline/planunexpectedlychangessomethingelse.

When the local pizza company delivers more than two pizzas to the samemeetingroom,payattention.Atwo-pizzateamisateamthatcanbefedwithjust two pizzas—about five to seven people depending on the size of theappetite.Ifthreetwo-pizzateamsneedtohaveajointmeetingtodiscusstheirdependenciesoneachother,thenyouhavehighcoordinationcosts.Fifteentotwenty-one people bantering their point of view can consume a lot of time.Whenwasthelasttimefifteenpeopleagreedonanything?Whencoordination

needsarehigh,peoplearen’tavailablewhenyouneedthemtobe.

Small teams canmove fast. Nothing beats a small, cohesive group of peoplewho communicate and collaborate effectively. The problem occurs whendependencies span across teams, causing things to break. When one teambreaks another team’s functionality by creating incompatible changes, theimpactscanbedestructive,asmentionedintheopeningofthissectionaboutmy friend’s $23 billion company. When we attempt to increase theperformanceof individual teamsbybreakingthemdownintosmallergroups,hiddendangersawaitifthereareunknowndependencies.

Crossteamcommunicationishard.Whenabunchoftwo-pizzateamswithlotsof dependencies between them spend a lot of time coordinating to avoidsteppingoneachother’scode(duetothemergingofthedifferentteams’codebranches),thebenefitofthesmallteamdiminishes.Smallerteamscanincreaseintegrationcosts.Welikesmallteamsbecausetheycanmovefast.Justrealizethatbymovingfastasanindividualteam,youmaybepayingthepriceofnotmovingveryfastasawholeorganization.

Lastly,rememberthefollowingcommonattributesfromThiefTooMuchWIP:costlycontextswitchingandexpensiveinterruptions.Distractionisoneofthebiggest hurdles to high-quality knowledge work, costing almost one trillion

dollarsannually.2

KEYTAKEAWAYS

It’sexpensivewhenteamsareunawareofmutuallycriticalinformation.Architecture, expertise, and activities on hold are some of the common dependencies youmightruninto.Everydependencyincreasestheprobabilityyouwillbelate.Ifpossible,reducedependenciestosavetimeandmoneyandavoidothercomplications.Conversely,everydependencyyoucanfindandeliminatedoublesyourchancesofdeliveringontime.Whencoordinationneedsarehigh,peoplearen’tavailablewhenyouneedthem.Thesameistrueforexperts—whendemandforaskillishigh,expertsareunavailable.

Whenitcomestodependencies,individualteamperformancecanincreasetothedetrimentofcompany-wideperformance.

Codewillbeused inwayswecannotanticipate, inways itwasneverdesigned for,andforlongerthanitwaseverintended.—JoshuaCorman

1.3

UNPLANNEDWORKDowntown,USA,TuesdayMorning

Picture this:A seniorbusiness executive seesbusinessvalueinanintegrationbetweenhercompany’sproductandanothersoftwareapplication.HehiresathirdpartytointegrateanewserviceandpromiseszeroimpacttoherProductDevelopmentteams.

Theexternal,offshoreteamdesignsanintegrationbutoverlooksthefastgrowthoftheuserbase,resultinginanoverburdeneddatabase.Thedatabaseserverrevolts,setting off alerts to the duty pager. Operations staffhave to stop work on a high priority task totroubleshoot the angry database. Two cups of coffeeand two hours later, the issue is resolved, and peoplecan get back to work on that high-priority task theyweredeepintobeforetheinterruptionoccurred…rightaftertheirmeetingthatstartsintenminutes.Thiswasnottheintentoftheseniorbusinessexecutive.

People were interrupted and pulled away from important work. Theinterruption (the unplanned work) caught people off guard. It was anunintended consequence of a move by a well-meaning executive which

negativelyimpactedhigh-prioritywork,workthatwillnowbedelayedbecauseoftheinterruption.Thetimespentawayfromworkingontheoriginalpriorityis irreversibly wiped out. This is the problem with unplanned work—it setsback planned work. It increases uncertainty in the system and makes thesystemlesspredictableasaresult.

Sometimesunplannedworkcomesintheformofanecessarystrategicchangein direction: “Let’s stop marketing to everyone and just focus on largeenterprises.” But often, unplanned work comes in the form of unnecessaryrework or expeditedwork requests. These are the fires that stem from somefailure.Demand from failure is called, predictably, failure demand, and it is afrequent target forThiefUnplannedWork. It canbe the case, though, that adependency froma team justdown thehall fromyou is a greater risk tonotrespondingtoyourneeds.Thisusuallyresultsinacommunicationupyourlineofcommandtothecommondenominatorleaderandthenbackdownthefoodchain to the person responsible, resulting in an interruption/delay to thatperson’slunch(hopefully,it’sstillinthefridge).

Letmebeclear,Iamnotsuggestingallworkshouldbe preplanned. It is irrespon-sible (maybe evendelusional), to believe that it’s possible to knoweverything up front while planning a complexproject. Quite the contrary—we don’t knowmuchaboutwhatwedon’tyetknow.Sometimeschangesindirectionarenecessary,becausenewinformationemerges as we work to solve problems. A majorvalue of the Agile movement is to encouragerespondingtochangeover followingaplan.Life isuncertain. Change is inevitable. It’s a law—the

secondlawofthermodynamicstobeprecise.

WhyUnplannedWorkMatters

Unplanned and expedited work steals time away from work that’s creatingvalue.The2016StateofDevOpsReportsurveydatashowthathighperformers

are able to spend 28% more time on planned work.1 Unplanned work isconsidered ameasure of quality because themore unplanned work, the lesstime for creating value. “All hands on deck” incidents tend to reduceperformanceandincreasevariability.

As mentioned above, unplanned work steals time away from planned work.However, there are times when it is understandable and necessary forunplanned work to muscle its way to the front of the priority queue. If therequesthappenstobe,“Pleaselookintowhynoonecanlogintothewebsite,”thenyoureallyhavenochoicebuttodropwhatyou’redoingtofixthe issue.Unpredictablefluctuationsindemandcanreducetheabilitytodeliverthingsasexpected.

YouknowThiefUnplannedWorkisstealingtimefromyouwhenurgentissuespullpeopleawayfromfocusedeffortsoncreatingvalue.Thiscouldmanifestinanything from an unexpected fire drill to amalfunction with a heavily usedprogram that then adds uncertainty and variability into our everyday work.Because of this interruption, something else is going to take longer thanexpected.Iftheworkisfrequentlylate,chancesareThiefUnplannedWork(i.e.,failuredemandorstrategicchangeindirection)isstealingnotonlyyourtimebutalsoyourpredictability.

The reality is thatwework inwebs of interdependencies. The complexity ofhuman interactions consistently produces things that no one wants. ThiefUnplanned Work is a mainstay in a complex world where change anduncertaintyflourish.

Unplannedwork not only causes its own problems but bringswith it all the

problemsoftoomuchWIP:contextswitching,interruptions,delayedwork,andincreasedcost.Whenunplannedwork(suchasfixingbrokenfunctionalityonawebsite)creepsintotheeverydayworkscene,itincreasestheamountofworkalreadysittingonyourplate.Themoreurgentunplannedworkthatinterruptsyour day, the bigger the pileup of partially completed planned work. Therelationship between unplanned work and too much WIP is a twisted,codependentliaisonresultinginheapsofunexpectedworkthatisimpossibletocatchupon.Unlessyoukeepsteppingup,plannedresponsibilitiesfallbehind.Overfunctioning becomes an automatic response, eventually becomingdysfunctionalandimbalanced.That’swhyitissoimportanttolearntoidentifyandtackle,asearlyandoftenaspossible,theissuesthatThiefUnplannedWorkcreates.

Unplannedworkincreasesrisk,uncertainty,reducespredictability,andsmacksdownmorale.Butthatdoesn’tmeanwehavetosimplylaythereandletThiefUnplannedWorkwalkalloverus.Therearewayswecanfightback.

Making work visible is a core component in combating thieves like ThiefUnplannedWorkandisalsoacorecomponentofkanban,thesystemthatwewillcomebacktoagainandagainthroughoutthisbook.Kanbancardsshowallkindsofinformationthattraditionallyhavebeenhardtosee.Kanbancardsliveon kanban boards, and kanban boards answer questions like, “What is beingworked on?” and “What state is the work in?” and “Who’s doing what?” Allessentialinformationisvisibleontheboard,soyoudon’thavetochasedownworkers to ask them what is happening and you don’t have to wait for apoliticallysanitizedweeklystatusreporttogetaglimpseoftransparency.

KEYTAKEAWAYS

Unplannedworkaddsunpredictabilitytothesystem.It’s all about predictability and expectations. Unplanned work eats expectations forbreakfast.

High-performing companies spend less time working on unplanned work than lowerperformingcompanies.Sometimeswehaveno choice but to drop current projects to focus onurgent unplannedwork.Unplannedworkstealstimeawayfromplannedwork.Unplannedwork is hard to see, but it can bemade visible. Kanban helps to combat andbetteranticipateunplannedworkbymakingworkvisible.Planforunplannedworkbyreservingcapacityforwhenitarrives.

Focusisamatterofdecidingwhatthingsyou’renotgoingtodo.—JohnCarmack

1.4

CONFLICTINGPRIORITIESImaginearoomofforty-oneITOperationsengineersworkingforasuccessfulgaming company. They are smart. They are engaged. They are well paid. Onoccasion, Nerf gun darts zoom across the room. All appears to be thriving.Exceptthatwhenyoustandbackandobservetheroomabitlonger,younoticethingsthatdon’tlookright.

The team’s poorly attended stand-ups (a fifteen-minutemeeting where teammembers check in with each other to discuss what’s preventing them frommoving forward)heldeveryday, includemore foot shuffling thandiscussion.The VP of Operations jokes half-heartedly about another trip to the CEO’soffice as if hewere being called to see the Principal. Peoplewearing anxiousexpressions frequently approach the twoprojectmanagers’ area.The anxiouspeoplearebusinessproductownerswhowantstatusupdatesontheirprojects.Itturnsouttheonlywaytheycangetvisibilityontheirprojectsisthroughtheproject managers, but the two managers are busy planning capacity andprocuringhardware.That list is jotteddownon stickynotesand innotepadsand on calendar tasks—in other words, it is spread out overmany differentlist-makingtools.

Theprojectmanagersfacilitateadailystand-upinfrontofaseventy-two-inchplasmascreenmountedonthewall.Seekingclarity,theyopenthefloortotheOperations team, asking them to identify blocked work and other issues

preventingthemfromfinishingtasks.Theprojectmanagers’hopeisthattheywill be able to help remove project impediments and provide accurate statusupdates to the product owners. But when the project managers ask theOperationsengineersifanyoneisblocked,theOpspeoplereturnblankstaresandsilence,becausetheydon’twanttopointfingersorgetateammemberintrouble.Theissuescausingtheblockagesareinvisibletotheprojectmanagers.

This isafamiliarsceneonanygivendayatorganizationswherestand-upsgowrong. The engineers are trying to get clarity on priorities from the projectmanagers,andtheprojectmanagersaretryingtogetclarityonstatusesfromtheOperationsengineers. Inbothcases, thecommondenominator isunclearpriorities.Invisibleworkandinvisibleprioritieshamperthealignmentneededamong engineers, project managers, and business people to make effectiveprogress.Again, it all comesback to thenecessityofmakingworkvisible foreveryone.

However,formany,thereisconfusiononwhatrealprogresslookslike.Ateamofengineerswholookreallybusybutwhoarenotcompletingprojectfeaturesarearedflag.Havingabunchofprojectsthataresittingat90%completedoesthecompanynogood.Salescan’tsella featuretoacustomer if thecustomercannotaccessthefeature;featuresareonlyvaluableifcustomerscanusethem.

Going back to the same daily stand-up, in the top left-hand corner of theplasma screen is a list of the Operations team’s top four priorities: capacityexpansion, disaster recovery, security, and site reliability. The assumption isthat,giventhesepriorities,theteamcanprioritizetheirownwork.

Ofthethirty-threeprojectsthatthisteamofforty-oneengineersisworkingon,morethanhalfofthemareidentifiedaspriorityone.Yet,nooneiswillingtopointoutthattheteamisbeingaskedtodotoomanyprojectsatonetime.Andno one is paying attention to the metrics that show just how long work issittingqueuedup,waitingforsomeonetobecomeavailabletodothework.Thecareless implication is that all the projects must get done now. The teambelievestheyaredoingeverythingintheirpowertogetalltheworkdone,yetmany of the thirty-three projects remain incomplete, and new projects arestartedbeforeexistingprojectsarefinished.

We’veallencounteredsituationslikethisinvaryingcontexts:groupprojectsinmiddle school in which no one yet knows how to prioritize, unreasonabledeadlines set by managers who want all the work done “yesterday,” and/ortrying to multitask through your to-do list by tackling everything at onceinsteadofaccomplishingtasksbylevelofvalue.

The negative impact of copious untracked dependencies, lengthy cycle times,and habitual overtime is invisible in the short term. Eventually, though,

network errors, security oversights, and missed delivery dates becomeuncomfortablyvisible.What’snothappeninghereistheacknowledgmentthatsomeoftheseprojectsshouldbesetasideuntiltheteamshavecapacity.

WhyConflictingPrioritiesMatter

ThiefConflictingPriorities cackleswithdelightwhenpeople areuncertainordisagreeonwhattoworkon.

Sayateamwasworkingonareportthattookagestocomplete.Notonlydidittake a long time, but it was also delivered six months later than leadershipwanted.Let’ssayweexaminedtheteam’sworkload,anditturnsouttheyhadthirteeninitiatives,whichwasmorethanthenumberofpeopletheyhadonthewholeteam.Furthermore,theirprioritymeetingstooklongerthananhourandhappened everyweek. Ifwe reduce this team’s initiatives to, say, seven, theynow have better focus and their prioritymeetings are shorter. Reducing thevolumeofWIPhelpspeopleprioritizemoreeffectivelybecausetherearefewerthings vying for attention. Lest we forget that Thief Too Much WIP is theringleader of all the thieves, one cause of too much WIP is the failure to

prioritizeproperly.

If people can’t prioritize effectively, they try to do too much at once andeverythingtakeslonger.Then,toomuchWIPequalslongercycletimes,whichleads to delays in delivering value to customers. Happy customers (whetherinternalorexternal)brightenourdayandourpocketbooks.Longercycletimesdelaytheopportunitytoreceivevitalfeedbackfromcustomersaboutourwork,whichinturncreatesacreviceformorethieverytosneakin.

If everything is priority one, then nothing is a priority one, and everythingtakestoolong.AsRossGarbersays,“Manythingsmaybeimportant,butonly

one can be themost important.”1 It could be that the greatest value for thebusiness today would be for you to go help someone else finish somethinginsteadofstartingsomethingnew.

YouknowthatThiefConflictingPrioritiesisstealingtimefromyouwhenyouhearpeopleask/say:

“Whenwillmythingbedone?”“Mythingisahighpriority!”“Ifmythingdoesn’tgetdoneby_________,then…”

Another indicator conflicting priorities are bogging down yourwork is whenyou find yourself spending countless hours inmeetings discussing priorities.ThiefConflictingPrioritiesisaclosecousinofThiefUnplannedWork.Andjustlike unplanned work, heaps of older, planned work pile up when prioritiesconflict. When today’s highest priority work supersedes yesterday’s highestprioritywork,looktotheringleaderThiefTooMuchWIPasthecoreproblem.Teamswillfallbehindonplannedresponsibilitiesunlesstheykeepsteppingup.

KEYTAKEAWAYS

Thereisonemostimportantthing—letpeopleknowwhatitis.Conflictingprioritiesoccurwhenpeopleareuncertainonwhat thehighestpriority is.This

leadstotoomuchWIP,whichleadstolongercycletimes.Conflicting priorities that compete for the same people and resources block flow andincreasepartiallycompletedwork.Whatweseeasapriorityfrequentlyclasheswithwhatothersseeasapriority.

Indelay,thereliesnoplenty.—WilliamShakespeare,TwelfthNight

1.5

NEGLECTEDWORKWhen Iwaswith Corbis,we relied on an ERP application called JD Edwards(JDE).Itwasanold,customizedversionandquitefragile.TakingJDEofflinetodobackupsordatabaserestoresimpactedtheaccountspayableandaccountsreceivablefunctions,andapplyinganewJDEupgradefromthevendorbrokethecustomizeddatabasetables.So,wedidwhatmanyITshopssurprisinglydo—thought short term and continued using the ten-year-old unsupportedversion. What could possibly go wrong? Well, the manual JDE build-and-deploy process routinely overwrote configuration files during deployments,causingneworderstodisappear.Eventually,everyonebecameafraidtotouchtheJDEserver,andasaresult,itwasleftaloneforthebetterpartoftenyears,until we replaced it with SAP. In someways, aging software ismuch like anolder car that needs regular oil changes and tune-ups to keep it functioningcorrectly. Old software in itself is not a problem. Old software that is notmaintainedandisnotpartofanautomatedbuild,test,anddeployprocessisaproblem.

Maintenanceoflegacysystemsisoneofthemostneglectedtypesofwork.Old,fragile systems decay,making them unpredictable as the technical debt addsup. The entropy of an isolated system always increases over time. If notrepaired or replaced, the system eventually bombs, blocking or delayingsomething important. This consumes time and energy, taking people awayfromotherimportantwork.“Ifithurts,doitmoreoften,”asthesayinggoes.

Frequency reduces difficulty. Until that principle is applied to systemmaintenance, the neglectedworkwill remain a problem.Whennew requestscontinuously jump over or bypass important maintenance work, neglectedwork sits alone, sadly, like a rejected child at a lunch table in the schoolcafeteria.

Thief Neglected Work often plants invisible technical debt in the system,knowingthatshort-termthinkingswaysprioritizationinfavorofnewfeaturesover protecting valuable assets. Like financial debt, technical debt incursinterest payments, which come in the form of extra effort required to fixsoftware bugs and develop new features. Conflicting priorities and neglectedworkarealsoclosecousins.(Iimagineyou’resensingapatternhere.)Neglectedwork doesn’t get the attention, the budget, or the resources needed to besuccessful, like that ten-year-old JDE configuration that was still using acustomizedversionofano longersupportedversion.The impactofthisagedand neglected system on our Corbis team was failure demand that occurredwhen the configuration files incorrectly pointed to the wrong instance. Thiswasahighmaintenanceproblemandcauseformanyatroubleshootingticket.

IfIhadtoidentifywhatkindofworkismostneglected,itwouldbetheworkrelated to improving quality, including deferredmaintenance, bugs, technical

debt,andcodewithouttests(legacysoftware,asdefinedbyMichaelFeathers).1

Timeandcostoftenwinwhenitcomestogettingproductoutthedoor(“Justskip those tests.Weneed to get thisdelivered.We’ll comeback to it later.”).Thecurrentcorporateculturethatfocusesonpeoplebeing“busy”allthetimeisabsurd.Work is neglectedwhen people are “busy.” Busy people, however, donotsignalproductivity—deliveredvaluedoes.

Twosignificantareaswhereworkstalls includeworkwaitingonfeedbackandwork deemed important but not urgent. A third factor is what Donald

Reinertsen termed “zombieprojects.”2 Zombieprojects are low-valueprojects

thatarebarely alive.They lurkaround looking forhandouts,but theygetnolove.Theyarestarvingformoney,resources,andpeople.

Nevertheless, they persist, and in doing so, these starving projects subtlysiphon people’s time and energy away fromhigher value projects.When youdiscover a zombie project, kill it. Kill it so themore important work will bedeliveredsoonerandwithfewerinterruptions.

Somepeoplehavedifficultykillingprojectsthathavestartedbecauseofadesireto avoid losing whatever time and money have already been sunk into theproject.Themorepeopleinvestinaproject,theharderitbecomestoabandonit,evenwhenamorerationaldecisionbasedoffoffuturevaluewouldbebetter.Thisisknownassunkcostfallacy.InThePrinciplesofProductDevelopmentFlow,Donald Reinertsen suggests we should only consider the incremental

investmentneededtofinishtheprojectincomparisontoitseconomicreturn.3

Purginglow-valuejobsfromtheworkqueuemakessensewheneverasurplusofhigh-valuejobsisinprogress.Inotherwords,it’sokaytokillzombieprojects.Iftheyare reallyneeded, zombieprojects can return fromthedead.The thingsthatmattermostmustnotbesidetrackedbythethingsthatmatterleast.

However,zombieprojectsarenottheonlycauseofneglectedwork.Businessesfrequently prioritize new feature releases over fixing technical debt. Theychoose to work on revenue-generating work instead of revenue-protectionwork.

This rarely works out as the business hopes, particularly as problemsdiscoveredduringthefinalstagesofuncompletedprojectsdragengineersawayfrom the newer projects. Because new work is started at a faster rate thanpartiallycompletedworkalready inthepipeline is finished,theworkpilesupandtakes longertodo(ThiefTooMuchWIPsneakingitswayinagain).Flowtimemetrics start to increase. It’s like rushhour traffic.Whenmore cars areentering the freeway than are leaving the freeway, drivers are greetedwith a

longercommute.Andlikethetrafficjamonthefreeway,theensuingbarrageofconstantinterruptionscanbringworkflowtoagrindinghalt.

WhyNeglectedWorkMatters

Important work sits waiting until it eventually becomes an emergency or itcauses distractions and interruptions. Neglected work is perishable. It ages.And like rotten fruit, it’swasteful.Fruit is expensive, consumes spaceon thecountertop,andgetsoldandmoldyandsmellsbad.Andwhowantsthat?

You know Thief Neglected Work is stealing time from you when you delayimportanttasksthatwilleventuallybecomeemergencies. It’s likeplanningtotake your better half out for your anniversary dinner, then deciding to skipgoingoutthisyearinfavorofbatchingitupwithnextyear’sanniversary.Howdoyouthinkthatwouldgoover?Ifyourpartnerwouldgiveyouanearfulaboutit,yourneglectedworkwillgiveyoutwo.

Shiningalightonhowlongworksitsneglectedisausefulexercisetoundergoin order to understand the relationship between old work (think zombieprojects)andnewer,competingprojects.Similartoalltheotherthieves,ThiefNeglectedWorkischeeredonbyThiefTooMuchWIP.

KEYTAKEAWAYS

Ifnotdealtwith,importantneglectedworkeventuallybecomesanemergency.Beware of invisible technical debt accruing while teams are sidetracked by short-termpriorities.Acknowledge zombie projects. Consider the impact they have on completing high-valueprojects.Eithergivethemtheattentiontheyneedorkillthem.

Learningisnotcompulsory...neitherissurvival.—W.EdwardsDeming

2.1

MAKEWORKVISIBLETakealookatFigure4,anillustrationbyPhilippeKruchten.1(I’vemodifieditto be hand drawn and I’ve used different colors.) It’s a superb visual forhighlightingimportantdifferencesbetweenvisibleandinvisiblework.Aquickglanceatthisvisualreturnsameaningfulmessage.Likeavisual language,wesee the relationship between invisible and visible work with respect to bothnegative and positive value. Highlighted in blue, “Architecture” radiatespositive value. Unless it’s an old, customized, fragile JDE implementation,whereinitwouldfallintotheyellowtechnicaldebtarea.

Figure4.VisibilityGrid

The image is a superb example of good visualization. It displays the fourelementsnecessarytomakeavisualwork:structure,usefulness,relevance,and

honesty.This iswhatwewanttoaccomplishwhenwemakeourworkvisible:easyontheeyes,accurate,meaningful,andefficientataglance.

Oncewehavevisibilityofourwork,wehavethetoolstomanagetheproblemsthat slow our workflow down and are able to create the solutions for theunexpectedwhenitarrives.

Anotherthingtoconsiderwhenlookingattheimpactofrelayinginformationvisually is that two-thirds of the population are visual-spatial learners.Visibilitymatterswhen amajority of people think in pictures rather than inwords.It’salso importanttoknowthatthis isn’ta learnedpreference,visual-spatial learners have a different brain organization than auditory-sequential

learners.Theylearnbetterbyseeingthanbyhearing.2Whatdoesthismean?Itmeans that it’s a struggle for potentially two-thirds of your teamwhen theycan’tseetheflowandpriorityoftheworktheyaredoing.Makingworkvisibleallows your team to level up because it works with how their brains workinsteadofagainstit.

Thissection isastartingpointformakingyourandyourteam’sworkvisible.I’ll be discussing different aspects of creating a kanban board as well asdifferent ideas and concepts that help in understanding why they areimportant in the larger context of your work. The goal here is to show howmaking our work visible is an extraordinarily simple way to show our workdemand—the amount and the type of work requested from us by everyone,includingourselves—andbeginfixingthetime-thiefproblem.Youjusthavetojumpinandstartdoingittoreapthebenefits.

And kanban boards are all about jumping in because they start with anamazinglysimplecore—theToDo,Doing,Donedesign(Figure5).Thegeniusofthisboarddesignisthatit’sself-explanatory—youneedtostartworkingonsomething (to do), are working on something (doing), or are finished withworking on something (done). Creating this straightforward kind of board is

easyforanyonetodo.Youcanapplyittoallyourtasks(yourdemand).

Figure5.TheToDo,Doing,DoneBoard

Italsoorganizesyourworkloadthatmayhaveotherwisebeenhardtosee,oroutright invisible, inasingleview. Imaginethisboardhanging inyouroffice:It’ssosimpleandsimultaneouslysoinformativethatyoudon’thavetoexplainittoothers—anyonewalkingintoyourofficewillbeabletolookattheboardandknowwhatyou’reareworkingonandunderstandthestatethattheworkisin,allwithoutinterruptingyouwithquestions.Talkaboutafastmeeting.

Hereisanexampleofoneofourself-explanatoryboards.

Mostboardsare,atminimum,madeupofToDo,Doing,andDonecolumns(orequivalent names) that represent the state that the work is in. The work isrepresented by the work item cards. In Figure 5, the blue squares representworkitemcards.

So,howdoesthiskindofkanbanboardactinreal-worldactionatyourwork?

First,therearesomethingstotakeintoconsideration.Forinstance,ifyouhaveato-dolistthatlookslikeit’sthesizeofWar&Peace,thenyoumaybeaskingifyoureallyneedtoputeverythingonyourboard?No,youdonot.Thequestionbecomes,whatdoyoutrim?Somethingsaresolowontheprioritylistthatit

doesn’tmakesensetoclutteryourToDocolumnwiththembecausethatwilldistractyoufromthemostimportantwork.Also,bythetimeyoufinishyourtop three to fivepriority items, yournext setofprioritieswill probablyhavechanged.

So, what to-dos are okay to leave off the board? What provides the rightamountofvisibilityfortheteamandtherightamountoftransparencyfortherest of the organization? The answer is: it depends. What you make visibledependsonwhat youdo andwhat causes your team themostpain.Anotherconsiderationistheimportanceonmakingvisibletheuncertaintiesimpactingyourteamandyourcompany’sbusinessvalues.We’ll tackleuncertaintiesandbusinessvalues inPart3, so let’s focuson identifyingwhatyoudoandwhatcausesyourteampain.

Theguidancehereforwhattoputonyourboard istoweighthetimecostofmanagingworkitemswiththevaluereceived.It’snotuncommontoseeteamsimplement a policy whereby they don’t create cards for work that takes lessthanfifteenminutes.Usuallythereareexceptionstothisrule.Knowingwhentobreaktherulesissomethingthatcomeswithexperience.

My takeonwhen tobreak the rules iswhen riskoruncertainty ishigh.Justbecauseataskonlytakestenminutesdoesn’tmeanit’snotimportant.Herearesomeguidelines.Aten-minutetaskprobablydoesn’tneedtobetrackedunlessoneofthefollowingistrue:

1. Only one person knows how to do it (Thief Unknown Dependencies). Making the workvisiblecanpromptsomemuchneededcross-training.

2. Theworkimpactsotherteams(ThiefUnknownDependencies).AsdiscussedinPart1,cross-teamdependenciescanbeveryexpensive.Theonetotwominutes it takestocreateacardonaboardislowenoughoverheadtobeworthcross-teamcommunication.

3. Someone’sjobprimarilyinvolvesdoingtasksthatlastfifteenminutesorless,meaningifthatperson’sworkisn’ttracked,thenit’sinvisible(ThiefTooMuchWIP).Ifalotofthework is invisible, then it’s awfully easy to pile too much WIP on top of that person’s normalworkload.

Whenyou thinkabouthowtoapply this guidance for whatshould not go on your boardandwhentomakeexceptions,you need to start askingquestionsaboutyourdemand,ifyouhaven’talready.Suchas,whatkindofworkdoyoudo?What kind of requests floatacross your desk, inbox, andchat window? What are theprioritiesoftheworkitemsonyourlist?Inotherwords,whatisthenatureofyourworkdemand?Theanswerswillvaryfromteamtoteam;differentteamswill havedifferentdemand.Here are examplesof someof the stuffdifferentteamsdo.

Each team has a different set of things they do, although sometimes theyoverlap. The Product Development team helps the IT Operations teamtroubleshootsecurityissues.TheMarketingteamhelpstestnewfeaturesthat

the Product Development team delivers. Visibility across departments issomethingwe’llgettoinSection2.3,whenwecoverthetopicofdependencies.

In the early days at Corbis, before we made our work visible, I experiencedendlessamountsofheroicsonholidays,weekends,at3:00a.m.Ialwayshadalist in my head of all the things outside of my control that constrainedmyefforts—a request from out of nowhere; a two-hour-long impromptu,unproductivemeeting;anewprojectdroppedontheteamwhilewewerestillfinishingupanolderproject…yougettheidea.

Iranted(mostlytomyself)asIworkedfuriouslytofinishthejobsonmyplateaboutnothavingenoughtimeinthedaytofitallthedemandin.

I’m here to say that unnecessary, overnight heroics suck. Here’s the thing:Everyone I workedwith at that timewas doing the same thing.Wewere alloverburdenedwithtoomuchWIP,conflictingpriorities,andadisjointedworkflow, resulting in negative impacts to our health aswell as to the company’sorganizationalhealth.

Lookingback,Icringeatallthatwastedtimeandhoweasilyitwouldhavebeento alleviate the mutual pain if we had had visibility on all the work and itsimpactstoalltheteams.Butwedidn’t,soourtimecontinuedtobesiphonedaway by invisible demand. This is why it is so important to identify whatprevents you and your team fromgettingworkdone.What randomizes yourday?Whatcausesyourteampain?

IdentifyingteampainisaportionoftheDemandAnalysisexerciseattheendof this section. It’s satisfying and valuable because it gives permission forpeople to speak theirminds instead of the typical restrainedwork language.(Whileranting,justbemindfuloftheLeanpillarofrespectforpeople.)

Thesearesomeexamplesofteampain:

Here are the top pain points that appear again and again in my DemandAnalysisworkshops(ThiefConflictingPriorities,atyourservice):

Toomanyinterruptions—Ican’tgetworkdone.Conflictingpriorities—everythingispriorityaone.

Ifthesearetwoofyourteam’sprimarypainpoints,youarenotalone.

One of the other lessons I took away frommy time at Corbis is the need toconsider other teams’ pain carefully, particularly customer pain (or businesspain).BythisImeanyourinternalcustomerswhoareunhappyaboutthewayyourworkresultsareimpactingthem.

It’simportanttomakeinternalcustomerpainvisibleforacoupleofreasons:

1. Youaregoingtoneedyour internalcustomersbuy-into limitWIP. IgnoringWIP limitsallowsthecontinuationofmoredemandthantheteamcanhandle.Thecycleofbeingoverloadedwillcontinueandyouwon’trealizethebenefitsofflow.Unhappypeoplearelesslikelytoparticipateinsolutions.Gettingbuy-intolimitWIPiseasierwhenyoueasecustomerpainalongwitheasingyourownteampain.

2. It’s not just all about us. We must consider the whole system by using a Systems Thinkingapproachtooptimizeworkflowacrossallteamsinordertodeliverbusinessvalue.Optimizinginfavor of one team can reduce the overall performance of the company.Organizational health

includesdiscoveringwhatourcustomersareunhappyabout.

Once work demand and pain pointsareidentified,you’llwanttoconsideryour work item categories. Creatingcategories for your work allows youto see different types of work—notallworkisthesame!It’simportanttohave this clearly articulated becausedifferent types of work may havedifferent levels of urgency anddifferent workflows need differentrules to accommodate them. Whenwecreatecategoriesforourwork,wecan collect the data necessary tocreate the metrics for the different

typesofworkthatshowsus(andleadership)thehealthofoursystem.

Yourworkitemtypecategoriescanbebasedonwhererequestsoriginatefromorwhorequeststhework.Alternatively,theycanbebasedonhowtheworkisprioritized or the states that your work flows through. Since there arenumerous ways to categorize your work item types, it’s important to alwaysconsiderwhat shouldbevisualized inorder to gather theproperdata and tobringvisibilitytoaddressproblems.

Figure6.BalancedWorkItemTypes

Sometimes,thebossorjustafewpeopledeterminethewholeteam’sworkitemcategories.Thisissomethingtoavoid.Thepeopledoingtheworkshouldalwaysbe involved with designing their workflow management system for tworeasons:

1. It helps ensure you have the right number and types of categories that cover the needs anddemandofyourentireteam.

2. Whenpeopleparticipateincreatingsomething,theyhaveownership,whichmotivatesthemtoinvestinsolvingproblemsandachievingdesiredoutcomes.

Whendecidingthenumberofcategories, I’ve foundthatsomewherebetweenthreeandsevenisgood.Anymorethanthatanditbecomeshardtomanage,becauseforeachcategory,youmayhavedifferentrules,differentmetrics,andpotentiallydifferentworkflows.

TheimportantthingtocalloutisthattheOpsteaminthisexamplebasedtheircategoriesontheirdemandandtheproblems(teampain)theywantedtomakevisible. In this case, the volume of unplanned work and lack of capacity formakingteamimprovements.

Whenyoudefineyourworkitemtypecategories,youarecreatinga legendtohelp your team work with a kanban board effectively. It also allows others,

frommanagementtootherteams,tointerprettheboardwhentheyseeit.

Onceyoulisttheworkyourteamdoes,determineyourteamandbusinesspainpoints,andcreateyourworkitemcategories,youcanstartfillinginsomemoredetailedaspectsofyourworktofurtherdevelopyourteam’svisibility.

Thisdetailedinformationwillgointhefieldsofaworkitem.Again,involvethepeoplewhodotheworkforagreementonwhatdatashouldtakeupspaceonthe card.The informationon the card should answer thesequestions: “Whatdatadoyouneedtomanageworkflow?”And“Whatdoyouwanttomeasure?”

Here’s a sample list that is in noway exhaustive but should get you up andrunning:

CardIDHeaderTitleDescriptionAssignee(s)CommentsectionTagsforquerycapabilityIconsforextravisibilityPrioritySubtasksorconnectedcardfieldsDatefieldfordatedrivenrequests

Figure7.WorkItemTypeExample

Once you have your work item card designed, your team can immediatelycreatecardsfortheworktheyarecurrentlydoingandpopthesecardsonyourboard.Ata singleglance, you’vegotaprettygoodviewofwhatyour team isworkingon.Figure8isfromanITOperationsteamwhocategorizedtheirworkbased on the combined desire to balance internal requests (teamimprovements)andbusinessrequests,andtheneedtosupportbothunplannedwork and regularly scheduled maintenance. On this board, the team justfinished an expedited request (yellow), and they’re currentlyworking on onebusinessrequest (blue)andonemaintenance item(green).Nextup intheToDocolumnisateamimprovementitemandabusinessrequest(orange)(blue).

Figure8.ToDo,Doing,DoneBoardwithColors

Often,teamsneedmoregranularitywithintheDoingcolumn.It’scommontoseeaDoingcolumnsplitoutfurthertorepresentsomekindoffeedback,test,orvalidationworkbeforetheworkmovestoDone,asseeninFigure9.Initially,there isadegreeof reader involvementnecessarytorenderboardtransitionsmeaningful,butafterseveralviewsitbecomesobviousthatcardsarenowinadifferent state: The business request is now in review, and themaintenanceworkisstillbeingimplemented.

Figure9.ExpandedDoingColumn

Things really start to become visible when you have a To Do, Doing, Done

boardwithsomeworkitemsonit(asseeninFigure10).Itbecomesevenmorehelpful when you’ve thought on those pain points you and your team havelistedandvoteonthetoptwoorthreethatyouwanttomakevisible.Ifit’sapainpoint,makingitvisiblemakesiteasierforyoutodosomethingaboutit.I’ve found that it just takesanengagedgroupofdedicated teammembers tobringextra,muchneededvisibilitytothoseitemsthatarepainpointsforyourteam.

As youdo all this, keep inmind that there isno sense in over engineering akanbanboarddesignupfront.Keepitsimple.Ifextragranularityisneeded,itwill become self-evident through usage. Once you have your current workvisibleontheboard,it’sgoingtochangeafteryoutakeitforatestdrivewithyourteam.Avoidanalysisparalysisatthispoint.Itwilltakesometimeforyourboarddesigntostabilize(twoorfourorsixweeks,dependingonthematurityoftheteam).

Inthemeantime,headonovertoSection2.2andgocatchathief.

EXERCISES

DemandAnalysis

PURPOSE:Toidentifythekindofworktheteamdoesandtherelatedworkproblems (the team pain and business pain). These items will becomeinputstotheboarddesignitselfinlaterexercises.

Time:30to60minutes

MATERIALS:

MarkersFlipchartpaperorawhiteboard

INSTRUCTIONS:Listoutthedifferenttypesofworkyourteamdoes.Refer

totheITOperations,Marketing,andProductDevelopmentteamexamplesonpages51and52forideas.

Then,listoutanyroadblockspreventingyouandyourteamfromfinishingwork. You can take a look at some examples of the teampain points onpage53.

Whenmakingyourlist,bespecific.IfyourITteam’sworkislatebecauseofconstantinterruptionsduetocompetingpriorities,noteit.IfyourMarketingteam’sworkislatebecauseworkpilesupintheDesigndepartment,noteit.Thisisyourtimetorantaboutthethingsyouusuallyjustmutterabout.Comeon,getitout.

Shininga lightonthesebottlenecks inyourLeankanbanflowdesignwillhelpmakeitpossibleforyoutobegintofixthepain.

Next, listoutyourcustomerand/orbusinesspainpoints.Seepage55forinspiration.Occasionally,someoneinmyworkshopswilltellmethattheirbusinessexecutivesareallreallyhappy,andtothatIwanttocallbullshit.Noproblemisaproblem.

IdentifyWorkItemTypes/Categories

PURPOSE:To categorize the different kinds ofwork in order to supportdifferentworkflows,differentdegreesofpriorities,andapplicablemetrics.

Time:20to30minutes

MATERIALS:

3x3multicoloredstickiesMarkers/pens

INSTRUCTIONS:Here’swhereyouandyourteamdecideonwhattypesof

worktomakevisibleviacardsthatwillflowacrossyourboard.Anywherefromthreetosevencardtypesisreasonable.Eachcardisassignedacolor.If your team struggles with wanting more card types, you can create acatch-allcategoryfortasksthathavethesameworkflow,andusetagsandiconstodifferentiatethem.Createalegendtoreferbackto.

CardDesign

PURPOSE:To design useful, relevant, and good-lookingwork items thatwillprovidepeoplewiththenecessaryinformationaboutthework.

Time:20to30minutes

MATERIALS:

3x3stickiesMarkers/pens

INSTRUCTIONS: Identify thedata thatyouwant tocaptureonyourcardsandcreatefieldsforthedata.Ifyou’reusinganelectronictool,thispartofthedesignwillbedoneforyou. Ifyou’reusingaphysicalboard,considerwhatfieldsyou’llneedtocapturethieveryproblems.

WorkflowMapping

PURPOSE:Tomakeworkvisibleinordertoseewhatisbeingworkedon,what state the work is in, and the problems associated with potentialdisruptionsanddelaystotheflowofbusinessvalue.

Time:40to60minutes

MATERIALS:

Flipchartorwhiteboard3x3multicoloredstickiesMarkers/pens

INSTRUCTIONS:First,askyourselfwhatpainpointsorhiddeninformationyouwanttomakevisible.Thisisthefunpart.Grabyourteamand,usingabigwhiteboard or flip chart (if youdon’t have awhiteboard or flip chart,use stickies on a wall or window), begin with three columns: Options(Backlog),Doing,andDone.MaketheDoingcolumnwidesoyoucanbreakitupintomorecolumnsifneedbe.Placeyourexistingworkontheboardanddiscusswhatworkstatesyou’llwanttohavevisibilityon.

Now,let’shavealookathowtomakethetimethievesvisiblesowecandosomethingaboutthem.

1. Listthedifferenttypesofworkyoudo(demandandwhereitcomesfrom).2. Grouptheitemsintooverallcategoriesofwork.3. Discusswhichworktypeseemstocausethebiggestproblem.Whyisitasourceofissues?

Thiswill be yourworking kanbanboard to use throughoutPart 2 of thisbook.

KEYTAKEAWAYS

Visual-spatial learners think in pictures rather than inwords. They have a different brainorganizationthanauditory-sequentiallearners.Theylearnbetterbyseeingthanbyhearing.Remember—two-thirdsofthepopulationarevisual-spatiallearners.Makingworkvisibleisoneofthemostfundamentalthingswecandotoimproveourworkbecausethehumanbrainisdesignedtofindmeaningfulpatternsandstructuresinwhatisperceivedthroughvision.Visualscanshowbusinesspainpointsandotherhiddeninformation.Wecanusevisualsystemslikekanbanboardstohelpmakeworkvisible.

Multitaskingismerelytheopportunitytoscrewupmorethanonethingatatime.—SteveUzzell

2.2

AMBUSHTHERINGLEADERBackporchoffice,8:35a.m.

Deepintoametricsreviewfocusedontheevaluationofmediancycletime,myeyeperceivesa tinyalert in thetopright-handcornerofmyscreen.Before itflashesaway,IseeablurbfromLizthatasks,“Hey,doyouhavefiveminutes?”So,whatdoIdo?IrespondtoLizandsayyesbecauseIadoreher.Wedothingsforpeoplewelike—it’soneofthefivereasonswetakeonmoreworkthanwehavecapacitytodo.Andwiththatshortexchange,I’veaddedmoreworktomyalreadyfullday.

AsmentionedinPart1,understandingandbeingawareofthesefivereasonsisgolden.Thefirstreasonisprettybasic:asteamplayers,wedon’twanttoletourtribedown,andasmentionedbefore,wegetendorphinsfromsayingyes.Thesecondreasonisfearofpublichumiliationorofgettingfired.Thethirdisthatwedostuffforpeoplewelike.Thefourthreasonrelatestothefactthatpeopleareoptimisticcreatures,whichleadsustothinkwecanfinishtasksfasterthanweactuallydo.Ofcourse,thingsalmostalwaystakelongerthanwethinktheywill.Andfinally,thefifthreasonisthatstart-ingsomethingnewandshinyismore fun than doing the grunt work it takes to finish something old andunglamorous.

All five reasons are a part of Thief Too Much WIP’s toolkit. As we havediscussed previously, Thief TooMuchWIP is the ringleader of all the other

thieves.Whileitisbadenoughonitsown,ThiefTooMuchWIPinfiltratesallthe other time thieves and exacerbates the problems that are created bydependencies,unplannedwork,conflictingpriorities,andneglectedwork.

Let’stakealookataquickrefresheronWIP.Remember,toomuchWIPmeansthat work arrives faster than you can complete it. It’s all the work you’vestarted butnot yet finished—all the partially completedwork. BecauseThiefTooMuchWIPscattersourattentionacrossmultiplethings,itstealsourtime,ourmoney,andourabilitytodeliverhighqualitywork.Thisresults inothershaving towait longer than they’d like to getwhat theywant, and you losingmoney due to the delay. Thief TooMuchWIP steals time away from gettingthingsdonesoonerratherthanlater.Andbecausewecramtofinish,theresultisnotthebeautiful,masterlycraftedproductwewant.

YouknowyouhavetoomuchWIPwhen:

Contextswitchingiscommon.Newtasksarestartedbeforeoldertasksarefinished.Inotherwords,wesay,“Yes,I’lldothat,”eventhoughwehaven’tfinishedabunchofotherstuffthatisalreadyonourplate.Workgetsneglectedandages.

When you find yourself frequently context switching or when you get askedthatfive-wordquestion,“Doyouhavefiveminutes?”andyousayyes,youallowyourselftobeyankedoutoftheflowzoneandintothedetourzone.

But there’s hope! Tracking WIP can help you avoid further distraction andpreventThiefTooMuchWIPfromstealingyoureveningsandweekends.

There aremanyways to trackWIP. The following example (Figure 10) helpsexposeWIP and starts by dividingWIP into threemajor categories based onwhorequestedthework:

Silverbulletsareurgentrequeststodosomethingrightaway,usually initiatedbysomeoneinaleadership position. They are in a category of their own because of their urgent priority(perceivedorotherwise).Businessrequests,includingfeaturework,content,anddesign,arethethingsITdoesthat“the

business”promotes,manages,tracks,andisheavilyinvolvedwith.Teamwork is the stuff IT does that teams initiate, such as dealing with bugs, technical debt,security,platformupgrades,andmaintenance.

Figure10.ExposeWIP

Again, this is justonewaytocategorizework; thereareotherwaysthatwe’llexploreabitlater.

Breakingwork down into categories helps people better visualize the flow ofwork,which inturnaidstheminunderstandingthecommunicationneedsoftheir teamandofpeopleoutside the team.Recognizingandaddressing theseneedswithintheteamisrelativelyeasy.Outsidetheteam,thistaskisharder,usuallybecauseextraeffortisrequiredtoensuretherightpersongetstherightinformation. And silver bullets probably require special communication withtheboss,becauseitwaslikelytheboss’sboss,ifnottheVP(orsomeonehigherupthefoodchain),whomadetheinitialsilverbulletrequest.Ifnotthem,thensomeonetheylistentoand/orwhoseopinionstheytakeseriously.(Wedostuffforleadershiptoreducefearofpublichumiliationorofgettingcanned.)

Horizontal swimlanes are sometimes added to kanban boards for extravisibilityorasawaytosetupWIPlimits.Swimlanesarelanesdedicatedforaparticularkindofworktoflowthrough.Acommonswimlaneseenatthetopofmanyboardsisalaneforexpedites(workitemsthathavesomeurgencyaroundthemandneedtobeexpedited).

WIP limitsareoftensetatthetopofthecolumnsofakanbanboardtokeepthe linear flowofwork in check, but theydon’thave tobe.Howyou imposeWIP limits is up to youand your team, those folkswho are impacted by thework.ThereareseveralwaystosetWIPlimits.WIPlimitsperworkitemtype,perswimlane,orpercolumnarecommonoptions.

WIPlimitsperpersonaresometimesthefirststepfolksnewtokanbantakeinordertostemthebleedingofoverburdenedpeople.MoreadvancedteamssetWIP limitsat team levels inorder tooptimizeworkflowat thebigger-picturelevel,insteadofjustthelocallevel.Thishelpsmultipleteamstocollaborateonteamgoalsinsteadoffocusingnarrowlyonindividualgoals.

Visuals like this help team members hold each other accountable withtransparency.TheWIPlimitsaddtensiontothesystem.Peoplearecompelledtoinnovateandresolveissuespreventingthemfromfinishingtheirwork.WIPlimits provoke necessary conversations. Some may start uncomfortably, butagain, the tensionfromWIP limits inspires teamstogetcreativeandprevail.WIPlimitsprotectpeoplefromhavingtoplaycatchup,andtheyimposerulesthat help you get stuff done. It’s the WIP limits that create the necessarytensioninthesystem.Theygivepermissionforpeopletosay,“No,Ican’ttakethaton rightnow;myplate is full.”They are the constraint that enables thecompletionofwork.

Asimplethreeswimlaneboard(suchas inFigure10)giveseachswimlane itsownWIPlimit.

Silverbullet requests tendtocomedirectly from leadership.This swimlane issometimescalledtheVPlane.CIOsandVPsoftendon’trealizethedisruptioncreated when they ask for things outside of the standard process. Bringingvisibilitytosilverbulletshelpsshowthecostassociatedwiththeserequests.Allwork,includinginvisibleWIP,hasanassociatedcost,somakeitvisible!Silverbullets may well be worth the cost, so to these we say, “We know this isimportant, andwe’ll do it, but youonlygetoneof theseat a time.”Limitingsilverbulletstooneatatimeisjustoneexample.

Teamwork in this visual consists of what I call revenue protection work;namely,fixingtechnicaldebtandsecuritywork.

Businessrequestsarerevenue-generatingwork.Thisswimlane isshadedpinkin the figurebecause they’ve goneover theWIP limit of five,whichpromptspeopletotakeastepbackandask,“What’sgoingon?”It’shelpfulwhenotherskeepushonest. It’sabit likebeingonadiet—it’seasiertoavoidsugarwhenpeoplearetheretowatchyouorderdesserteverytimeyougoout.

Remember, categorizing work by who requested it brings visibility to thecommunicationinvolved—internal,external,orleadership.

NoticethattheitemsinthefeedbackcolumninFigure10arenotdeliveredyetand will now take longer to do. Flow requires clear and prompt feedback.Waitingonfeedbackfromothersisoneofthebiggestdelaysinworkflow.Thelongerittakestogetfeedback,theeasieritistoforgetdetailsandtheharderitistobeabletopicktheworkbackupagain.Likerottenfruit,knowledgedecaysfasterthanwe’dlike.Promptfeedbackhelpsusnegotiatechallengingdemandsandadjustourtacticstomaintaintheflowstate.Italsoencouragesustofinishworkalreadyinprogressbeforelaunchingintoshinynewwork,nomatterhowmuch it glitters. Visualizing work through the lens of flow improves teamcommunicationandunderstanding.

Speakingofvisualizingwork,Iamremindedofthevisuallanguagethatoccurswithkanbanboarddesigns.The “pictures”ontheboard (theboardstructure,cardavatars,icons,andsymbols)areeasilyreceivedinformation.Weneedverylittleeducationtogetthemessage.Thewritingontheboard(thelaneheadersand on the cards) is perceived information. It can take some specializedknowledge to decode the abbreviated text and acronyms, but once they areknown, a quick glance at the board returns volumes of informationimmediately.Thecombinationofpicturesandwritingrespondstoourneedforanimble,unifiedlanguage.

Ifyouthinkaboutit,meteredonrampsareacombinationofpicturesandtext.The traffic light is the picture, and the sign below it provides the text. Afreewayonrampconstrainstrafficflowtoallowforsafermerging.Themeterisonly neededwhen there is high traffic.Have you ever tried tomerge onto afreewayatrushhourwhenthemeteringlightsaren’tworking?Yikes.

Ifyouwanttobemorepredictable, limitWIPtotheteam’scapacity.What isthe team’s capacity, you ask? Good question. Don’t let it be 90–100%.We’llcoverwhyinPart3.

Keep inmind that it’s okay to startwith a simple approach to limitingWIP.TakingsmallstepstowardagoalissomethingLeancoachescounselpeopletodo all the time. You’ve got to start somewhere. Startingwith a thin layer ofsomethinggoodissometimestheonlywaytogetstarted.Sometimestryingtomaketoobigachangetoyourcurrentprocess, suchas implementingastrictset ofWIP limits, could cause you to crash and burn. No one wants that tohappen.

LimitingWIP also has the advantage of limiting interruptions. I have a tinywood stove, which comes in handy when the power goes out. To conserveenergy, I used to use it every day during thewinter, regardless of the powerbeingoutornot.Unfortunately,theoverheadforstokingandattendingasmallwoodstoveeverythirtytoforty-fiveminutesishighduetotheinterruptionofwork.But,ifthefirelastedninetytoonehundredtwentyminutes,I’dbefinewith it because ninety to one hundred twenty minutes allows me time toconcentratelongenoughtomakeprogressoncomplicatedwork.

Whenjugglingfiveballsintheair,theamountofattentiongiventoeachballisbutafractionofasecond.Thereisanalmostcontinuousneedtoswitchyourfocus to the next ball. The same is true when handling five different workitems.Youcanonlygiveeachonesomuchattentionbeforeyouareinterruptedbyoneoftheotheritems.ThemoreWIP,themoreinterruptions.It’seasiertojugglethreeballsthanfiveballs,andinthebusinessworld,it’seasiertofinishsomethingandgetitdeliveredandoffyourplatewhentherearefewerthingstofocuson.

EXERCISE

ExploretheFiveReasonsWhyWeTakeonMoreWIP

PURPOSE:Toacknowledge thatmany (if notmost) people take onmorework than they have the capacity for, to hear and empathizewith teammembers on why it happens, and to discuss countermeasures (actionstaken to counteract problems) for how to deal with this commonphenomenon.

Time:15to30minutes

MATERIALS:

OnepenperpersonSeveral3x3inchstickynotesperpersonStopwatch

INSTRUCTIONS:Participantsbeginbypairingoffwiththeirneighborsandaskingeachotherthisquestion:“Whydoyoutakeonmoreworkthanyouhavethecapacitytodo?”

Allow two to three minutes for the interviewee to respond, while theinterviewerjotsdownoneanswerperstickynote.Thenswitchroles.

Onceeveryoneisdone,haveagroupdiscussionaboutthereasonspeopleoffered.Then,discussideasforhowtodealwithpeople’sdesiretosayyeswhentheydon’thaveenoughcapacity.Besuretospecificallyexplorewhattodowhentherequestcomesfromsomeonetheylikeorfromaboss.

Variation1:Askeachpersontowritetheirownanswersonstickynotesifnetworking is unnecessary, such as with a group of people who alreadyknoweachother.

Variation2:Askthegrouptocollectivelygroupsimilarresponsestogetherandpostthemonawalltobringvisibilitytothemostcommonresponses.

KEYTAKEAWAYS

ThiefTooMuchWIP infiltratesall theothertimethieves,uppingtheirdamageandmakingthemallthemoredifficulttocontrol.Therearemanyways tosetWIP limits.WIP limitspercolumn,perwork-itemtype,orperswimlanearecommonoptions.WIPlimitscreatethenecessarytensioninthesystem.Theyaretheconstraintthatenablespeopletocompletework.InvisibleWIPhasacost,somakeitvisible!Categorizingworkbywhorequesteditisjustoneapproachtovisualizingwork.It’sonethatbringsvisibilitytothecommunicationinvolved—internal,external,orleadership.Visualizingworkthroughthelensofflowimprovesteamcommunicationandunderstanding.Thecombinationofpicturesandwritingmeetsourdesire foraunified language.Use thiscombotoyouradvantage.

Thehardestthingwedoiscommunicateacrossteams.—TroyMagennis

2.3

EXPOSEDEPENDENCIESSanFrancisco,2012

AlargeorganizationlaunchestheirthirdAgiletransformation.Anewteamofconsultantsarrivestoassessthesituation.ThenewAgileteamsareorganizedinto subteams of five to nine people. Pizza is frequently delivered to theseteams.(RememberthepizzaproblemfromSection1.2?)Theyhadheardaboutthe success Google hadwith two-pizza teams, and so this large organizationdecideditwouldworkforthemtoo.Manystoriesimpactedotherteams.Theycalledthesestories“away”storiesbecauseyouhadtogoawaytoanotherteamtosolvetheproblem.Awaystoriesimpactedapproximately92%oftheteams.Lotsofpeopleseemedtobeaway,well,alot.

ThiefInvisibleDependencies(akaThiefUnknownDependencies)sneaksuponteams.Astealth-liketimethief,ThiefUnknownDependenciesremindsmeofsomethingsomeoneoncesaidtomethatstuck:“Bythetimeyoufindoutyousuck, you have sucked for a very long time.” Similarly, by the time invisibledependenciesarereported,youarealreadyindeepwater.Thedamageisdone.But have heart—There is, as always, hope. This thief tends to thrive,predictably,whenmanyteamsworkondifferentpartsofonebigsystem.Themoreteams,thehighertheprobabilitythatmorefeaturesareworkedonatthesametime,andthisopensthedoorformoredependencies.

YouknowThiefUnknownDependenciesisrobbingyouwhenyouhear,“Oh,by

theway, so and somade this change/did this thing,” and your jawdrops.Orworse, you get blindsided by the arrival of an unexpected instant messagesfromanotherteamindicatingadevastatingproblem.Thehardestthingwedois communicate across teams. So, what do you do now? All the pizza in theworldcan’tplacateThiefUnknownDependencies.

Whenacrisisoccurswithoneprojectteam,peoplefromotherprojectscanbepulledoffoftheirworktoaddressthecrisis.Thisleadstowhatisknownasahigh coordination cost. When the coordination cost between teams is high,people aren’t availablewhenyouneed them tobe—andaprojectmanager isoftenputinplacetocoordinate.

AlargeorganizationontheirthirdAgiletransformationattemptedtodealwithcoordination issues by using spreadsheets. Dependency spreadsheets beganfloatingaroundbetweensomeoftheteams.Thismeantthatsometeamswerein the know and some teams were blind to it. The hardest hit teams weresharedservicesteamsthatsupportedmultipledepartments.

The consultants decided to try andminimize this risk. After some thinking,theycameupwith the following ideas tobringvisibility toanddealwith thedependencies:

1. Usecross-functionalteamstand-upstoflagdependencies.2. Identifydependenciesusingadependencymatrix.3. Implementexplicitrulesforworkflowingbetweendifferentteamkanbanboards.4. Create a rotating dependency scout role—an enterprise-wide system architectwho knows the

systeminsideandout.

The idea of using cross-functional team stand-ups to flag dependencies wasquicklyabandonedbecause itwas impractical, ifnot impossible, forgroupsofimpacted people to attendnumerous daily team stand-ups. The coordinationcostwastoohigh,anditwouldresultinpeoplespendingalldayinmeetings.

The dependency matrix idea, however, had some real merit. One clever

consultant createda floor-to-ceilingdependencymatrix,which looked sortoflikeFigure11butmuchbigger:

Figure11.PhysicalDependencyMatrix

Witheverythinglaidoutlikethis,peoplecouldbegintoseetheproblemsthatresulted from having so many dependencies. It took a long time to finishstoriesbecausetheexpertscapableofsolvingtheissuesweren’tavailablewhenneeded.

Whenabunchof two-pizza teamshave a lot ofdependenciesbetween them,howmuch time is spent coordinating? It’s true,we like small teams becausethey can move fast. Just realize that by moving fast as an individual team,you’re less able to move fast at the organization level due to the highcoordinationimpactsfromhavingalargenumberofdependencies.

Now,whattodoaboutdependencies?If theyareaproblemforyouandyourdependencymatrix,orifsoftwareisn’tprovidingthenecessaryvisibility,then

youmustfindawaytomakethemvisibleacrossimpactedteams.

Snazzy electronic automatic dependencymappers exist, but notmany teamsseem to take advantageof them. If you’ve gotone inplace and itworkswell(your dependencies on specialists or tightly coupled architecture don’t causeyou grief), then that’s great. It’s still good to have a clear understanding ofotherwaystovisuallydepictdependencies.

Sometimes,dependencymappinginvolvesgood,old-fashionedartsandcrafts.Partofthejoyofmakingworkvisibleliesintappingintothoselong-dormant,construction-paper-and-colored-markers skills from your elementary schooldays.

Notethe lastcolumnontheboard inFigure12titled“ARCH,”which isshortfor architecture review. The intent of the architecture review is to provideexpertguidanceandsupporttoenablethedesiredbusinessfunctionality.Itisnotmeanttobeanauthoritativeapprovalmoment.

Figure12.Arts&CraftsDependencyBoard

Iftheseoptionsdonotappealtoyouorarenotaccessibletoyou,thentakea

lookatFigure13whichshowsanotherwaytovisualizedependenciesviayourkanbanboardusingaswimlaneofitsown.

Figure13.DependencySwimlaneBoard

Designyourboard(s)tokeepThiefUnknownDependenciesfarawayasshowninFigure14.

Figure14.DependencyTagsonKanbanCards

Or, visually call out dependencies between different teams so they can be

broadcastedwidelytoreduceexpensivebusinesspain.

Figure15.ShowDependenciesBetweenDifferentTeams

Teams rarelywork in isolation. If a skill set isneeded fromadifferent team,thenhand-offs between the two teamsneed tohappen.To avoid the “out ofsight, out of mind” issue, visualize work flowing between teams. This helpspeopleanticipateworkthat’sheadedtheirwayandavoid“Oh,bytheway,youneedtodothisthingtoday”situations.Italsoprovidesthemwithaheads-upon potential issues that are coming their way, which is usually muchappreciated.Visualizingimportantcross-teaminformationhelpscommunicateacrossteams.

Small Agile teams are pervasive across the IT industry, and it makes sense.Nothing beats a talented, motivated, and cohesive team that makes fastdecisions and canproduce amazing results. In situationswhere the teamhaseverythingitneedstodesign,build,anddeploytheproduct,thenluckyyou—ThiefUnknownDependenciesisnotstalkingyourneighborhood.

However,largeorganizationswithmanyteamsarenotsoluckyinthisregard.Atacertainpoint,theorganizationhasgrowntowhereit’simpracticalforlots

and lots of peopleworking on lots of different projects to be aware of everydecisionthatimpactsthem(suchasarchitecturechangesandnewthird-partyintegrations). The more teams, the higher the probability that moredependencies are going to sneak into your day, project, objective, and so on.The more WIP, the greater the likelihood of Thief Unknown Dependencieshuntingyoudown.

This is why you must expose dependencies—to avoid teams fromunintentionallybreakingexistingfunctionality.Askanyteamwhosecodehasbeen broken by another team because they were unaware of a dependencycreated by that team—it’s no fun troubleshooting a broken productionenvironment while customers complain about your company on Twitter. Beinformed and inform others, use whatever method works best in yoursituation, and remember that starting simple is better than not starting.Increasedvisibilityhelpsgetyoutothenextlevel.Itmayevenhelpyoutogetleadershipbuy-infororganizingbyproductteamsinsteadofbyprojectteams—somethingtoconsider ifyourcurrentorganizationalstructure isn’tservingyouverywell.

Theproblemisthatprojectteamsareshortlived.ProjectpeopledisbandafterdumpingtheirprojectontotheSustainmentorOperationsteamandheadofftoworkonanotherproject.Thistransitioniscostlyandtimeconsuming.Intheprojectteam’srushtomakeprojectduedates,thingsget leftout—youknow,like dependency information. This introduces more WIP into the systembecausetheteamonthereceivingsidemustinterrupttheteamonthedeliverysidetogetthenecessaryinformationneededtorunthenewstuff.Inthisway,the project team’s expertise becomes a domain knowledge dependency. TheWIPpiles up because by the timeOperations interrupts the people from theprojectteam,theprojectteampeoplearealreadyoffandrunningonsomethingnew,andnowtheyneedtocontextswitchbackandforthuntiltheoldprojectisstable or the new caretakers are effectively cross-trained, because, well,

dependencies.

Organizing teamsaroundaproduct allows thepeoplewhodeveloped, tested,and delivered the functionality to stay in their area of expertise. There is noneedforcomplex,dependency-drivenhand-overs.Instead,organizingteamsbyproduct decreases dependencies during hand-off to ongoing operationalsupport.

Theproject/productdistinctionisanimportantoneformanyreasons,solet’stouch on that here for a moment so we don’t conflate or confuse the two,especiallysinceoneismoreproductivethantheother.Projectsaredeliveredasonebigmonolithicthing,meaningthatcoordinatingalltheactivitieswithinabig release is difficult and slow. Projects create big batches of work that arehandedofftoothersattheendoftheprojecttodeliverandmaintain.Projectscomeandgoandrequireextracoordinationandcommunicationtosetupandorganizetemporaryteams.Manyissuescanarisewhenmaneuveringthroughacumbersome,project-orientedprocess.

In contrast, organizing and managing by product keeps the same group ofpeople with the necessary expert domain knowledge consistently involved.Those who develop the product features don’t leave; they stick around todeliverchanges toprodandmaintenance.Project teamstendtobemeasuredbyvanitymetrics (e.g., test teamswithinaproject teamaremeasuredby thenumber of software bugs), whereas product teams are measured by thebusinessvaluederived.

SeniorDirectorofTechnologyatPivotalCorneliaDavisnotedinconversationatthe2017DOESForum,“Architectureissotiedtohowwedoourwork.Thepreferredarchitectureislooselycoupledcomponents,individualmicroservices,

builtbyindividualteams—autonomousproductteams,notprojectteams.”1

EXERCISE

The“Oh,BytheWay”DependencyMatrix

PURPOSE:Tobringvisibilitytodependenciesacrossteams,tohelppeopleanticipatewhat’sheadedtheirway,andtopreventdelays fromunknownorinvisibledependencies.

Time:60to90minutes(possiblylongerforverylargeteams)

MATERIALS:

LargewhiteboardorlargepaperorwallspaceStickynotesMarkersPizza(absolutelyessential)

INSTRUCTIONS: Gather up the detectives on your teams. Their mission,should they choose to accept it, is to investigate and visually capturedependenciesacrossallteamsthatcouldnegativelyimpacttheirwork.

Draw a large square graph with columns and rows. Using sticky notes,populate the column headers with your team names. Populate the rowlabelswiththesameteamnames.

Figure16.ExerciseExample

The rows show teams that are impacted. The columns show teams thatimpactothers(theimpactors,ifyouwill).

Identifytheoutputsfromeachteamthatcreateworkforanotherteamandwritethatnumberintheintersectingsquare.

For example, a customer success effort to provide self-service trainingcontentimpactsMarketingandSalesbecauseoftheawarenessrequiredtopromoteandsupportpresales.Salesrequiresmonitoringofthecustomerexperience and collecting customer data for personalized offers andpromotions, which impacts Product Team 1 because of changes to thewebsiteanddatacollection.ProductTeam1inturnimpactsITOpsduetosecurityanddatastoragerequirements.

Yourjobhereisto identifydependenciesbetweenteamsforanimportantupcoming feature or project, and mark an X in the intersecting squarewheredependenciesexist.Eachcellofthematrixrepresentsoneormoredependencies between the two intersecting teams. Capture the

dependenciesthemselvesonthematrix.

Once your cross-team dependencies are identified in thematrix, discusswhich actions can be taken to reduce the risk of breaking or negativelyimpactinganotherteam’swork.

Variation1:Includeupcomingrisksinthematrixadditiontodependencies.

Variation 2: Instead of teams, call out software components in thedependencymatrix.

KEYTAKEAWAYS

Smallteamscanmovefast,butiftherearedependenciesbetweenthem,youpaythepriceofnotbeinginapositiontomovefastasawholeorganization.Design your board(s) to highlight dependencies to keep Thief Unknown Dependencies faraway.Visually call out dependencies so they can be broadcasted widely to reduce expensivebusinesspain.Visualizedependenciesbetweendifferentteams’kanbanboards.Organizearoundproductteamstoreducetheproblemsassociatedwithprojects.

InthesetimesIdon’t, inamannerofspeaking,knowwhatIwant;perhapsIdon’twantwhatIknowandwantwhatIdon’tknow.—MarsilioFicino

2.4

THEPERFECTCRIME—UNPLANNEDWORKLosAngeles,2013

IaskedtwoITprojectmanagersresponsibleforsupportingtheteamofforty-oneengineerswhatpreventedthemfromgettingworkdone.Theirimmediateresponsewas constant interruptions. Theywerebeingbombardedwithprojectstatus questions from other teams and product owners. So, we conducted aone-week experiment to capture interruptions. Every time an interruptionoccurred, theywrote itdownonayellowstickynoteandplaced it in the toplaneofafour-by-six-footmobileboardonwheels(Figure17).

Figure17.AStudyinInterruptions

The yellow stickies flowed across the top swimlane. Green sticky notesrecording internal team improvements and otherwork impacting the projectmangers (represented by miscellaneous colored stickies) drifted across themiddle swimlane. And in the bottom swimlane, multiple colors of stickiesrepresented thedifferent projects theymanaged.At the endof theweek,wecounted ninety-two yellow stickies. Ninety-two! Most came from walk-uptraffic asking about project status. The project manager’s primary internalcustomers,theproductowners,wereinthedarkaboutthestatusandnumberofprojects.

Forthefirsttime,theproductownerscouldseetheotherprojectscompetingwiththeirown.And, importantly,theycouldunderstandwhyanotherprojectmightbeprioritizedhigher.Theycouldalsoseeforthemselvesthedestructivenatureoftheir interruptions.Ontheflipside,theprojectmanagerscouldseehowlittlevisibilitytheproductownershadregardingthestateoftheirproject.

The experiment also served as a chance to try out one board design beforecommittingtoadesignusinganelectronictoolthatisnotbeeasytochange.

Visualizing interruptions is quite useful for exposingThiefUnplannedWork.Figure 18 shows another way to visualize interruptions. This team placed apinkdotontheirworkitemeverytimethatworkwasinterrupted.

Figure18.AStudyinPinkDots

Thepinkdotsgavethemaquickwaytovisualizehowinterruptionsimpacteda

strategy used elswhere is the grawlix string ( ) that shows theinterruptionsinanamusingmanner.

Here’s how it works. Each time work is interrupted, add one grawlix to theticketontheboard.Thelongerthegrawlixseriesontheticket,thelongertheleadtimeand(presumably)themoreirritatedtheengineer.

Itwouldbeinterestingtocomparethegrawlixcountonworkthatisconsideredfunandengagingversusworkthatisnotfun.Icanimagineaticketwithworkthatnoonewantstodo(sayalegacysystemwithafunfactorofzeroorsometediousmaintenancejob)havinganextremelylonggrawlixstring.

Ours is aworld filledwith uncertainty,where everything is always changing.Sometimeswe justdon’tknowwhatwewantorneeduntilwesee it. It’swhytherewillalwaysbeunplannedworkandwhyweneedtoplan forunplannedwork.Becauseunplannedworkcanwipeoutyourobjectives,unplannedworkdeservesvisibility.Allofusperceiveasituationthroughtheexperienceofoursenses.

TheboardinFigure19exposestheexperienceofunplannedworkvisually.It’shardtorepairinvisibleproblems.Makingworkproblemsvisiblehelpsuncovertheunknowns.Someofthesemaybescarytosee,butoncethey’reoutintheopen, theproblemscanbegin toget fixed.Asmentionedbefore,kanbanmayscareawaythemeekbutiswellsuitedforthosewiththecouragetofaceissues,adapttochange,andfixproblems.

Figure19.ExposeUnplannedWork

Thinkofthisboarddesignasawaytovisualizethingsthatinterruptyourweekorruinyourday,thingsthattakeyouawayfromdeliveringimportantbusinessvalue because you are busy putting out fires. This unplanned work flowsthroughitsveryownswimlane.Acardiscreatedforeverypieceofunplannedwork.

There’susuallyresistancetothisintheformofPlatformOperationsManager

Erik, who says, “I don’t have time to create a ticket every time I getinterrupted!” But after weeks of interruptions, the CIO wants to know whyAzureisn’tupandrunninginproduction.AndwhatdoesEriksay?“We’vebeenbusy.”Thisistheproblem—thereisnoevidenceoftheissuespreventingErikfromcompletingtheplatformwork.It’saperfectcrime.AndThiefUnplannedWork goes home and collects $200 by passing go.When unplanned work ismadevisual,otherpeoplecanseeitandunderstandwhyworkisn’tdone—andsteps canbe taken to prevent, or at least limit, unplannedwork from takingoverinthefuture.

Figure20.MonthlyDeltaTrendforUnplannedWork

Knowing the ratio of unplannedwork to plannedwork helpswhen planningyourworkload capacity.Why?Because you’ll have an idea of how to setWIPlimitstoaccommodateimportant,unexpected,andurgentwork.Ifeveryweekthere is 25–50% unplanned work, then allocate 25–50% of your WIP forpotentialunplannedwork.Here’sawaytodothis:

1. Calculate your ratio of planned work versus unplanned work by dividing the number ofunplannedwork items completed lastmonth by the total number of planned and unplannedworkitemscompletedlastmonth.Forexample,ifyoucompleted100workitemslastmonthand

fortyofthemwereunplanned,theratiois40%unplannedwork.2. Giventheaboveexample,40%ofallyourWIPshouldbereservedtohandleunplannedworkand

low priority work. Low priority work is important work, just not the highest priority. Thisstrategyprovidesyouwithcapacity tohandleunplannedworkand/ortoset the lowerpriorityworkasideintheeventofanewcriticalrequest.Italsohelpsyougetimportantworkdonethatmightotherwiseturnintoanurgentcriticalrequest(asin,thisserverjustreachedmaxcapacity)inthenearfutureorfurtherdowntheroad.Eitherway,importantworkgetsdone.

3. Eachmonth,viewyourratioofplannedworktounplannedworktoseeifthetrendismovingupordown,andadjustyourWIPallocationaccordingly.

NotethatwearetalkingabouttheamountofWIPhereandnottheamountoftime.Allocatingtimeisadifferentanimal.Sometimesyouhearmanagerstellpeople,“Allocate50%ofyourtimetodoblah,blah,andsetaside25%ofyourtimetodoblah.”Buthowdoyoureallydothat?Youareararebirdifyouarefortunate enough tohaveawide-open calendarwhichallowsyou to schedulehalfyourday(sayfourhours)onblah,blahandaquarterofyourday(saytwohours)onblah.Becausemuchofthedayislikelyslicedupwithmeetings,email,andcontextswitchingbetweentasks,gettingtwohoursofuninterruptedtimetoworkonanythingcanbealuxury.

You may be thinking that allocating your workload (your capacity) by WIPlimitswon’tflybecauseyourworkitemsaren’tallthesamesize.Thisisanareawheresizedoesn’treallymatterbecauseyoucanonlyworkonsomanythingsatatime.Itdoesn’tmatterhowbigorsmallsomething iswhenyoucanonlytrulyfocusononethingatatime.Itcouldbeassmallasamouseorasbigasanelephant (metaphoricallyspeaking).When it’sdone,youmoveonto thenextthing.

InterruptionBusters

Trythesestrategiestocreateaconsistentformatformanagingyourtimewhentacklingthechallengeofmakingunplannedworkvisible:

The goalie: Someone designated to run interference for therestoftheteam.Itisarotatingrole,withthedualpurposesofprotecting others from interruptions and cross-training teammembers.Office hours: Just like with your teachers back in school,schedule a specific amount of time (a time-boxed amount oftime)acoupleofhoursaweekandletpeopleknowwhentheycangetyourtime.Officehoursgiveyourcolleaguesachancetodropinatyourconvenience.Donotdisturbhours:Amirror imageofofficehours.Puta“Backin1hour”signonyourdoororsomewhereinyourspaceto notify people when you’ll be back. I block off 6:00 to 8:00a.m. every day onmy calendar. This is when I write and do yoga.My boss once askedme toattenda6:30a.m.meeting.Isaidno.Ifeltguiltydoingthat,buttogettheimportantworkdone,wemust ferociously protect time to dowhatmattersmost. And by saying no, I established aprecedent.Nooneexpectsmetobeavailablefor6:30a.m.meetingsnow.Pomodoro: A timemanagementmethod using a timer to break down work into time-boxed

intervalsseparatedbyshortbreaks.1Setatimerfortwenty-fiveminutesandworkonyourtaskuntilthetimerrings.Whenthetimerrings,takeafive-minutebreak.Afterfourpomodoros,takealongerbreak(twentytothirtyminutes).Pomodorosgiveyoupermissiontoreallyfocus.Iusedthismethodtofinishwritingthisbook.

These four strategies create consistency thathelpyour teamseewhenyou’retacklingyourWIP.

Duringthe1960s,thecoffeecartatHProlledaroundat10:15everymorning.Alltheengineersdrankcoffeeandcasuallydiscussedtop-of-mindissues.Itwasagoldmineconditionthatgeneratedspontaneouscollaborativeadvances.Alotofproblemsgotunstuckatthecoffeecart.Inthe1970s,cost-cuttingdecisionswere made to replace the coffee cart with a self-service coffee pot on thecounterintheminikitchen.Engineersstilltookabreaktogetcoffeebutnotatthesametime.Nomoresetcoffeebreak,nomorespontaneousbrainstorming.Goneweretheunplannedcollaborativedevelopments—allinthenameofcostsavings. Some felt it was the beginning of the end for the great days of HP

researchanddevelopment.2

Just as regularly scheduled rendezvous (same place, same time) work as a

catalyst for frequent dialogue on top-of-mind issues in need of discussion,consistency ofwork times andbreak timeshelp establishhabits tominimizeinterruptions.Consistencysignalstopeoplewhentheycanhaveyourattentionand when they cannot. Consistency helps to minimize the damage fromunplannedwork.

EXERCISE

TheInterruptionReductionExperiment

PURPOSE:Toreducedamagefrominterruptionsusingempiricalevidence.

Time:45to60minutes

MATERIALS:

WhiteboardMarkers

INSTRUCTIONS: Gather the team to discuss ways to reduce the cost ofinterruptions.Thesemightincludehavingagoalie,schedulingofficehours,scheduling do not disturb hours, and using pomodoros or a variantinvolvingdedicatedninety—minutesessions.Whichofthesemethodsmightapplytoyourteamandwhy?

Comeupwithahypothesisandexperimentforaweek.Regroupaftertheexperiment todiscussyourobservationson the impactof theexperimentonyourteam.Whatworkedandwhy?Whatdidn’tworkandwhy?

Forexample:Hypothesis—schedulingofficehourswillreduceinterruptions.Schedule office hours from1–2p.m. onMonday,Wednesday, andFriday.Let everyone know you are available during these hours for impromptuquestions. This signals people (creates a visual) when you are availableandwhenyouareclosedforbusiness.

KEYTAKEAWAYS

Therewillalwaysbeunplannedwork,thereforeyoushouldplanforunplannedwork.Knowingtheratioofunplannedworktoplannedworkhelpsyouplanworkloadcapacity.Size doesn’t reallymatterwhen it comes toWIP limits—it doesn’tmatter howbig or howsmallachunkofworkiswhenyoutrulyfocusonjustonethingatatime.Aconsistenttimeandplaceforofficehoursanddonotdisturbhourshelpstominimizethedamagefromunplannedwork.

Manythingsmaybeimportant,butonlyonecanbethemostimportant.—RossGarber

2.5

PRIORITIZE,PRIORITIZE,PRIORITIZELosAngeles,2013

The six-by-four-foot mobile experiment board on wheels was positioneddirectlynexttotheprojectmanagers’desksandcoincidently,butconveniently,positioned alongside the main footpath between the entrance to theengineering space and the kitchen. Everyone had to pass this board on theirwaytotheirdeskinthemorningandontheirwayouttolunch/coffee/home.Itcaughtpeople’s attention.Productowners studied it to see if they could findtheirprojectsonit.Scrummastersstaredatit—unsurewhatitwasallabout.ITOpsEngineersdartedglances at itwith theuncomfortable familiarity andcommiseration from being overwhelmed by massive interruptions that theinterruptionexperimentexposed.

Thevisibilityon interruptionswasattentiongetting—andnot justbecause itcapturedhowoftenpeoplewerepokingtheirheadsintoofficeswithunplannedrequestsbutbecauseitalsoshowedthelackingprioritizationpolicy.

Returning to interruptions, the opening experiment with the two IT projectmanagers from the previous section—the one that captured interruptions—wasdeemedasuccessbytheITOpsProjectManagersforseveralreasons:

1. It caught theattentionof theVPofOperations,whoused it to re-emphasize specifickindsofwork thathewantedvisibilityon.Thesewere requests related to capacity expansion, security,site reliability, anddisaster recovery.Thesekindsofwork itemswere flaggedwithsymbolsonthephysicalboardandlaterfoundtheirwayintotheelectronicversionoftheboard.

2. Itraisedquestionsfromproductownersandotherteammembersaboutthecompetingprioritiesof the thirty-threeprojects.Thirty-threeprojects!That’s thirty-threeprojects inprogress foragroup of only forty-one people. Seems ridiculous doesn’t it? But that’s what happens whenprioritiesareunclear—peopletakeonmoreWIP.

So, the experiment continued for a secondweek. This time,we cornered theleadengineer into stack ranking the thirty-threeprojects.Hegave ithisbestshot.Wewere delighted to finally see that as heworked on it,we saw someprioritizedorder(Figure21).Butshortlyafter, theVPcamebyanddisagreedwith the priorities. This forced a necessary conversation, which was good,becausetherationalebehindhowvaluewasdetermined(atleastfromtheVP’sperspective)wasnowtransparent.

Figure21.AnExperimentinTaggingandPrioritizing

Thisisthepowerofplacingafour-by-sixrollingvisualalongthemainrouteforboth the department entrance and the very important coffee room. It’sunavoidable. The eye is drawn in and people stop to ponder. It provokesessentialcommunicationthatwouldotherwisebeconvenientlydodged.

After that discussion happened, the projects were reprioritized based on theVP’s input, a classic prioritization strategy known as HiPPO (highest paidperson’sopinion).Andthenaninterestingthinghappened.Thenextday,fournew projects appeared, seemingly out of thin air, in the top right corner, allvyingforthepriorityoneslot.ThiefConflictingPrioritiesscoredbigthatday,chucklingattheresultingdramacreatedbyinsertingmorechaos.(Iwishwe’dhadawebcamaimedatthatboard;itwouldhavebeeninterestingtoseehowtheygotthere.)

Bewarethelackofgoodrulesforprioritization—remember,wheneverythingisapriorityone,nothingisapriorityone.

Asalways,itispossibletogetoutofthismessbyemployingthekeyconceptswe’vebeenbuildinguponinthisbook—namelyexposingtimetheftbymakingworkvisible.

Forexample,proposeaprioritizationstrategytokickofftheconversation.YoumightusetheA3thinkingprocesstodothis.A3isaproblem-solvingapproachusinganinternational11x17inch(297x420millimeters)pieceofpaper.(A3getsitsnamefromthesizeofpaperofthesamename).A3encouragesprecisecommunication in an effective structure that cultivates understanding andagreement.OneresultofusingA3 isthatvariousoptionscanbe investigatedandfloatedacrosstheorganizationinadiplomaticfashion.

BecausetheA3methodhelpsyougetunderstandingandagreement,itcanbeused to discuss prioritization methods. The goal with prioritization is todetermine what to complete next in order to get maximum value in theshortestamountoftimeandtoavoidmultitaskingduetocompetingpriorities.

Figure22.A3Example

There are severalways toprioritize. Let’s look at someof themore commonmethods:

Highestpaidperson’sopinion(HiPPO):Each job is assignedapriorityby themost seniorperson in charge andprocessedaccordingly.Remember the thirty-threeprojectsprioritizedbytheVP?Cost of delay (CoD): A way of communicating value and urgency, CoD is a measure of theimpactoftimeontheoutcomeswewant.Thisisanexcellentapproachfordeterminingbusinessrisk,butdifficultformanypeopletoactuallydo.First-in, first-out(FIFO):First-in, first-outwork isprocessed ina first-in, first-out fashion.It’s a simple and fair process, such as at the movie theater, where the person ahead of youreceivestheirmovieticketbeforeyoudo.Weighted shortest job first (WSJF): WSJF gives preference to the shortest job with thehighest CoD. WSJF is calculated by dividing the CoD by the job duration. The Scaled AgileFramework(SAFe)modelusesavariantofWSJF,whichattemptstoincludetimecritically.

Therearemanywaystoprioritizework.Theabovelistshowsjustsomeofthe

options.

SwingingbackaroundtotheLAITOpsEngineeringteamthatweopenedthesectionwith,noticehowtheirprioritizationapproachevolvedfrom,“Doallthethings” to the lead engineer’s opinion of what was a priority to the VP’s oranotherexecutive’sopinionofwhatwasapriority—theHiPPOapproach.IftheVP or other executive has sufficient knowledge and experience, it is possiblethattheirprioritizationcanbegoodenough.Butproblemsoccurduetotheircognitive bias, misaligned goals, or overconfidence. We are often confident

evenwhenwearewrong,1anditcanbehardforustoseewhenwearewrong.This is whymaking prioritization policies visible is vital—it drives the rightconversationsfordeliveringidealoutcomes.

InPart1,clueswereprovidedfordetectingwhenThiefConflictingPrioritiesisstealingyourtime.Theyinclude:

Wastingtimeinmeetingsdiscussingpriorities.Takingonnewworkbecauseyouareunsurewhattheprioritiesare.Beingrepeatedlyasked,“Whenwillmythingbedone?”

Remember: Thief Conflicting Priorities is a close cousin of Thief UnplannedWork,whichproducesquestionslike:

“Whenwillmythingbedone?”“Thisthingisahighpriority!”“Ifwedon’tget___________doneby___________,wemightallbelookingfornewjobs.”

Here’s one way to see the things that conflict with each other. Unplannedexpeditedwork,projectwork,maintenancework—it’sallcompetingwitheachother.Whenworkgoesonholdbecausesomeonesaysyouneedtodothisotherthing now, show that in your visibility efforts. Show that implementing thenewsecurityvulnerabilityfixgotdeprioritizedbecausepeoplehadtodomeritreviews. The idea is tomake prioritization rules explicit.Otherwise, how arethey made? By the loudest, most aggressive person? By the highest paidperson?

Figure23.ExposingConflictingPriorities

Worktakesalongtimetocompletebecauseitsitsinqueueswaitingforstufftohappen.It’snotunusualforwaittimestobemorethan80%ofthetotaltime.Many organizations are blind to the queue problem. They tend to focus onresource efficiency instead of applying systems thinking to improve theefficiencyofthewholesystem,endtoend.

CostofDelay(CoD)

Priority is only a problem when things aren’t delivered in the desiredtimeframe.Weknow that trying todoeverythingat the same time increasestheriskofdelayingeverything.Andweknowthatchoosingtoworkononetaskputsothertasksonthebackburnerandcausingtheirdelay.Thegoal,then,istounderstandwhatworkshouldbecompletednextinordertoachievethebestpossibleoutcomes.Quantifyingthecostoftheworkthatgetsdelayedisuseful.

Figure24.InputsThatContributetoCostofDelay

AsmentionedinSection1.1,costofdelayisameasureoftheimpactoftimeontheoutcomesdesired.CoDboilsdowntothreethings:

1. Reducedrevenueorbusinessvalue(CoDisnotalwaysaboutmoney)2. Increasedcosts3. ThesumoftheCoDofotherworkthatisdependentontheworkwearecalculatingCoDfor

Let’s takeoneof the thirty-threeprojects from the ITOpsEngineering teamboard, “Build outNewDatacenter (BoNDC for short) to understandhow theCoDcouldbecalculated.”PartoftheDonecriteriaforProjectBoNDCisfortheold datacenter to close.Hence, the costs to keep the old datacenter runningneedtobeincluded.

Old datacenter costs include floor space rental, power, heating and cooling,operatingcosts,andmaintenancetotaling$7,400/week.Additionally,aprojectmanager spends 20% of their time managing the transition to the newdatacenter:$400/week.AlsoincludedintheCoDistheCoDofProjectP,whichwill improve performance but it is dependent on the completion of the newdatacenter: 8 engineers (50%) × $2,800/week = $22,400/week. Furthermore,

the cost savings from Project P (due to automation of manual processes) =$8,600/week.

Figure25.CostofDelay

TheweeklyCoDofBoNDCis$38,800/week.IfBoNDCisdelayedsixweeks,thetotalCoDis$230,400.

CoDcanbeused tonegotiateprioritizationofworkand tobringvisibility toprojects that have a bigger impact on the bottom line than the others.VisualizingCoDdrives therightconversationsanddecisionsaroundcostandrevenue.AsJoshuaArnold,thoughtleaderandfounderofBlackSwanFarming,says,

CostofDelaycombinesurgencyandvalue—twothingsthathumansarenotverygoodatdistinguishingbetween.Tomakedecisions,weneedto

understandnotjusthowvaluablesomethingis,buthowurgentitis.2

Hence, CoD communicates the impact that time has on value. Value usuallymeansmoney,butnotalways.Anon-profithealthorganizationmayconsidervalue as the number of lives saved. Work then is prioritized based on the

highest business value returned (economic or otherwise) rather than thetheoreticalROI.

CoD considers all the impacts towhat generates new revenue,what protectsexistingrevenue,andalltheexpensesassociatedwithrunninganorganization.

Simplyput,CoDhas todowith twoconstantly changingvariables: valueandtime. CoD asks the question, “What value is lost by the delay of something?Howmuchwillweloseifwedeliverthisthingtwelvemonthslater?”

TheLineofCommitment

The lineofcommitment isavertical linebeforeaspecificstatethatsignalsacommitmentonyourparttodothework.Thetasksinthebacklogareoptions,andtheymaynevergetdone.Butonceworkpassesthelineofcommitment,itexplicitlysignalsthatit’sbeenprioritizedandismovingforward.Itisnolongeranoptionbutafullyagreeduponandprioritizedpieceofwork.

There is no second-guessing once work passes the line of commitment. Theworkwillhappen—anditwillhaveacost.Howhighthecostisdependsonhowlong it’s delayed due to the competition with other planned and committedworkallvyingforthesamepeople’sattentionandresources.

Figure26.LineofCommitment

EXERCISE

VisualizePriorities

PURPOSE :Tohelpyoubringvisibility to competingprioritiesandbringclarity tohowwork isprioritized.Mostorganizationshave toomany topprioritiestoachievetheleveloffocustheyneedinordertosucceed.

Time:60minutes

MATERIALS:

Yourcurrentworkfloworkanbanboard

INSTRUCTIONS: Ensure those impacted have a voice in the discussion.Time-boxeachperson’scommentstonomorethanfiveminutes.

Questionstodiscuss:

Whatisyourprioritizationpolicyandhowisitvisualized?Howwillyousignalwhenworkhasbeenprioritizedandisreadytobeworkedon?Inotherwords—whereisyourlineofcommitment?Howdopeopleknowwhichworktopull?Howwillyouvisuallydistinguishbetweenhigherandlowerprioritywork?

KEYTAKEAWAYS

People take on more WIP when they are unclear on priorities. Establish an explicitprioritizationpolicytoavoidtoomuchWIP.The A3 problem-solving approach encourages precise communication in an effectivestructurethatcultivatesunderstandingandagreement.FormoreonA3,readManagingtoLearnbyJohnShook.Therearemanywaystoprioritize.Assignedpriority,CoD,FIFO,HiPPO,andWSJFaresomeoftheseoptions.Visualizeprioritiessopeoplearecrystalclearonwhatconstitutesthemostimportantwork.Thinkaboutthelineofcommitment.Bringclearvisibilitytoworkthatisprioritizedandfullycommittedbut,ifcompetingwithotherprojects,willincreaseitscostofdelay.

Neverletsomethingimportantbecomeurgent.—EliyahuGoldratt

2.6

PREVENTINGNEGLIGENCESmallTown,USA,2015

CTOFranksighsheavilyoverhisthirdMountainDewofthemorning.It’sthesecondtimethisweekthatthedatabaseserverhashitmaximumcapacityutilization.The team can’t get their heads above water longenoughtoperformseriouslydelinquentmaintenance.

Likeanoldbrokenchair,it’seasytocastasideanagingrequestwhenanewrequestarrivesinthebacklog.Butaswe’vediscussedpreviously,thecostofstartingnewthings before finishing old things is high. When toomuch work is in process, bad stuff happens: contextswitchingincreases,bottlenecksdevelop,dependenciesrise,windowsofopportunityclose,andholidaysarrive.Andbecausethingsaredelayed,peoplewanttopileonmorethings—youknow,“Let’saddthisotherfixwhilethehoodisup.”Thisallcausesworktotakelongerthanitshouldanddelaysthedeliveryofbusinessvalue.

An ITOperations teamatonecompany IworkedwithhadaworkflowwherethelaststepbeforeDonewasValidate.Therewasnogoodprocessatthetimetoprovidetimelyfeedback,soworkitemsjustsat,waitingtobevalidatedafterthey were delivered. The norm was to assume that all was okay if no one

complained.WhileworksatidleinValidate,peoplepulledmoreworkoutofthebackloginto“Implement.”TheValidatequeuegrewandgrewuntilithadcloseto100work items.Manyof the itemsdeliveredwere indeedokay, but somewerenot,andthosecustomerswereunhappy.

Whenthosedownstream,internalcustomersmentionedthattheworkhadn’tbeen delivered as expected, the IT Operations people took a long time torespondbecausetheywerealreadyworkingonthenexttask.AsthetimegrewbetweencustomerfeedbackandOperation’sresponse,thosecustomersbecameveryunhappyandeventuallygaveuptryingtocommunicatewithOperations.They complained among themselves and to other departments. “Ops neverresponds,”theysaid.“They’relikeabigblackhole—hellwillfreezeoverbeforethey do anything,” they groused. “What a worthless team.” The Operationsteam’sreputationsliddownthedrain.

The board in Figure 27 represents the IT Operations workflow. I show thisboard image in my workshop and ask folks, “Where is work stuck here?” Itdoesn’t take more than a second before people respond with, “The Validatequeue!”

Figure27.TheValidatePit

This is the power of the unified language of kanban boards—themessage iscommunicated instantaneously. Because the bottleneck lies in Validate, wemusttackletheValidatequeuefirst.

Here’stheexperimentwedid:WesetatimeframetoclosetheitemssittingintheValidatequeue.All items (called ticketsby this group) thathadnotbeenupdatedinfourteendaysormorewouldbeclosed.Weaskedafewkeypeoplewhattheythought,andwebegantosocializetheideaamongtheITOperationsandDevelopmentteams.Noonereallyseemedtooconcernedbyourproposal,sowepreparedboilerplatecommentstoaddtoalltheticketssittinginValidateand communicated our plan across the Development and Operations teams.We toldpeople thatwewere going to close the ticketsona setday and thattheyshouldletusknowiftheyhadanyissuewithit.Thatday,weclosed104tickets, and only two people responded to say they still wanted to keep theticketopen.Itfeltsogoodtogetallthoseticketsclosed.

Youmaybethinkingthatthevalidatefunctionwasworthlessifallthosetickets

couldjustbeclosedlikethat.Manyoftheticketsclosedthatdaywereancientandalreadyobsolete.Remember,otherteamshadgivenupontheOperationsteamandfoundothermeanstogetthejobdone.Butwedidn’tstopthere.

The next part of the experiment included gradually reducing the number ofdaysaticketcouldsitidle(gettingstale)inValidate.Wegavetheteamsthirtydaystoadjusttoidleticketsbeingautomaticallyclosedandthensqueezedthetimeframefromfourteendaystotendaysforthenextmonth.Teamsadjustedandstartedvalidatingticketssooner.Unlikebefore,whentheticketswerestaleandallbutforgotten,itwasrelativelyeasyforthemtocompletethevalidationbecausetheworkwasstillfreshintheirminds.

Thisisoneofthemajorproblemswithneglectedworkandanimportantreasonwhywe should avoid it.Once theworkhas sat untouched forweeks or evenmonths,we forget thedetails, and it takes a long time todiveback in.ThiefNeglected Work loves it when work sits around so long that you no longrememberwhatwashappening.SmackThiefNeglectedWorkupsidetheheadbykeepingworkclippingalongatasmoothpace.

Ofcourse,continuousimprovementisalwaysthegoal,andthirtydayslater,wecutthetimeframefromtendaystosevendays.Thirtydaysafterthat,wewentfrom seven days to five days. Why didn’t we just start out with five days?Becausefivedaysneverwouldhaveflowninthebeginning.Thechangewouldhavebeentooabrupt.Twoweeks is,relativelyspeaking,a longtime—it’snotthreateningtoanyone.Thegradualchangeallowedpeopletoadjustandadapt.Because the Validate queue usually empties every five days now (sometimesthingstakelonger,butnowpeopleupdatetheticketwithcommentskeepingitactive), the cycle time of the tickets has decreased dramatically.Work flowsthroughthesystemquickly,andproblemsinValidatesurfacerightaway.WiththeValidatepitimproved,wemovedontothenextbottleneck.

Multitasking is aneffectiveway toget lessdone. It’s aneffectiveway togive

negligence free reign.When you try to do toomany things at one time, youwon’t do any of them great, and sometimes you won’t finish them at all.Neglected work is another term for partially completed work. Consider apartially completed bridge. It is already expensive, but it provides zero valueuntilit’sfinished.

Thief Neglected Work steals time from you when important work becomesurgent.Goodtime-managementmeansspendingtimeonimportantthingsandnot justurgentthings.Maketimetoworkonfirepreventionsothatyoucanreducetheamountoftimeyouspendputtingoutfires.

Revenue-protectingwork is amajor target of ThiefNeglectedWork. Becausethebusiness is oftenunawareofwhat’s involved inkeeping a system secure,reliable, and functioning, revenue-generating work is considered a higherpriority than intangible maintenance and sustainment work. In manycompanies, short-term revenue-generating work is top-of-mind for businesspeople,butthelong-termhealthoftheplatformsystemisrarelycontemplated.Asaresult,revenueprotectionworktakesabackseatuntilsomethingblowsup,

suchasadistributeddenial-of-serviceattackoracorruptfinancedatabasethatbringsdowncreditcardfulfillment.

It’snot intentional—people just tendto focusondoingtheir jobandassumetheircoworkerswilldothesameandthateverythingwillbefine.Humansarelikethat.Wetendtoavoidtheannualdoctor’sexamuntilsomethingisreallywrong.

CallOutNeglectedWork

Here’s a way tomake neglected work obvious. Flag work items that haven’tmovedorbeenupdatedwithinacertainnumberofdays.InFigure28therearetwoflaggeditems,oneforninedaysandoneforthirteendays,respectively.

Figure28.ExposeNeglectedWork

Thenyoucanmakeagingworkvisible.QueryallyourWIP(so,notthebacklogandnotthedoneorarchivework)inyoursystemthathasn’tbeentouchedforthirtydays. Ifthatproducestoomanyitemstoseeatonetimeonthescreen

withoutscrolling,bumpupthesearchcriteriatosixty,ninety,or120days.I’veseenteamswithWIPover300days, sodon’t feelbad ifyou’vegota few likethattoo.

Todealwiththosearchaicitems,scheduleaten-minutemeeting(yes,justten-minutes)everydaydirectlyfollowingstand-upwiththecreatorandthecurrentassigneeofthework.

Here’ssomelanguageforyoutouse:

Subject:WorkItem#nnnnn(entertitleofitemhere).

Invitees:Thecreatoroftheworkitemandthecurrentassigneeoftheworkitem.

Location:Dominica’sdesk(orvideochatifremote).

Description:Thisworkitemisnowfamous.It’stheoldestWIPinthesystem.

Canwepleasetake10minutesdirectlyafterstand-uptomorrowtoseewhatitwouldtaketocloseit?

Manythanks!

Dominica

Thesemeetings should help you determine which projects are worth saving,whichcanbequicklywrappedup,andwhicharezombieprojectsthatneedtobepurged.Remember,zombieprojectsarelow-valueprojectsthatneverthelessconsume time, energy, andmoney.Killing themhelps reduceneglectedworkand allows you to deliver more important projects faster and with fewerinterruptions.

Even though neglectedwork is not a priority, it still consumes somementalbudget. Like that reminder on the fridge to schedule your annual physical,Thief Neglected Work stalks you every so often, distracting you just longenough to divert your attention away fromwhat you should be focusing on.Here’swhatyouneedtodotogetThiefNeglectedWorkoffyourbackinthreesteps:

1. Acknowledgetheneglectedworksittingonyourboardthat’snotmovingandtheimpactithasonotherwork.

2. To create space to finish the most important work, lower WIP to a discussion on priorities.Something has to give. Either givework the attention it needs, kill it, ormove it back to thebacklog.

3. Ruthlesslyprotect your timeusing someof the techniquesmentioned inSection2.4 and stopstartingnewworkbeforefinishingoldwork.AsauthorArneRoocksaysinhisbookStopStarting,startfinishing!

EXERCISE

CreateanAgingReport

PURPOSE:Toimprovetheflowofvaluebymakingstaleworkvisible.AnagingreportshowsWIPthatisstale.

Time:40to60minutes

MATERIALS

LargewhiteboardorlargepaperorwallspaceStickynotesMarkersComputer(and,ofcourse,apizza)

INSTRUCTIONS: Query your work-tracking tool to find high-priority, yetpartiallycompletedworkthathasnotmoved

or been updated in thirty days. If thirty days results in a huge list, thenincreasetosixtyorninetydays.Randomlyselectseventoelevenofthose

highpriorityitems.Statistically,seventoelevenissufficient—aslongasit’strulyarandomsampleset.

Foreachoftheseventoelevenitems,notethefollowing:

The number of days the item has been stale (not been updated or moved or made anyprogress).Theaveragecycletimeforcardsofasimilarworkitemtype.Howmanydaystheitemhasbeeninprogressincomparisontoothersimilaritems.

Now, write down what happens if this item continues to be delayed foranother week. Consider lowering your WIP limit and reprioritizing yourWIPbasedonwhatmighthappeniftheworkisdelayed.Identifytheutmostvaluablework currently on your board and separate out the lower valueitems.Ifpossible,doanimprovementblitztopushthroughtheonehighestpriority item to get it delivered. Improve flow by making work visible.Remember the goal is to improve flow bymaking important, stale workvisible.

KEYTAKEAWAYS

Importantworkthatisdelayedbecomesurgent,unplannedwork.Visualizedelays.Calloutneglectedworkandpurgelow-valueprojects.

Youdon’tlearntowalkbyfollowingrules.Youlearnbydoingandfallingover.—RichardBranson

2.7

USEFULBOARDDESIGNEXAMPLESInthissection,youwillfindexamplesofkanbanboarddesignsthatareusefulfor making work visible in situations where time thieves cause particulardamage.Thinkofthissectionasakanbansmorgasboard.Selectwhatworksforyourcontextandpassbywhatdoesn’t.

MultilevelBoardDesign

In the same way that cross-team boards make hand-offs visible, multilevelkanbandesignsmakemultipleprojectsandcross-functionalteamworkvisible.Amultilevelboarddesignprovidesabig-pictureviewfromtheportfolioleveltothe team level, providing visibility of all the WIP. In Figure 29, high levelprograms one and two are connected to team boards where the work isperformed.Teamsbreakdowntheirworkintosmallerbatchesandcreateworkitems.

Figure29.MultiLevelBoardDesign

If the work is handed off to another team, say from Development toOperations, then the parentwork items are connected to cards on the otherteamboard.Thissignalshand-offsandmakesworkvisible.It’snotuncommonto see organizations with three levels of boards: a high-level board for theportfolio level, amid-levelboard fordifferentprogramsorvalue streams, andlower-levelboardsforteamwork.

Donevs.DoneDone

It’sthelaststepoftheprocessandworkishandedofftosomeoneelsetowrapupthelooseends.Wecanconsidertheworkdone—right?Notsofast.There’sachancethatyoumightwanttoholdoffonmovingtheworktoDoneinordertogetvisibilityonworkthatisn’tprovidinganyvaluetotherequestoryet.

Think of a box of cereal sitting on a grocery store shelf. Corn flakes don’tprovide any value to Kellogg’s until a customer buys them. Like inventorysittingonashelf,anewlydevelopedfeatureorabugfixdoesn’tprovidemuchvaluetotherequestoruntiltheycangettheirhandsonit.

Figure30.DoneVs.DoneDone

ThisiswheretheDoneDonelanecomesintoplay.Teamswhowantvisibilityonworkthathasyettoprovidevaluetosomeone(eitheracustomerorinternalteammember) use the singleDone lane to visualize “inventory”work.WorkonlymovestotheDoneDonelaneonceithasachieveditsrealgoal.

OnecouldarguethattheMarketingteamgetsvaluefromthesingleDone,asitsignals the pointwhere they can amp upmarket buzz in anticipation of theDoneDone.

Customers, however, don’t give a rip about the singleDone. To them, a newfeatureisn’tdoneuntilit’sworkingrightinProduction.DoneDonesignalstherealfinishline,oftenquiteawhileafterthecodeisconsideredshippableorevendeliveredtoproduction.

ThePDCABoard

W.EdwardsDeming,authorofOutoftheCrisis,isthemanwho,inthe1950s,taught hundreds of Japanese engineers and businessmen statistical processcontrol (SPC) and concepts of quality. Japanese manufacturers applied histechniques and went on to experience unheard of levels of quality andproductivity.Demingusedaniterative,four-stepapproachforchange,problemsolving, and continuous improvement of processes and products known asPDCA (Plan-Do-Check-Act). For the Edwards Deming fans, Figure 31 is anexampleofaPDCAboarddesignthatallowsyoutoseeworkflowthroughtheseiterativestates.

Figure31.Plan-Do-Check-ActBoard

TheHomeProjectBoard

Technologydoesn’town theproblemof toomuchWIP.Earlier in thebook, Imentionedthatpeopleusuallysayyestotheirspouses,whichgoesbacktoourdesire to say yeswhenwe like thepeoplewho ask.This promptedourhomeprojectboard(Figure32):

Figure32.HomeProjectBoard

We grouped home project work into three categories, showing problems inyellow, maintenance in green, and everything else in blue. This work flowsthroughthissimple, five laneboard,whichshowswhatournextprojectsare,whatwe’re currentlyworkingon,whatwe’ve just finisheddoingand isbeingchecked,andwhatweagreeisdone.Thetop-fivecolumnisourcurrentpriorityof work. The backlog is huge and not shown here. It doesn’t make sense toprioritizeoreventalkaboutthestuff inthebacklogbecausethingswill likelychangebythetimewegettoworkingonit.

Remember the story from earlier in the bookwhen I askedmyhusband thisreally stupid question about building a greenhouse while he was perilouslyperchedontopoftheoldcrumblingoutbuilding?Well,that’swhywehaveWIPlimitsnow inDoing andNext.You see, I can’tpull inor even talk about theupcomingworkuntiltheworkinDoingisintheCheckcolumn.That’sabigwinforthehubs.Buthere’swhatIgetoutof it—thehubscan’tsaysomething isdoneuntilIgettocheckitout.It’sawin-win!

ManagingYourMove

Whilewe’re on the topic of using kanbanboards around thehome, here is aboard that JuliaWester used tomanage her and her husband’smost recent

move(Figure331).Thisoneisinterestingbecausetheynotonlyusedtheboardtoprioritizethemoveandtracktheirmovingprogressbutalsotookitonestepfurtherandusedthecardstovisualizehowmucheachitemcost,givingthemvisibilityontheoverallcostoftheirmove.

Figure33.ManageYourMoveBoard

RepetitiveTasks

There are a number ofways to visualize repetitive tasks. I’m inspired by thesequesteringapproachthatJimBensonandTonianneDeMariaBarrydescribe

intheirbookPersonalKanban:MappingWork,NavigatingLife.2

It’sasimple,yetelegantapproachdesignedtohandlerepetitivetasksthatcanclutteryourboard.I’vetakenthelibertytomodifythedesigntoprovideabitmoredatathatelectronictoolscanprovide ifyouareusinganelectronictoolversusaphysicalboard(Figure34).

Figure34.RepetitiveTasks

The board is split in half horizontally. Repetitive tasks are shown on thebottom half and standard work appears on the top half. Template cards arecreatedfortherepetitivework.Thetemplatecardisprepopulatedwithalltheinfo.Allyouhavetodoiscopythetemplateandpopitinthedoinglane.Theoverhead of filling out fields is gone, alongwith the excuse for keeping thatworkinvisible.

Thefirstexampleunder“Templates”inFigure34isformeetings.Here,ateamofninepeoplemeetdaily for a stand-up. Insteadof creatinga card for everystand-up,thissinglecardsitsintheDoingcolumnfortheweek.Thenumbers

on the card show the number of people in attendance (nine) alongwith thenumber of stand-up meetings (incremented for each stand-up during theweek).

Similarly, a training template card can be easily copied and used again andagaintokeeptrackofallthesmallrepetitivetasksrequiredtodotraining.

Separate repetitive tasks in a dedicated area of your kanban board. It’simportant to keep these tasks visible because they increase WIP, and theirimpact should be acknowledged. Remember Operations Manager Erik fromSection2.4who got clobberedbyThiefUnplannedWork?Don’t be likeErik,friends.

PurchaseOrderBoardDesign

SometimesIthinkonecouldsailaroundtheworldinthetimeittakestogetapurchaseorder(PO)processed.EveryoneItalktowhoisworkingseriouslytoimplement Lean thinking in their organization eventually bumps up againsttheir accounting systems. It has become painfully clear that traditionalaccounting systems are restrictive and sloooooow. Finance departments areoneofthelastremainingholdoutsinLeantransformations.

Thefivetimethievesmakethemselvesathomeintraditionalcom-paniesthatdrive theirbusinessbasedoncostsandmargins. It’sharder for themtostealtime fromLean organizations that focus on customer and/or business value.The good news is that there are some CFOs out there who get it and areevolvingfromtraditionalproject-basedresourceallocationandcostaccountingto a faster, lower-overhead financial management model. Yippee! (For someexamples,checkoutBrianH.Maskell,BruceBaggaley,andLawrenceGrasso’sPracticalLeanAccounting:AProvenSystemforMeasuringandManagingtheLean

Enterprise.)

Ifyou’restillsufferingfromaccountingwoes,considermeasuringthetimeyourworkwaits in financequeues. (Likehow long it sometimes takes topayyourfavoriteconsultant.)

TakealookatthetwohorizontalswimlanesinFigure35.Oneisfortheworkthatdoesn’trequireaPOandoneisfortheworkthatneedsaPO.TheWaitingonApprovalandPOcolumnsbringvisibilitytoworkthat issitting idlewhilewaitingonFinance.Remember,thegoalofkanbanistomakeproblemsvisibleso they can be fixed. In this example, the time the work spends waiting onFinance can bemeasured and used as supporting data to help your Financedepartmentseethecostofthispractice.

Figure35.PurchaseOrderBoardDesign

Likewise, professional adults struggle to keep on top of everything, yet weadmonish young people when they don’t. Between schoolwork, homework,chores,andextracurricularactivities,studentsjugglemanyactivities—someofwhichgetlostsomewhereinthecalendar.Thisleadstoaflurryoflast-minutework crammed into a late night tomeet the deadline. The result is a subpar

experienceforallinvolved.

Figure36.StudentBoard

Theadditional stress thatcomeswithprocrastination isavoidable. Insteadofthe out-of-sight, out-of-mind that occurs when long-term assignments arehiddenondifferentcalendarpages,makeallthestudentthingsvisibleononepage.

Nowthatwe’veexploredsomeofthemanywayswecanempowerourselvestocalltimethievesoutintotheopenbymakingourworkvisible,let’stakethingsastepfurther:What, forexample,dowedowithall thisknowledgenowthatwe’vestartedtoimplementthesepractices?Howdoweprovetothehigher-upsthat our boards and visualizations andnewpractices areworking?Whydoesthismatter?Readontofindout.

KEYTAKEAWAYS

Multilevelboardsprovideabig-pictureviewofworkflowacrosstheorganizationasawhole.Smalltasksmaynotneedtobetrackedonaboard,butconsiderifaddingthemtotheboardishelpfultopromotecross-training,givevisibilitytosomeone’swork,ornotifyanotherteamabouttheexistenceofadependency.POboardscanprovidethemetricsneededtogetbuy-inonchangestotraditionalaccountingsystems.Kanbanboardscanbeappliedtonon-worksituationstoo!Fromhomeprojectstomovingtostudentactivities,kanbanbringsasmoothflowtohomelife.

Beapproximatelyrightratherthanexactlywrong.—JohnTukey

3.1

YOURMETRICSORYOURMONEYHere, we’re going to be using metrics to set better expectations for ourcustomers,whoalwaysseemtobecriticizingthelengthoftimethingstaketocomplete and complaining that they can’t see the progress of their requests,sincemakingworkvisibleisthesecrettoexposingthetimethievesandallthehavoctheywreakonyouandyourorganization.Capturingdataformetricscanhelpyoubecomethevoiceofreasonandpromptchangeinyourorganization.Metricsgetattention.

Because many teams use arbitrary due dates to set expectations and thenultimatelyfailtohitthatdeadline,it’stimetotryasmarterapproach.Namely,aprobabilisticapproach.

For those of you panicking at the idea of no due dates, know that they areappropriateinsomesituations.Thecharacteristicsthatinformusofwhetheraduedateisarbitraryornotdependonthelensthatwelookthrough.

The 2018 FIFA World Cup, scheduled to start June 14, 2018, is one suchexample of a real due date. So is a scheduled security audit with the federalgovernment.

A date that the CFO selects to roll out a new CRM system so that half theAccounting team can be laid off is a suspect due date. A change to a UXenhancement? Almost certainly arbitrary. It probably does not matter that

muchifitshipsonTuesdayoronThursdayorevennextweekonMonday.

If expectations are set correctly, not everything needs a due date. Beingpredictableiswhatcounts.Beingpredicablesavestime.

Becomingmorepredictablemeansweneedtotalkaboutprobabilityandhow,onceweareabletoshifttoamoreprobabilisticapproach,expectationsaroundtimeframes can be improved. It’s all about the expectations. Setting betterexpectations makes leadership happy and helps make those uncomfortablestand-upsandretrospectivessurprisinglyfun.

Goodmetrics help us to see progress and understand how long things reallytake.This is importantbecause thebiggest complaintheard from technologycustomersisthatthingstaketoolong.Withmetrics,wecandemonstratejusthow long things actually do take and then extrapolate on why they took solong.

Again, the problem usually starts with an arbitrary due date. That due date(oftenbasedonfaultyestimates)setsthewrongexpectationandthen,despiteall our best efforts, everything goes to hell because of conflicting priorities,unknown dependencies, and that curveball unplanned work, which alwayscomesoutofnowhere.

Everythingtakeslongerthanwethinkitwill,especiallyiftheworkiscomplex.Hofstadter’sLawisastatementmockingtheaccuracyofestimatingcompletion

fortaskswithsubstantialcomplexity.1(Italwaystakeslongerthanyouexpect,evenwhenyoutakeintoaccountHofstadter’sLaw.)Weknowthisistruefromourownexperience.Whenyourequestsomething,howoftendoyoureceiveitearlierthanexpected?Ifyoufrequentlydo,callme—Iwanttocomeworkforyou.

Throughout my career, no matter my role or what company, feedback from

customers has invariably noted that projects take too long to complete, newfeaturestaketoolongtobedelivered,andnewcloudcomputingplatformstaketoo long to get set up. Everyone wants stuff sooner than they get it, andeveryoneisunhappyaboutthedelays.Yes,thatisthetimethievescacklingtothemselvesinthebackgroundastheycapitalizeonourdistress.

Thecascadeofdelaysonceaduedatehasbeendetermined(oftenbyasalesormarketing executive), usually starts with the Development team.While theybegintheprocessthinkingtheyhavesufficienttimetodevelopanddeliverthefeature,alltoooftentheyfallbehindasthingsdon’tgoasplanned.Maybetheydidn’t know about the dependencies on another web team, or they didn’texpecttheirrelationaldatabasemanagementsystemtobesolimiting,ortheyhadn’t plannedon their coreAPIdeveloper getting sick.Whatever the cause,theybegintomissduedateswhentheseunanticipatedobstaclespreventthemfromstayingonschedule.

FurtherdelayshappeninOperations,suchaswhenachangethatwaspromisedtohavezero impactonproductionactuallydoeshavean impact,orwhenthedatabase upgrade usurps an important cross-teamplanning session, orwhentheautomationengineerleavestopursuetheirPhDinphysics.

It’s tough tomake decisions about delivery timeframes and due dates in theabsence of compelling evidence. A lack of data often leads to peoplemakingdecisions based on opinions, which are likely to be more problematic thandecisionsbasedoffofgooddata(i.e.,visiblework).Wouldyourathertaketheadvice of a licensed structural engineerwith ten years of seismic retrofittingexperienceoryourbrother-in-lawaccountantwhoswearsthathisDIYretrofitmethod is best? This is whymetrics are so useful—they help usmake gooddecisions.

Imagine trying to fly a plane without a fuel gauge, compass, or airspeedindicator.How riskywould that be for a pilot?There’s a reason for all those

flight instruments in the cockpit.Without them, thepilotwouldhaveahardtimegettinginformationaboutthefuellevel,airspeed,anddirection,whichinturnwould impacttheflighttowerastheytrytoguidemultipleplanes inforlandings. Similarly, lack of visible data in IT makes us blind to problemsbecausewedon’thaveanythingtotellushowweareactuallydoing.Whenwecan’t see the problems, it’s hard to analyze them, which in turn makes itdifficult to knowwhat direction to take. This iswhat goodmetrics do—theysteerusintherightdirection.

When it comes to forecasting how long things are going to take (rememberHofstadter’sLaw),it’susefultolookatmetricsthatmeasureprogressinsteadof activities. Some of the best metrics that show actual progress (or lackthereof)are leadtime,cycletime,WIP,andagingreports.Fortherestofthissection,thosearethemetricswewillconcentrateon.

FlowMetrics

Theboss (or thecustomer/spouse/teacher/coach)wants toknowwhenXwillbe done. The way many people go about answering this question is todeterminehowlongeachstepoftheprocesswilltakeandthenaddthemallup.Usually, a contingency buffer is added in for good measure because thingsalways take longer. Humans are predictably horrible at estimation, even intheirareaofexpertise,myselfincluded.

I estimated that the seismic retrofitmy husband and I did in our basementwouldtakefourweeks—ittooksixweeks.Wedidn’tconsiderscenariossuchashowlongitwouldtakeustoreplacegasandwaterpipesinneedofconversionfromhardmetal to flexiblehoses, even thoughwe’vepreviously done lots ofgasandwaterpipeworkinthepast.

Optimism is a near universal human trait when it comes to answering thequestion “When will it be done?” Software estimates are no different.Developerspresentoptimisticestimatesthatmanagementapprovesbecauseofthe implied achievable business targets. Furthermore, we take onmore workbecauseweareanoptimisticbunch—it’soneofthefivereasonsforsayingyesthatwerementionedinPart1.

The problem with the traditional estimation process (add up how long eachpartoftheprocesswilltakeandthenaddabuffer)isthateachstepinaprocessonitswaytocompletioniscloudedwithuncertainty.Eachstepisvulnerabletodistractions fromThiefUnknownDependencies,ThiefUnplannedWork, andThiefConflictingPriorities,alongwithholidays,snowdays,andaboutamillionotherthings.Estimatesarefilledwithmoreuncertaintythanconfidence.

The estimation process matters because, traditionally, people are heldaccountablefornotablyhorribleestimateswhenprojectsfailtobedeliveredontime,whichismostofthetime.TheNoEstimatesmovementgainsmomentumdailyasmoreandmoreprojectsfinishlateandmoreandmorestafffailtomeettheirgoals.Whattodo?Looktoflowtimeforhelp.

Flowtime isameasureofhow longsomething took todo frombeginning toend. You might be thinking, “Wait, that’s cycle time.” And you’d be right.Although—it depends on the context as to which definition you use.Dependingonwhomyouask,cycletimehasdifferentmeanings,whichI’llgetto shortly. Just know that cycle time is an ambiguous term and that’swhy Iprefertouseflowtimewhendiscussingspeedmetricsingeneral,becauseitisattunedwithLean.It’sactuallyamainpillarofLean.Theterm“flowtime”hasbeenaroundforawhile—Ididn’tmakeitup.

Flowtimehasastarttimeandanendtime.That’sall.Flowtimedoesn’tstoptheclock justbecause theweekendrollsaround. Itdoesn’tdostartandstop,start and stop, start and stop. What flow time does do is quantify the

probabilityofcompletingx%ofworkinsomanydays.

Figure38.FlowTimeMetrics

Collectinghistorical flowtimes thatshow, forexample, that90%ofa certaintypeofworkgetsdeliveredwithintendaysallowsustosaythatnineoutoftentimes,wedeliverthesekindsofrequestswithintendays.Weknowthenthatthere isa10%probability that someworkwill take longer.This is importantbecauseithelpsusbecomemorepredictablewithourcustomers.

Lead time and cycle time are types of flow timemetrics. They bothmeasureduration. Using pizza order and delivery as an example, the lead time clockstarts tickingwhen the customer orders thepizza,while the cycle time clockdoesn’tstarttickinguntilthecookbeginsmakingthepizza.Peoplewhoorderpizza care about lead time. They want their pizza delivered quickly. Internalteams care about cycle time.They try to reduce thewait time in thedeliverypipeline to be more efficient. Lean organizations optimize for speed andeffectiveness.

Figure39.LeadTimeandCycleTime

Traditionally, in a manufacturing sense, cycle time is calculated as a ratioresulting in theaverage timebetweencompletionofunits.Forourpurposes,I’m defining it as the elapsed time from when you start work until thecompletedunithasbeendeliveredtothecustomer.Thisishowmanysoftwaretechnologyorganizationsdefineit.Likeleadtime,cycletimealsoquantifiestheprobabilityofwhenworkwillbecompleted.Theclockjuststartstickinglater.Cycletimeisimportantbecauseitrevealshowlongthingstakeinternallyonceworkhasbeenstarted.Onecanseehowwaitingondependencies fromotherteamscanimpacttheschedule.

TheoddsofbeingpredictabledecreasewhenWIPconstantlyincreasesandflowtimes elongate. Remember—WIP is ameasure of howmany different thingsarebeingjuggledatthesametime.Unlikemostothermetrics,WIPisaleadingindicator. The more WIP there is in the pipeline, the longer things take tocomplete,period.Wecan lookatLittle’sLawtounderstandthemathbehindwhy WIP extends completion times. Recall that lead time equals WIP overthroughput.GivenWIPisthenumeratorofthatfraction,weknowthatwhenWIP goes up, so does lead time. Algebra and theory aside, the proof is inmeasuringtheday-to-dayexperience.

Little’sLawcomeswithsomeassumptionsthough.DanielVacantitalksabout

these inhisbookActionableAgileMetrics forPredictability:An Introduction.Hediscusses how the true power of Little’s Law lies in understanding the

assumptionsnecessary for the law towork in the firstplace.2Allmetrics arebasedon assumptions, andLittle’s Law isnodifferent.All youhave todo todiscreditametricistoquestiontheassumptions.Inorderforyourmetricstobetakenseriously,carefullyconsiderandidentifytheassumptionsinplace.

A Training team that concentrates together on producing trainingmaterialsprogressesfasteronthetrainingcollateralduringweekswhentheyarenotalsotraveling to customer sites and speaking at conferences. A Marketing teamprogresses fasterwhen theywork on seven initiatives at one time insteadofthirteen. College students finish their homework soonerwhen they take twoclasses instead of three classes. One can argue that it depends on thecomplexity of thework. The homework for three freshman-level classesmaytake less timetocomplete thanthehomework for twograduate-level classes,and this is why categorizing types of work is important. When work iscategorized,youcangetfancyandobtainWIPreportsforeachworkcategory,whichinturncanimproveyourWIPallocations.

Figure40.TheWIPReport

QueuingTheory

Whydosomedoctor’sofficesalwayshavelongwaittimesandothersdon’t?Isit because the busy doctors are better and have more patients? In myobservation, that hasn’t been the case. I’ve waited just as long for a crappydoctorasforgooddoctor. Ioncewaitedanhourforadoctortotellmetherewasn’t anything I could do to improve my atrophied quad muscles afterdislocatingmykneecap.

I’mconvincednowthatthedoctor’sofficesthatdon’thavelongwaittimesaretheonesthatunderstandwhy100%capacityutilizationdoesn’twork.

There is a direct correlation between WIP and capacity utilization. Capacityutilizationisthepercentageofthetotalpossiblecapacitythatisactuallybeingused.Ifthedoctorisintheofficefortenhoursandhastenhoursofscheduledappointments, then she is considered to be fully loaded at 100% capacityutilization.Ifsheisintheofficefortenhoursandhassevenhoursofscheduledappointments,thensheisloadedat70%capacityutilization.

Whathappenswhenacustomerwithasorestomachcallsandneedstobeseenthat day? We’ve all been there, right? I’ve been there. We’re sitting in thewaitingroomforthirtyminutesforanappointmentwemadetwodaysbefore,when someonenewwalks in, checks in, and sits down.Another tenminutespass,andthereceptionistcallsinthenewguybeforeme.Thoseofuswhohavenow been waiting for forty minutes roll our eyes at each other, sigh, andgrumble internally about how this is fortyminutes of your life you can’t getback.

Attempting to loadpeople and resources to100%capacityutilization createswaittimes.Thehigherutilization,thelongerthewait,especiallyinfieldswithhighvariability,likeIT.

Note that when I say “variability,” I refer to a lack of consistency and howthingsaresubjecttochange,suchasanunexpectedevent.AnexampleiswhenThiefUnplannedWorkdisruptsyourdeployment, andanetworkswitchgoesbad,takingout200or2,000servers.Orwhensomeonehackstheirwayontoour now unsecure DB servers. Or when Brent calls in sick—and cuts in lineaheadofusatthedoctor’swaitingroom!

Unpredictable events cause variability, and the more variability, the morevulnerable we are to capacity overload. The more people and resources areutilized, the higher the cost and the risk. Computers stop responding whentheygetcloseto100%utilization.Freewaysclogupandslowwaydownwhenthey are fully utilized. I like to think that good doctor’s offices understandqueuing theory and allow for variability for thedrop-inpatients.Themore apersonorresource isutilized,thebiggerthe lines(queues)get.Andwhile it’sintuitiveatsomelevel,thereissomesciencebehindit.

It’s called queueing theory.When we look at themath, we can see why thesinglemostimportantfactorthataffectsqueuesizeiscapacityutilization.Thereasonwe care about queue size is because the bigger the queue, the longerthings take.What I’vedrawn is the curvedescribedby theKingman’s formula(anapproximationfortheaveragewaitingtime inaqueue).What itshows istherelationbetweenutilizationandwaittime(Figure41).

Figure41.QueuingTheory

Queueing theory is a field of applied statistics that studies waiting lines. Itallowsustoquantifyrelationshipsbetweenwaittimesandcapacityutilization,even when arrivals and service times are highly variable. If requests arrivefasterthanthesystemcanservicethem,theyqueueup.

Aswemovefrom60–80%utilization,thequeuedoubles.Aswemovefrom80–

90%utilization, thequeuedoublesagain.Andagain from90–95%.3Onceweget past 80% utilization, the queue size begins to increase almostexponentially, slowing things down to a grinding halt as it pushes 100%capacityutilization.

Haveyoueverworkedforacompanythathasa20%“creative”timepolicy?I’veread that the main reason they do that is not for innovation (that’s just a

bonus)buttokeepcapacityutilizationat80%ratherthanat100%.4In1948,3M gave its workforce 15% slack time, which years later resulted in sticky

notes.5

Wedon’tletourserversgetto100%capacityutilization,solet’snotdothattoourselves.

WatchtheWork,NotthePeople

Agingreportsrevealhowlongworkhasbeensittinginthepipelinenotgettingdone(Figure42).Lookingattheworkthat’sbeeninthesystemformorethansixty days (or ninety or onehundred twenty days) shines a valuable light onhowmuchwasteisinthesystem.

Figure42.AgingReport

Whenpeopledescribesituationsintheabsenceofcompellingevidenceinorderto persuade others to agree with them or take some kind of action, they’rerelyingontheirowncredibility.Dependingonthedynamicsoftheteam,thiskind of anecdotal evidence can lead to doubt, skepticism, or even suspicion.And this is why quantitative measures are good—they are usually moreaccurate than personal perceptions and experiences. Good metrics help usmakegooddecisions.

When itcomestoefficiency, time iswastedwhenthere is toomuchfocusonresourceefficiencyoverflowefficiency.

Figure43.FlowEfficiency

Good metrics help others see a clearer picture and help set more accurateexpectationswhenitcomestoquestionslike,“Whenwillitbedone?”Duedatesdon’ttakewaittimeintoconsideration.Andtheproblemisusuallynotintheprocess time—it’s in the wait time. Focus on the wait time and not on theprocess time.What should the batch size of items be for delivery purposes?Whatistheoptimaldeliveryrate?

Therearetwofactorsinvolvedinthisthatweneedtolookat:

1. Thereisacostfromdelayedfeedbackandfromdelayedrelease.2. There is a cost to initially set up and work on the item, plus transaction or coordination

overhead.

Figure44.OptimalBatchSize

The optimal batch size for delivery then depends on the combination of theimpactofeconomiesofscaleandthecostofdelayingresponses(holdingcostandtransactioncost).

Some people have a bias for large batch sizes because of the concept ofeconomies of scale. Economies of scale is the cost advantage that arises withincreased output of a product. The cost advantage is seen in some areas ofmanufacturing. Boeing produces large quantities of a single product on itsassembly line.The transactioncost tocreateairplaneenginesneeded for justoneairplaneatatimeistoohigh,sotheycreatealargerbatchofenginesatonetimetoreducetheassociatedoverheadcost.

Focusingonefficiencyproducesbettercostaccountingresultsforlargebatch-sizeprojects,suchasmanufacturingcommercialairplaneenginesorpublishingbooks. In knowledge work, however, problems with coordination costs grownonlinearly with batch size. Old school management assumptions abouteconomyof scaledonotapply toknowledgeworkproblemssuchas softwaredevelopment.

Mylittlewoodstoveisworkingbetterformenow,eventhoughIdidn’treplaceitwithanew,bigger stove. I changedmyworking routine toPomodorostyleworkingsessionsandsetatimerforthirtytoforty-fiveminuteswhereIfocusheadsdownuntiltheringergoesoff.EnoughhotcoalsremaininthestovethatIcankeepthefiregoingbytossinginmorewood.ThenIhaveanotherthirtytoforty-fiveminutesof“don’tinterruptme”time.ThislittlebreakofthreetofiveminuteshelpsmemorethanIrealized.

Istandupandstretch,evaluatehowmuchIaccomplished(ordidn’t),andaminspired to do better with the next thirty to forty-five minutes. Previously,whenworkinginninetytoonehundredtwentyminutechunksoftime,I’dletmyself spend toomuch time on things that were not essential. The shorterwindow provides a sense of urgency to get something done faster, and itencouragesmetobreakdownworkintosmallerchunks.Ultimately,Iammoreefficient.

Imagineyou’regroceryshoppingforbananas.Ifyoubuyasix-monthsupplyofbananasatonetime,yourtransactioncostislow,butmostofthebananaswillberottenwithintendays,soyou’vewastedmoney.Ifyoubuyaone-daysupplyofbananasatonetime,theywon’trot,butyourtransactioncostswillbehigh,because you’ll be grocery shopping every day. Somewhere in between is therightbatchsizeofbananas.

ThereductionofbatchsizeisacriticalprincipleofLeanmanufacturing.Smallbatchesallowmanufacturerstoslashworkinprocessandacceleratefeedback,which,inturn,improvescycletimes,quality,andefficiency.Smallbatcheshaveanevengreateradvantageinsoftwaredevelopmentbecausecodeishardtoseeandspoilsquicklyifnotintegratedintoproduction.

Oneofthebestpredictorsofshortleadtimesissmallbatchsizesofwork.TheaverageamountofcurrentWIPisdirectlyproportionaltobatchsize.LikeWIPlimits, smallbatchsize isanenablingconstraint.Workbatchedup insmaller

sizesconstrainstheamountofworkneededtobecompletedbeforereceivingfeedback.Fasterfeedbackmakesforabetteroutcome.

Smallbatchsizesenablefastandpredictableleadtimesinmostvaluestreams,whichiswhythereisarelentlessfocusoncreatingasmoothandevenflowofwork.

Whenyoucan implementthepracticesabove inyourworkand lifeandshowhowthesepracticessaveyoutime,money,andstressthroughmetrics,youaremoreabletoproactivelybanishthetimethievesfromyoursphere.

KEYTAKEAWAYS

Delaysarecommon;usemetrics,particularlyflowmetrics,tohelpyoumakegooddecisionsonpriorities,WIPlimits,andcapacityutilization.Stoplettingyourselfandyourteamreach100%capacityutilization.Lookfortheoptimalbatchsizetohelpyouachieveefficiencywhilekeepingtransactioncostsdown.

Investenergyincollectingmetricsthathelpyoumakedecisions.—EricRies

3.2

THETIMETHIEFO’GRAMLadiesandgents,IgiveyoutheTimeThiefO’Gram.

Think of this tool as a spotlight, shining a light on a criminal lineup of theuncertaintyacrossyourorganization.

The intent here is to look atmetrics that reveal high risk. Commonmetricssuchasthenumberofstorypointscompletedoveraperiodoftime(velocity),the number of bugs that escaped into production, or the number ofdeploymentstoproductiondon’trevealverymuchaboutrisk.

I’mgoingtoargueherethattimetheftshouldbemeasuredbythethingsthatcause theproblems in the firstplaceandpreventyour teams fromdeliveringquality work quickly: too much WIP, unplanned work, neglected work,conflictingpriorities,andunknowndependencies.Thesearethecausesbehindmediocreworkdonetooslowly.

Eachofthesetimethievescanbecapturedusingworkitemtypesandcardtagsoracombinationofboth.Ifwecancountthesethingsandmeasurethem,wecanseeboththestolentimeandwhichthiefstoleit.It’sgoodtoknowifThiefUnplannedWorkdiditinthekitchenwiththeknifeorifitwasThiefTooMuchWIP in the library with the candlestick. The Time Thief O’Gram shows whocommittedthecrimeandrevealshowmuchwasstolen.

Figure45.TheOriginalTimeThiefO’GramTheTimeThiefO’Gramcanbegeneratedbytaggingallyourthievesonyourkanbanboard.Inthisexample,eachthiefhasbeencolorcodedtomakeitevenmorevisual.Onceyouhaveallthesetagged,youcanvisualize,count,andtrackthemweekoverweekor

monthovermonth.Indoingso,youcanbegintoseepatternsandconnectionsthatwouldotherwisebescatteredacrossmultipleteamboards.

We can look at the congregated Time Thief O’Gram (Figure 46) to comparethievesandtheirtrendsweekoverweektoseethefluctuationintheamountoftoomuchWIPacrossalltheteamsovertime.

Figure46.CongregatedTimeThiefO’GramThereis,ofcourse,avarietyofwaystomaketimethieveryvisible.Figure47showsabalancedscorecardtohelppeopledeterminewhichthievestofocusonfirst,

whichonestheyaredoingwellwith,andwhichonesarerobbingthemoftimeandpredictability.Trackingandmeasuringthesethievesputspeopleinapositiontodosomethingaboutthem.

Figure47.BalancedScorecard

Inmywork,IamfrequentlyaskedaboutthebestwaytoinfluencetheCIOtouseaDevOps/Lean/kanban/whateverapproach.TheCIOs I’ve talked towanttwothings:lessriskandmorepredictability.TheTimeThiefO’Gramshowsyouthat. It shows risk (*ahem* uncertainties) with respect to unplanned work,neglected work, conflicting priorities, and unknown dependencies. The TimeThiefO’Gramalso showswhenWIPexceeds limits (basedon the team’sownsetWIP limits). BecauseWIP is a leading indicator, trackingWIP levels cansignalproblemsearlyandwarnyouthat thingsaregoingto take longer thanexpected.Ifpredictabilityfordeliveringontimeisimportant,thenwhatwouldit be worth to your CIO to have this kind of transparency? Effectivecommunicationrequireskeymessagesfromdifferentsources.LooktotheTimeThiefO’Gramasoneconsistentsourceofimportantmessages.

KEYTAKEAWAYS

TheTimeThiefO’Gramrevealswhichthievesareatworkinyourorganizationandhowmuchtheyarestealing.ThemetricsfromtheTimeThiefO’Gramprovidetransparencyforleaderseagertoseetheissues facing teams. These are an excellent source for getting executive buy-in to do thethingsnecessarytoimprovepredictabilityandreducerisk.

Lettheflowmanagetheprocesses,andnotletmanagementmanagetheflow.—TaiichiOhno

3.3

OPERATIONSREVIEWDavidAndersonintroducedustooperationsreviewsatCorbis.Everyoneonhismanagement team had to present their team’s metrics. It was my firstexperience as a team lead presentingmetrics to a group ofmore than thirtypeople. I was terrified standing in the conference room:my voice shook,myheart pounded, and I thought I was going to maybe throw up. What if mypresentationlettheteamdown?

Thisexperiencetaughtmethatmanagementresponsibilities includeknowingthedemandonmyteamandbeingabletopresentwhatthedemandlookedlikeinrelationtothecapabilityoftheteamtomeetthedemand.

Figure49isacumulativeflowdiagram(CFD)—astackedlinechartthatshowsthe amount of WIP and items delivered over time. I presented this CFDmonthly at the operations review on behalf of the Build and Release team.WhenJunerolledaround,incomingrequestsspiked,andbecauseIwasabletodemonstratethisspike,myrequestformoreheadcountwasapproved.

Figure48.CumulativeFlowDiagramforOpsReview

Thegoaloftherecurringmonthlyoperationsreviewsistocollectively lookatthe data to see and understand the health of the organization. A disciplinedandconsistentreviewoforganizationalhealthprovidesagreatopportunityforcontinuous improvement.Operations reviews provide feedback loops to helpyouunderstandhowtheorganizationisdoingandtoenableyoutomakesounddecisions concerning your next move. They are objective data-drivenretrospectivesontheorganization’sperformance.

Thereviewsareorganization-wideandinvolveseniorleaders,managers,leads,and individual contributors to communicate that the organization takesperformance seriously by setting an expectation for objective, data-driven,quantitativemanagement. Inotherwords,operations reviewsareoneway inwhichwecanmakeworkmorevisiblebyseeingwhathascomebefore.

Here are some logistics for how to run a successful operations review: Each

managerhasfiveminutestopresenttheirteam’smetricsforthemonth.Allowtwo to three minutes for questions or comments from the audience. Time-boxingthepresentationsandthefollowingquestionandanswersessionhelpskeep people focused on what’s important, gives speakers their fair share oftime, and avoids going off track or consuming too much time. Time-boxingavoids the common problem that occurs when you give someone a mic andcan’tgetthemoffthestage.

Sampleoperationsreviewagenda:

Openingremarksfromleadership.Presentationsfromteamleadsorfrontlinemanagers.Closingremarks.

MetricstoPresent

In order to understand how we are doing, what the risks are, and how toimprove predictability, Ops reviews show how the teams actually performedthepreviousmonthagainstpromises/expectations.

Whenbeginningoperations reviews, the suggestion is for each lead to reportonthefollowingmetricsanddataforthefirstfewmonths:

1. Throughput:Throughputishowmanythingswerecompletedoveraperiodoftime.Onewaytoshow throughput iswith aCFD.TheCFDwill also show the ratio of incoming requests againstcompletedrequestsandtheamountofWIPateachstageoftheworkflow.

2. Flowtime:Lookathowlongittookitemstomoveacrosstheboardfromthetimeworkwaspulledinto an in-progress column through delivery. The visual (hopefully a report easily generated byyourworkflow tool) shows the actual flow time for each completedwork item for the previousmonth.Thisisusefulforlearningsoyoucanmakechangestoyourprocesses,yoursystem,ortodecreasevariabilityandincreasepredictability.Or,ifthecardsizeissimilarinnature,wemayfinditinterestingtocalculatetheaveragecycletimeorleadtime.

3. Issuesandblockedworkitems:Identifyanymajorissuesorblockedworkthatispreventingtheteamfrommakingprogress.Thiswillhelppeopleunderstandwhythingstooksolongandknowwhatchangesareunderwaytopreventthoseproblemsfromhappeningagain.

4. TheTimeThiefO’Gram:Selectoneormoretimethievestoreportontoexposetheirtheft.

Additionalmetricsyoumaywanttoconsiderpresenting:AgingreportsCardtypedistributionFailureload(valuedemandversusfailuredemand)asameasureofqualityFlowefficiency

FutureOperationsReviews

For each metric, we want to track the trend over time so we can see theimprovements(ornot).Ifwearetodemonstratecontinuousimprovement,wewantthemeantrendtoimproveovertime.Todemonstratepredictability,wewant the spread of variation to decrease over time. For example, todemonstratethatonecanpredictablyarrivetoworkontime,ithelpswhenthefrequency of late arrivals decreases over time and the length of the latenessshrinks.

Anotherexample:AmtrakhasapassengertrainfromPortlandtoSeattlethatissupposed to run on a set schedule throughout the day. According to theschedule,thelasttrainofthedayissupposedtoarriveinSeattleat8:05p.m.Sometimesthetrainarrivesat8:25p.m.Othertimesitarrivesat2:30a.m.So,the train isunpredictable.Thewidespread inarrival timecausesvariation inthetrainsystem.

Thevariablearrivaltimesarecausedbyseveralfactors.Forexample,ourrainyPacificNorthwestweather generatesunpredictable landslides, blocking trainsuntil the rails are cleared (Thief Unplanned Work). Furthermore, becausefreighttrainshaveahigherprioritythanpassengertrains,andthereareonlysomany train tracks that run through the tunnel, the Amtrak train takes abackseattothefreighttrains(ThiefConflictingPriorities).

InorderforAmtraktodemonstratepredictability,theywouldneedtoaddresslandslidesandchangeprioritizationpoliciestodecreasethevariationofarrivaltimesofthetrainfromPortlandtoSeattle.

These are the types of decisions that operations reviews make possible.Withoutgood,objectivemetrics, it isverydifficulttounderstandexactlyhowthevariousthievesstealawayourtimeandenergy.Butbymakingworkvisible,we are able to see the patterns and communicate to our organizations justwheretheproblemslie,sowecanlearnandadjustandimprove.

KEYTAKEAWAYS

Operations reviews are an opportunity to present objective metrics that can form afoundationforimprovement.Usetime-boxingtokeepoperationsreviewsandspeakersfromrunninglong.Goodmetricsforoperationsreviewsincludethroughput,flowtime,andtimetheftalongwithissuesandblockeditems.Trackmetrics over time so that you can seewhat improvementshavebeenmadeor stillneedtobemade.

Frictionaroundprioritiesandpracticeshappensonthewaytoimprovement.—ScottNasello

3.4

THEARTOFTHEMEETINGWednesday,Seattle,9:00a.m.

Nine people sip coffee around a table at a South Lake Union coffee shop inSeattle.AlleyesareonCarmen,whoismakingapointabouttheimpactofthelatest round of layoffs at his company. The group is discussing howreorganizations impact different teams. People politely wait for Carmen tofinishhis sentencebefore voicing their opinion.Aminute later, a timer goesoff,andafterasimplevotefromeveryoneatthetable,thesubjectchangestothenextvisuallyqueueduptopicwrittenonastickynote, “Howto influenceleadership.”

ThisisLeanCoffee—astructuredmeetingwithfewrules.Participantsgather,buildanagenda,andbegintalking.

CreatedbyJimBensonandJeremyLightsmithin2009,LeanCoffeeemerged

as one of the greatest ways a group of people can discuss ideas.1 Theconversations are productive because the agenda for the meeting isdemocratically generated. People are engaged because they get to talk abouttopicsthatmattertothem.LeanCoffeeworksbecauseattendeesareinchargeoftheagendaandeveryone’svoiceisheard.Theminimalrulescombinedwithmutualrespectforeachotherprovidesasettingthatencouragesopendialogueandcollaboration.AdamYuret,whowroteHowtoHaveGreatMeetings:ALeanCoffeeBook,says,

LeanCoffeeturnstraditional,one-directionmanagementmeetingsonitsheadbyhelpingteamsuncoverthemostimportanttopicstothemajorityofpeople,byallowingeveryonetohearandtobeheard,andbyproviding

real-timefeedback.2

Iwill add to that by saying that LeanCoffee doesn’t just change the tone ofteammeetings,itshiftstheoverallculturewithintheenterpriseusingit.

Reversing conditionswhere time thieves thrive requires change.Manyof theproblemsrelatedtotimethieveryhavetodowithorganizationalproblemsorcompanyculture.Toputitanotherway,whenthecompanyhasaculturethatis focused on keeping people busy (instead of on keeping work flowing), itinvariably results in overloading people with too much WIP. This is not aproductive culture. We want to avoid the mistake of having a goal to keeppeople busy all the time when the goal should be to generate value for thebusiness.

Inorderforchangetooccur,people’sbehaviorsmustchange,and inordertochange behavior, the hearts and minds of people must be open to change.Casual, in-person conversationswith someonewith anopposing viewpoint isone of the easiest paths to changing someone’smind.Nothing accomplishesthisbetterthanapersonalrelationshipgeneratedbyface-to-faceconversationinasafe,calm,andrespectfulsettingsuchasLeanCoffee.

HowtoLeanCoffee

MyexperiencefacilitatingLeanCoffeessince 2012 has led me to develop adecentplanof action forgetting themgoing within your team ororganization.

First,blockoffsixtytoninetyminutes.

Next, gather up sticky notes andmarkers and place them around a table.Once everyone is seated, review theLeanCoffee rules: only one person speaks at a time, andparticipants shouldattempttolistenmorethantheytalk.

Then,inviteparticipantstotakejusttwotothreeminutesandusetheprovidedmaterialsto jotdownasmanytopicsastheywould like,but instructthemtowrite only one topic per sticky note (sticky notes are our friends!). Aftereveryoneisdonewriting,participantsshouldbriefly(twosentencesareusuallysufficient) summarize their own topic(s) for the group so others canunderstandwhattheyarevotingon.Eachparticipantgetstwovotes.It’sokaytovoteonyourown topic(s). It’s okay tousebothvoteson a single topic ordistributethemacrosstwodifferenttopics.

Tallythevotestoprioritizethetopics.Then,runthetopicsthroughakanbanboard right on top of the table. Three columns are essential: To Discuss,Discuss, and Done. Place the topic with the highest number of votes in theDiscusscolumn,andsorttheothersinpriorityorderintheToDiscusscolumn.If you wish, you may create a fourth column for decisions, epiphanies, oractions.

Seta timer for fiveminutesand invite theauthorof the topic in theDiscusscolumn to lead off the discussion. The facilitator should take care to ensureeveryonehasachancetospeak. (Bewarethe loudextrovertmonopolizingthe

conversation!) When the alarm goes off, allow the speaker to finish theircurrentsentence,thenvoteusingthumbsupordown.Amajorityofthumbs-upvotesmeansthediscussionmaycontinueforadditionalminutes.Amajorityofthumbs down votes signals the group to move on to the next topic. Thefacilitatorcanbreakanyties.

RepeatthisprocessuntiltheLeanCoffeesessionends.LeanCoffeeconcludeswitharoundofclosingcommentsfromeachparticipant.It’sokaytopass.

Figure49.LeanCoffeeSetup

While Lean Coffee usually occurswith just a small group of people, don’t beconstrainedbygroupsize. I’ve facilitatedLeanCoffeeswith fifteen to twentytablesoftenpeopleeach.

Stand-Ups

Earlier, I described stand-ups that saw more foot shuffling than discussion,where projectmanagers tried unsuccessfully to get status from attendees. Aceremony where you go around the room and everyone says what they’reworking on today, what they did yesterday, and what they plan on doingtomorrowisastatusmeeting,whichisunnecessarywhenworkisvisibleonaboard. It’s alsoboring.Whenwe go around the room,people spend the timefiguringoutwhattosaywhenit’stheirturninsteadofgivingotherstheirfullattention.

Whenthestand-upisdoneinfrontofavisualboard,itisobviouswhatpeopleare working on. When it comes to the stand-up, get to the point. Is thereanything blocked? Is there any invisiblework? Is there somethingwe shouldknowabout?Somethingthatwillimpactus?Somethingthatshouldbeontheboard,but it isn’t?Make it snappy, andget to theafter stand-up,where realproblemsgetsolved.

AtonecompanyIworkedat,agroupofthirty-fivepeoplemetdailyat9:30a.m.for stand-ups. Initially, the “go around the room” process was used. Somepeopleweresouncomfortablespeakinginthespotlightthattheirvoicesweremerewhispers.Otherpeopleenjoyedbeinginthelimelightandtendedtohogalot of precious and expensive time. Imagine the cost of thirty-five engineersandmanagersbeinginanineffectivemeeting,nottomentionthatlisteningtothirty-fivepeoplereporttheirstatusmadeforadreadfullyboringmeeting.So,themeetingruleschanged. Insteadofgoingaroundtheroom,wesetapolicythat theboardmustbeupdatedandaccurateprior to9:00a.m.This allowedpeople to simply look at the board to see the latest status, and the stand-upcouldbespentfocusingonriskanduncertainty.

Thefollowingthreequestionsbecamethenewagenda:

1. Whatworkisblocked?Noticetheemphasisisontheworkandnottheperson.ThiefUnknownDependenciesoftenchimedinherewithblockagesduetodatabasearchitectureissues.

2. Whatwork isatriskofbecomingblocked?Here’swhereThiefConflictingPriorities typicallyshowedup.

3. Isthereworkbeingdonethatisn’tontheboard?Thisquestionwouldthenevolvetoincludequeries about work thatmight currently be invisible to the team or problems thatmight haveoccurredduringproductionthepreviousnight.ThesequestionsoftenrevealedthedirtyworkofThiefUnplannedWork.

Thischangeallowedtheteamtoimmediatelyseeandrecognizemajorblockerspreventingimportantworkfrombeingdelivered.Thethreequestionsmadethestand-upsimpleandfast. Itwasoverby9:45a.m. (just fifteenminutes),andthisallowedanamazingthingtohappenthatwascompletelyunexpectedandspontaneous. Engineers hung around after the stand-up (because they hadfifteenminutes freebeforeheadingoff to theirnextmeeting) and started totacklesomeoftheengineeringproblemsthatblockedwork.Wecalledthistimethe“afterstand-up.”Previously,Iwouldhavehadtoscheduleameetingeightdaysoutinordertofindavailabletimeontheengineers’calendarsandtofindanavailablemeetingroom(meetingroomrealestatewasinhighdemand,andpeoplefrequentlyhadtogotothecoffeeshoparoundthecornertomeet).

Theresultwasthatmeetingsdecreasedbecausepeoplehungaroundafterwardstosolvetheproblemsthathadjustbeendiscussed.Halfpastthehourisagoodtimetoscheduleafifteenminutestand-upbecausewhenit’sover,peoplewillhavefifteenminutesbeforetheyhavetogettotheirnextmeeting.

Interruptions also decreased because instead of people popping in andinterruptingmewith“Gotfiveminutes?”theyknewthattheycouldcatchmeatafterthestand-uptoasktimelyquestionsorgetquickfeedback.

Thestand-uptogetherwiththeafterstand-upallowedustorevealwherethetimethieveswereatwork,whichinturnsavedmuchtimefrombeingstolen.

One last comment on the topic ofmeetings is the acknowledgment that theregular cadence of ameeting held at the same time, in the same location is

extremely helpful for all involved. This relatively simple guideline reduceduncertaintyforthirty-fiveexpensivepeople.

Andnow,inthenextsection,weturnourattentiontopracticesthatIconsiderproblematic.Thesepracticesrangefromisolatedoccurrencestocommonplacemisfortunes,buttheyallhinderus inourpursuitofkeepingourworkvisibleandholdingthethievesatbay.

KEYTAKEAWAYS

LeanCoffeeallowspeopletodiscussthetopicstheywantinafun,respectful,andefficientsetting.OrganizeandflowLeanCoffeetopicsthroughakanbanoncetheyhavebeenvotedon.Using a board to show the status of work allows stand-up time to be spent discussingproblemsanduncoveringinvisiblework.Holdingstand-upsataregularcadenceatthesamelocationreducesuncertainty.

TellmehowyoumeasuremeandI’lltellyouhowI’llbehave.—EliyahuGoldratt

3.5

BEASTLYPRACTICESIdecided to throwthis section inaswenear theendbecause thereare somethings thatmust be said in the spirit of full disclosure, but I didn’t want toscareyouawaywithmyopinionatedmessagesatthebeginning.Nonetheless,Iwanted to include them so you and your teammemberswill recognize themwhenyouseethem—akindofredflagcheatsheetofcompanypracticesthatmightbenegativelyimpactingtheworkplace.

So, listed in no particular order aremy rants and othermusings that inciteparticularly visceral reactions. They concern systemic pressures that occurthroughout theorganization.Theseare thepressures that lay the foundationforThiefTooMuchWIPandotherthievestosabotage improvementsand, insomecases,jeopardizetheemotionalandpsychologicalsafetyoftheworkforce(whichinturnmakespeopleupdatetheirrésumésandsearchforhealthierandkinderworkplaceclimates).

FlowTimeMetricsthatExcludeWeekends

There are three reasonswhy excludingweekends from speedmetrics is a bigproblem.

1. Allmetricsarebasedonassumptions.Andallyouneedtodotodiscreditametricistoquestion the assumptions. Allow me to demonstrate. Are you saying you never work

weekends? What about holidays? What about vacation days? Should we count thevacationdayspeoplequalifyfororonlytheonestaken?Whatifwejustlookatthetimepeopleworked?Doeseveryonealwaysworkaforty-hourweek?Howmuchtimeareyouwillingtospenddebatingassumptionsinordertomakeyourmetricscredible?

2. Flow time exclusions encourage gamingof thedata.What happenswhen resourceutilizationdecisionsaremadebasedoffoftimesheets?Howaccuratearethoseestimates?Whenpeopleworkweekendsbutdon’t count that time, themetrics canbedeceptivelyreassuring. Furthermore, I’vewatchedworkers attribute100%of their time to a singleprojectbecausetheyfearedthatitwouldotherwisereflectpoorlyonthem.Thedatawaswrong and everyone knew it. Using metrics to shame people encourages gaming. Thesamethingistruewithtarget-drivenmetrics.Whenthefocusisonthemetricinsteadofthegoal,it’saproblem.Ifpeopledon’trevealthingstous,thenwelosetransparency.

3. Business customers care about duration. If you’re my customer and I tell yousomethingwillbedoneinthirtydaysbutthendon’tdeliverinthirtydaysbecauseIwasonlycountingworkingdays,you’renotgoingtobehappy.Customersdon’tcarehowlongtheirthingsatindevelopmentortest.Theywanttoknowabouttheduration.

There will always be variability in our systems due to weekends, holidays,unplannedwork, and sick days. Credit leadership for being bright enough toknow that requests arriving the day before a three-day weekend will have alongerleadtime.Visibilityofaccuratemetricshelpsusmakegooddecisionsbutisdependentonthetransparencyofothers.Helpothersbeokaywiththetruth.If you want more predictability in your organization, measure flow timeaccurately. Once you exclude weekends from flow time, or any other timepeople aren’t thought to beworking, you open the door tomany questionedassumptions.

IneffectiveAccountingMethodswithTimeSheets

Correlating activity with business value is risky. High activity levels do notequate to high business value. High activity levels equate to hidden queues,whichcontribute to teams finishingprojects late.Usinga timesheet to trackthenumberofhoursToddspentworkingonataskdoesnotreflectthespeedof

delivered business value. Furthermore, the customer doesn’t care howmanyhoursToddspentworkingonlineitem236.

I’ve experienced these problems first hand. I once waited eight weeks for apurchase order from a customer who wantedme to train their team withinthree weeks, and another time I waited twelve weeks to be added to acompany’s payment system. The processes used in traditional accountingsystems do not seem to be able to move as quickly as other parts of thebusiness need them to. Some organizations are turning the tables onineffective accounting methods that are based on costs and margins byfocusinginsteadonthebusinessvaluecreatedandtheeconomicprofitabilityofthevaluestream.

With other viable options available, organizations that are bleeding frombudget process landmines can turn the tables and staunch the wounds bylooking into other methods, such as those offered up by Brian H. Maskell,Bruce Baggaley, and Larry Grasso in their book Practical Lean Accounting: AProvenSystemforMeasuringandManagingtheLeanEnterprise.

GanttCharts

Likeafalsepromise,Ganttcharts(jokinglycalled“can’tcharts”bysome)foolus into believing that timeline accuracy based off estimates is doable.DevelopedbyHenryGantt in the1910s,aGantt chart isa typeofhorizontalbarchartthatillustratesthestartandfinishdatesofallthetasksinaproject.The problem is that Gantt charts don’t consider the wait and blocked timesthatoccurduetohighcapacityutilizationofworkers.

Ganttchartssubdivideprojecttasks intotimeintervalsthatgetbrokendowninto smaller groups of subtasks. Due dates are identified, and people are

incentivized tomeet the timeline.One can recognize this in the promise, “IfprojectVgetsdonebyJulyfirst,youcantaketwodaysoffduringtheweekoftheFourthofJuly!”

In response, people insert contingencybuffers into theplan toprevent tasksfrommissing their deadline, and these aggregate into even longer timelineswith more variability. Each contingency buffer opens the door to includingmorework.“Oh,that’snotdueuntilThursday.Canweaddthisotherlittlefixin?”

Each bucket in the timeline is already prone to some kind of variance. Forexample,someonegoesofftoatwo-dayconferenceortakestheircarintotheshop, the internet connectivity goes down, or the database server is runningslowly.

Whenwe add in contingencybuffers,weunintentionally extend the timelineeven further because the workers with the highest demand get blocked andaren’ttherewhenyouneedthem,or ittakesthemlongertorespondbecauseprojectVisn’ttheonlythingthey’reworkingon(alikelyscenarioifthey’reinhighdemand).Asaresult,thereisworknotgettingdoneuntilsomeonewiththerightskillsethascapacitytodoit.Sowewait.

We wait while flow time goes up, other tasks dependent on that task getdelayed, and people start asking, “Is it done? Is it done? Is it done?” Moreprojectstatusesgetrequested,morevariabilitycreepsin,andcostsgoup.Thisincludespsychologicalcosts,whichgoupbecauseasqueuesandwaittimesgetlonger, they thwart motivation. When something is ready to use within anhour, there is a sense of urgency.When it’s a threeweekwait, there is littlevalue in hurrying to finish it. The work begins to decay like perishable fruitwhen it sits around for too long. Partially completed work can be veryexpensive.

Instead of managing work with Gantt charts, consider managing work withqueues.Weknowthelongerthequeue,thelongerthewaittime.Thefocusonqueues and wait times changes the game. Projects no longer need to beaccomplishedbyheroic,sleep-deprivedpeopledrivenbyGanttcharts.

Insteadof givingduedates, reduceWIP, prioritize byCoD, and reduce batchsize. Instead of organizing by projects, organize by product and decoupledependencies on architecture or single threaded skill sets that increase waittimesandlengthenqueues.

IndividuallyNamedSwimlanes

Figure 50 shows the results of a team subjected to begrudgingly using theirboss’skanbanboarddesign.

Figure50.IndividuallyNamedSwimlanes

The boss mandated this board design, carving out swimlanes for individualteammembers.Hewantedvisibilityonwhathisteamwasworkingon.It’snot

difficulttounderstandwhytheteamhatedit.Herearesomeoftheproblems.

There are four main problems with individually named swimlanes. Given aconstantly changing environment, there is a desire to feel in control of yourownwork,buttherearecoststothisdesire:

1. Becausetheboarddesignwasfocusedonindividuals,stand-upsfocusedonindividualsinsteadoftheworkitself.Thestand-upbecamean“I”fest.“Ididthis,”and“I’mdoingthat,”and“I’mgoingtodothisotherthing.”Keepthespotlightonthework,notthepeople.

2. Therewasaperceptionofpoorperformancewhensomepeopledidn’tmove ticketsacross theboardasfastasothers.Notallworkisthesame.Someworkismoreplaguedbytimethievesthanotherworkis.Unplannedworkcaninflatetasksandcausevariation.(Remembertheimpactoflandslidesontrainschedules?)

3. Peoplefelttheycouldn’ttouchworkoutsidetheirlane.Insteadofbeingencouragedtobroadentoward T-shaped skills (Figure 51), they concentrated on greater specialization, which madeThiefUnknownDependencieshappier.

4. Focusing on utilization prevented collaboration. People were incentivized to not help others.WhyshouldAlangohelpRussifitmeanstheworkinAlan’slanewilltakelongertocomplete?Peoplewillprioritizebasedonmakingthemselveslookgood.Thisbehaviorcandecreasebusinessvalue.ItcouldbethatthesinglemostbeneficialthingAlancoulddotoincreasebusinessvalueistogohelpRussfinishsomething.

Figure51.T-ShapedSkills

Ifthisisyoursituation,andyouwanttotakeanincrementalstepinadirection

thatmoves you closer to optimizing the flowofwork, then consider a boarddesign thatbringsvisibility to thenatureof thedemandand to theskill setsrequired to get the job done (Figure 52). Thisway,whenwork is stalled, it’seasiertoseewhichskillisinhighdemandandworktocrosstrainmorepeopleto reduce the risk of bottlenecks. Instead of trying to keep individuals busy,changehowyouvisualizetheworktoinspirehowtofocusontherightthing—smoothingtheflowofworkfromstarttofinish.

Figure52.Specialization

WorkScatteredEverywhere

Meetingscanbehardenoughontheirown.Justmakingsuretherightpeopleshowupat the right time todiscuss the right topics tohopefullyachieve thecorrect outcome is a difficult task. Add in three different boards, sixspreadsheets, four slack channels, a video conference tool with poor videoconferencing bandwidth, twenty-seven open browsers, and umpteen other

toolsandapps,andyou’veplungedintomeetingmalaise.Whowantstojuggleallthatchaos?Noonewhowantstokickthetimethievesinthebutt,that’sforsure.Keepthingssimplesopeoplewaste lesstimesearchingfor information.TimespentsearchingincreasesWIP.

GarishCardColors

Information/datacanbebeautifultogazeupon,butnotwhensurroundedbycolors that are at visualwarwith eachother orwith thebackground.Beautyattracts. Design your visual kanban user experience with beauty in mind.Author of three bestselling books on visualizing information and TED talkspeakerDavidMcCandless identifies four elements he believes are necessaryforavisualizationtowork:

Information:Thedatamusthaveintegrityandmustbeaccurate.Function:Thegoalmustbeusefulandefficient.Visualform:Themetaphormusthavebeautyandstructure.

Story:Theconceptmustbeinterestingandrelevant.1

Makeboardsvisuallyappealingtokeeppeople interestedandengagedandtoavoid confusion and wasted time. Communication can be hard when manydifferent tools are used. Improve communication with elegant visuals thatintegratewellwithothertools.

BestPractices

It was inHonoluluwhileworking at Boeing that I heard someone (my boss,actually)say,“Doitrightthefirsttime.”Iinstinctivelyknewthatwasn’tright.The only time anyone does anything right the first time iswhen they followdirectionsgivenbysomeoneelsewhohasdoneitmanytimesbefore.Thefirst

timedoinganythingisanexperiment.Andsoitiswithkanban.Thefirstboarddesign attempt is an experiment to help you discover how to improve yourworkflow. That’s why there really are no “best practices” when it comes todesigningyourkanbanboard,aswellasmanyothersituations.Unlessyouaredoing something simple that has been donemany times before—somethingwhere cause and effect are known, we cannot know exactly what specificallyshouldbedone.Wejustdon’tknowwhatwedon’tknowyet.

Today, when I hear the term “best practices,” I attempt to use my magicintrovert powers to avoid physically cringing. I have to remind myself thatthereareappropriatetimesforbestpractices.Whenthepilotgoesthroughthechecklist before heading down the runway.When the nurse cleans a woundbefore applying a bandage. When the system admin pulls the server out ofrotation before restarting IIS. Best practices might sound cringeworthy, buttheydohave theirplace, especiallywhenyou’redoing something routinebutimportant.

Now that my ranting is done, let’s head to the conclusion of our journeytogether,wherewe’ll answer a tough question, look at improving alignment,andreduceresistancetochange.

KEYTAKEAWAYS

Don’t exclude times when people aren’t “supposed” to be working from metrics or themetricswillbeskewed.Look for alternatives to ineffective accounting methods. Just because they are “the waythingshavealwaysbeendone,”doesn’tmean they represent theonly (orbest)way todothem.ConsiderreplacingGanttchartswithqueues.Bewareofindividuallynamedswimlanes.Simplifymeetingtoolswheneverpossible.Make kanban boards (and other presentation materials) visually appealing to engageviewers.Bestpracticeshavetheirplace,particularlywhen itcomestosimpleroutinetasks,but it’softenabetterideatoconductyourownexperimentstodiscoverwhattrulyworksforyoursituationandorganization.

CONCLUSION:CALIBRATION

Neverletformaleducationgetinthewayofyourlearning.—MarkTwain

MountainView,California,September2011

FollowingthefirsteverKanbanforDevOpsclassinMountainView,California,amansportingakiltandlonglocksasked,“Howdoyouintegratekanbanwithticketsystemswithoutslowingdownhigh-throughputOpsteams?”Theman,whosenamewasBen,wrotethequestionona large,orangestickynotewhilestandingatthebackoftheDevOpsmeetuproom.

Now,lookingattheorangenote,whichIsaved,IrememberthatIdidn’tknowhowtoanswerthequestionatthetime.It’sasvalidaquestiontodayasitwasin2011. My response today begins with one comment and two questions: Mycommentisthatanychange,evenagoodchange,impactsperformance.Addingnewpeopletotheteamrequiressomelevelofon-boardingandgettingpeopleup tospeed. In theshort term, the teamwillbe impacted.But is itworth it?Therehastobesomewininordertodisrupttheteam.Next,mytwoquestionsare:

1. Whydoyouneedkanbanifyourteamsalreadyhavehighthroughput?2. What’stheproblemthatyouwanttomakevisible?

Iwouldhavetomakealotofassumptionstospitoutananswerworthmuch.But for the sakeof it, let’s assume thatwhile the teamhashigh throughput,theyareoverloadedandrelyonheroicstomeettheirdemand.Theproblemyouwanttomakevisibleinthisscenariowouldbetheoverwhelmingdemandyourteamishavingtoworkthrough(ThiefTooMuchWIP)andthereasonwhyit’slike that.Making thisvisible allowsyou to see theproblemandhavea thinkaboutwhattodonext.TheanswercouldbetoreduceWIPandreprioritize.ItcouldalsobetobringinmorepeopletomanageWIP(andtokeepyourcurrent

peoplefromseeinggreenerpastures).

LimitingWIPcankeepthroughputhighwhilereducingthelevelofdemandaswell as the problems that create and then feed into that demand, such asconstantinterruptionsfromunplannedworkandburned-outteammembers.

Thisiswhatwe’vebeentalkingaboutthroughoutthisbook:limitingthetimethieves’abilitytomesswithyourlifebyexposingthem,andthencontinuouslyimproving on ineffective practices across teams, departments, andorganizationstoreapmaximumbenefits.Remember,exposingthetimethievesis important because it’s ridiculously hard to manage invisible work. WhenthereistoomuchWIP,thereisnotimetosimplythink.

Whenwelookattimetheftwecannarrowitdowntocalibrationbecausethat’swhatmakingworkvisibledoes.Whenyoucanseethethievesforwhattheyare,it allows you to realign and calibrate the systemic issues that hinder yourorganization.

Alignment(Coherence)ofTechnicalandBusinessTeams

Aligning technical and business teams is a matter of gaining clarity andconsensus aroundwhy teams are doing what they’re doing. Your teamsmay(and in fact, should) argue about thewho,what, andwhen, but the contextaroundwhyshouldbewellunderstood.Alignmentproblemsareoftenrelatedto conflicting priorities resulting from toomuch demand. If all requests gotdone,therewouldbenoproblem.PrioritiesconflictbecausethereistoomuchWIP.

Gaining crystal clear understanding on why a company is in business in the

firstplacecanfundamentallytransformthecultureofanorganizationbecauseleaderscangobacktothistomakeprioritydecisions.

Changeishardforhumansandisoftenmetwithresistance,especiallyduringdramatictransformations.LeancoachesrefertothisastheJ-curve(Figure53).Bigchangescausedipsinperformanceduetoavarietyofreasons:learningnewmaterial,hiringmorepeople, installingandusingnewtools,yada,yada,yada.That’swhysmall,gradualchangesareeasiertoimplement.Smallchangemeetswithlessdefiance.TakeeBayforexample.

Figure53.TheJCurve

Oneday, eBaydesignersdecided that abright yellowbackgroundwasn’t coolanymore,sotheyreplaceditwithawhiteone.Customersdidn’tlikeitonebit.SomanypeoplecomplainedthateBayrolledthechangebacktoyellow.Then,overaperiodofseveralmonths,theymodifiedthebackgroundcoloroneshadeofyellowat a time,until all theyellowwasgoneandhadbeen replacedwith

white.1Hardly a single usernoticed. This is the power of gradual change—itgetsmetwith less resistance as people are only asked to adapt to one smallpieceatatimeratherthantoeverythingallatonce.Satisfypeoplewithgradualchangeinsteadofdramaticchange.

Otherchallengeswithrollingoutanewwayofworking:

1. LimitingWIPisscaryandnotintuitive.LimitingWIPmeanspeoplecan’tsayyesallthetime,andthismakesthemuncomfortable.ButifWIPlimitsareignored,thinkthroughwhatwillhappentoyour flowtimeandchaos levels.SelectaWIP limit that isdoablebutchallengesyoutosaynosomeofthetime.

2. There’salwaysmoredemandthancapacity.Bewaryoffallingbackintotheoldhabitofstartingeverything because of the pressure to say yes to everything. Remember,WIP limits are yourfriendandthekeytogettingthemostimportantworkdone.

3. People who are afraid may not cooperate with the level of visualization you are seeking.Leadershipshouldstrivetorelievethefear,sopeoplearecomfortablewithtransparency.Don’tpunishthepeople;changethesystem.

Howyoucanveeroffcourse:

1. Focusing too much internally and not enough on interactions with stakeholders/customers.Strikeabalance.Rememberthecostofdelaytothebusinesswhenprioritizingyourwork.

2. Neverfinishingbecauseyouwantittobeperfect.Youaregoingtokeepimproving.Acompletedblog post is better than no blog post. Deliver something so you can get feedback as soon aspossible.

3. Tryingtofixeverythingatonce.Visualizeyourcurrentrealityandthenmakesmalladjustmentssoyoucanmeasuretheimpactofeachchangeandlearnfromit.

4. Expectingovernightresults.Itcantakemonthsforakanbandesigntostabilize.Eventhen,youwillbecontinuallyimproving.Thereisalwaysroomforimprovement.

AsaLeancoach,IoftengetaskedwhatthedifferenceisbetweenkanbanandScrum (an Agile framework used to complete projects). They are both Agilemethodsthatuseconstraintstoenableproductiveoutcomes.

Scrumandkanbancanworkwell together.Theyaremorealikethantheyaredifferent.Differencesincludereleasecadence,roles,andthetypeofconstraintsthemselves.Scrumusesatime-box(usuallytwoweeks)tolimitdemand,whilekanbanuses theWIP limit to constraindemand.Someteamsuseahybridofboth and call it ScrumBan (originally designed as a way to transition fromScrumtokanban).

IaminspiredbyKlausLeopoldandSiegfriedKalteneckerwhowroteakanbanelevator pitch in their book Kanban Change Leadership: Creating a Culture of

ContinuousImprovement.

Kanbanisamethodforthecontinuousimprovementofyourownareaofwork.Youdon’tbeginwithabigchangemanagementproject,butratherfocusonaseriesofsmallchangesteps.Youidentifythemostimportantbusinesspartnersandtogetherinvestigatethestrengthsandweaknessesinyourcurrentworkprocesses.Basedonvisualizationoftheseprocesses,youuseasimplemeanstomakethingsmoreefficient,improveleadtimes

andcreateaddedvalueforyourcustomers.2

To investigate strengths and weaknesses with business partners, usingvisualization reduces resistance to change in the organization.Making workvisiblehelpswithinvestigation,whichinturnhelpspeoplebecomecomfortablewithchange.

Old school language tends to block progress. Combatting time thieves takesagility, daring, and new words to describe new ways of doing things. Thinkpeople (not resources),goodpractices (notbestpractices), anduncertainty (notexactness).

Changing how we work requires a mental shift in the way we think aboutthings.

WhenSomersetCounty,England, changed their traffic system from lights tono lights, the townsfolk thought it wouldn’t work. “Drivers are not going todovetail in,” they said. Surprisingly (tomost citizens), after the traffic lightswere removed, the queues instantly disappeared. It was less congested andeasier for pedestrians to cross the street. A drive through town that used totake twenty minutes now took just five minutes. The difference wasextraordinary.

Ittooksometimeforpeopletoadjust.Mostpeoplewereusedtojustlookingat

the lights. Some drivers still assumed the right of way in the time-honoredfashiondictatedby traffic controls.Reformrequireda change in culture, andpeople had to unlearn their bad habits. It was a new way of thinking thatrequiredamindshift,andittookawhileforpeopletocatchon,butitresultedinasystemthatwasultimatelysaferandfaster.

Thesame is truewhenmovingtoaLeankanbanflowapproach. It’snewanddifferent.Peoplethinkitwon’twork,andthereisresistanceacrossteamsanddepartments. Culture change is often required to unlearn bad habits. Theresultsarerarelyasinstantaneousasthetrafficlightexample,butoftenpeoplesee some improvement early on in the form of fewer interruptions, moretransparency,andafasterflowofwork.

Theoryandexpertiseinthescienceofquantitativemeasuresisavaluablepartofexposingtimethieveryandimprovingworkflow.Itinterestsme,andsoI’vedabbledinit.It’sthemath—Ilikesolvingforx.Myfirstambitionwastobeadetective likeHoneyWest. I’ve experienced firsthand the power to influencedecisions using metrics. I’ve gotten budget and headcount approvals andagreementtopursueonedirectionoveranotherbecauseofmetrics.I’veappliedthemintheinterestofimprovingtheworldofwork.

But theory and metrics is not why I do this work. Articulating abstract orunintuitive ideasvivavoce isnotmystrongsuit.Mybraincompeteswithmyownnarrative.ThereasonIknowalotaboutmakingworkvisibleisbecauseinmanyways (maybe inmostways), it’s easier forme to communicate visuallythanvocally.Theactofmakingworkandideasvisiblesothatpeoplecaneasilysee the problem or situation is exciting. Creating useful, relevant, visuallyavailable and beautiful information to help people understand what’s reallygoingonisatruedelight.Butit’snotwhyIdothiswork.

IdothisworkbecauseIgettoconnectwithpeopleatan intrinsic level. Ican

sensewhat people are dealingwith fromwatching them atwork and duringworkshops.Muchof itcomesfrominstinctwhileobservingpeople. Ideas justcome to me for how to visually shine a light on their pain. I see what’shappeningattheemotionallevelandtransformitintoaphysicalvisualtohelppeoplecommunicate.Forme,interpretingotherpeople’senergytounderstandtheirsituationisanautomaticprocess.I’mavisualempathiclistener,andthisiswhatIdobest.Putmeinthecrimesceneandletmededuceaway.

Iwouldnotbeable todo thisworkwithout theability tomakeworkvisible,optimize for flow, and enable important discussions. I’ve provided you withsomeoftheessentialtoolsandknowledgenecessarytodothethingsthatwillhelpyoubecomethevoiceofreasoninyourorganization.It’snowuptoyoutotakethesetoolsandrunwiththem.

Stop banging your head against a system that doesn’twork. The practices inthisbookcanputyouinapositiontocontinuouslyexperienceimprovements.Small improvements over time make a difference. So get out there andencourageotherstojoinyouinyourjourneytodotherightthing.IhopeIhaveinspired you to get started. Just start doing the exercises, visualizing theproblems, and provoking the necessary conversations, and see where theseactionsleadyou.

Idoubt that I’ve coveredeverythingyouneed toknow to successfully exposeyourtimethievesandoptimizeyourworkflow,butI’vetriedmybest,whichisall any of us can do. There are many aspects of visualizing and improvingworkflowthatIamstilllearning.

When I first thought about writing this book, my brain seemed to beoverflowing with ideas and examples for how everyday workers trapped incorporate bureaucracy or business stalemate could improve their world andcould become the voice of reason in their organization. And that’s who Iwantedtoreach/teach—theeverydayperson just tryingtodoagood joband

enabletheirteamtoreachabetterplace.Ahappierplace.So,carryon.IfIcandoit,socanyou.

Goodluck!

GLOSSARY

A3:Namedaftertheinternationalpapersizeof11x17inch(297x420mm)pieceofpaper,anA3isusedtostructurediscussionstogainunderstandingandagreement.

AGILE:Incrementalanditerativeimprovementsdoneonaregularcadence;analternativetotraditionalprojectmanagementmethods,suggestsfrequentreassessmentandadaptationofplans.

BOOLEANLOGIC:Aformofalgebrainwhichallvaluesareeithertrueorfalseandrepresentedinthebinarynumberingsystemwhereeachdigithasavalueofeither1or0.

BUILD:Whencodeiscollectedfromasourcecoderepository,compiledintoexecutables,andpackagedupwithallthenecessarypartstobeinstalledintoaplacewheretheresultingnewfunctionalityisseenbyothers.

CAPACITYUTILIZATION:Thepercentageofthetotalpossiblecapacitythatisused.Ifapersonhasthecapacitytowork10hoursadayandtheywork7hours,thentheircapacityutilizationis70%.

CHURN:Customersorsubscriberswhocuttieswithyourserviceorcompany.

CONSTRAINT:Abottleneckinthesystem;somethingpreventingforwardmovement.

CONTEXTSWITCH:Theactofstoppingworkononethingtostartworkonadifferentthing,duetoaninterruption.

COSTOFDELAY(COD):Awayofcommunicatingvalueandurgency,CoDisameasureoftheimpactoftimeonvaluableoutcomes.

COUNTERMEASURES:Actionstakentocounteractaproblem.

CYCLETIME:Theelapsedtimeittakestocompletearequestfromthetimetheworkbegantothetimetheworkwasdelivered.

DEPENDENCY:Filesrequiredinorderforsourcecodetocompile;peoplewithspecializedskill-setsneededtodosomething;aneventthatneedstooccurbeforesomethingelsecanbeachieved.

DEPLOYMENTLEADTIME:Theelapsedtimeittakestobuildanddeployachangeoncecodeischeckedintosourcecontrol.

ECONOMIESOFSCALE:Aconceptineconomicsthatrelatesadecreaseincostwithanincreaseinproduction;thecostadvantagethatariseswithincreasedoutputofaproduct.

ENTERPRISERESOURCEPLANNING(ERP)SYSTEM:Amanagementinformationsystemthatintegrateselementssuchasplanning,purchasing,inventory,sales,marketing,finance,andHR.

ENVIRONMENTISSUES:Problemswiththeconfigurationofserversthatpreventwebsitesandotherthingsfromworkingcorrectly.

FAILUREDEMAND:Demandcausedbyafailuretodosomethingortodosomethingrightforthe

customer.

FEATUREDRIVENDEVELOPMENT:AtypeofAgiledevelopmentfocusedoncross-functional,collaborative,andtime-boxedactivitiestobuildfeatures.

FIRST-IN,FIRST-OUT(FIFO):Aprioritizationmethodwhereworkisprocessedinafirst-in,first-outfashion.

FLOW:Valuepulledthroughasystemsmoothlyandpredictably;thepositiveaspectsandjoyofbeing“inthezone.”

FLOWEFFICIENCY:Thepercentageoftimewhereworkisdonevs.waitingforworktobedone.Calculateitbydividingworktimebywork+waittime.

GANTTCHART:Theillustrationofthestartandfinishdatesofallstepsinaproject.

KANBAN:Japanesewordforvisualsignal;usedthroughoutthisbooktorefertoavisualmanagementpullsystemforknowledgework.

KINGMAN’SFORMULA:UsedtocalculateautilizationpercentagefromtherelationshipbetweenWIPandflowtimes.Itshowshowwaittimesincreasedramaticallyasutilizationapproaches100%.

LEADTIME:Ameasureoftheelapsedtimeittakestoworkonarequestfromwhenitwasfirstrequestedtowhenitwasdeliveredtothecustomer.

LEAN:ASocraticphilosophyusedtomakeimprovements.Leanputsjust-in-timepracticesandvisualmanagementfrontandcenter.

PULLSYSTEM:Whennewworkispulledintothesystembasedonavailablecapacitytohandleit;wherethepeopledoingtheworkhaveautonomytostartitwhentheyhavetime.

QUEUE:Apileupofworkwaitingonattentiontobeworkedon;workthatisinawaitstate.

RESOURCEEFFICIENCY:Thepercentageoftimethatresourcesarebusy.Sometimesreferredtoaskeepingpeoplebusyallthetime.

SCRUM:AnAgileframeworkusedtocompleteprojects.

SILVERBULLET:Anurgentrequesttodosomethingrightaway;usually,initiatedbysomeoneinaleadershipposition.

SMOKETEST:Atesttoensurethatcodefunctionsasexpectedafterabuildiscompleted.

SOURCECONTROL:Aplacewheredeveloperscheckintheircodeforsafekeeping.

STAND-UP:Abrief(usually15minutes)meetingwhereateamdiscussesissues.Becauseit’sonly15minutes,peoplestandupforit.

SUNKCOSTFALLACY:Whenyoukeepdoingsomethingbecauseyou’veputalotofeffortintoitandyoudon’twanttowastethateffort.

SYSTEM:Anetworkofinterdependentcomponentsthatworktogethertotrytoaccomplishagoal;includesthepeopledoingtheworkandtheimpactingrulesandtools.

SYSTEMSTHINKING:Aholisticviewofthesystemwherethegoalistooptimizethewholesystemversusjustindividualfunctionsorsilos.

TECHNICALDEBT:Extraeffortrequiredtofixsoftwarebugsanddevelopnewfeaturesbecauseofpreviousquickanddirtydesignchoices.

THEORYOFCONSTRAINTS(TOC):Awaytoidentifythemostimportantlimitingfactor(theconstraint)thatstandsinthewayofachievingagoalandthensystematicallyimprovingthatconstraintuntilitisnolongerthelimitingfactor.

THROUGHPUT:Thenumberofthingscompletedoveraperiodoftime.

TIME-BOX:Aspecificperiodoftimewithadistinctstartandendtime.Ex:Pencilsdownafterthetwohourexamthatbeganpreciselyatnoon.

VALUESTREAM:Theactivitiesdonefrombeginningtoendforaspecificproductorserviceinordertoprovidebusinessvalue.

VELOCITY:Thenumberofstorypointscompletedoveraperiodoftime(usuallytwoweeks).

WATERFALLAPPROACH:Atraditionalsoftwaredevelopmentmethodwhereworkcannotproceeduntilallthepartsoftheentirepreviousstagecomplete.

WEIGHTEDSHORTESTJOBFIRST(SWJF):Aprioritizationmethodwherethejobthathastheshortestdurationisprocessedbeforeotherjobsofequalvalue.

WORKFLOW:Theflowofworkthroughthepipeline(orsystem)frombeginningtoend.

WORK-IN-PROGRESS(WIP):Alltheworkstartedbutnotyetfinished.

WORKITEM:Anythingbeingworkedon;workthatencompasseseffortbothlargeandsmall.

WORKSTATE:Thestatethattheworkisin.Workflowsthroughdifferentstatesonitswaytocompletion.Theworkstatesshowuswheretheworkisinthepipeline.

NOTES

Introduction

1. In Time, directed by Andrew Niccol (Los Angeles: 20th Century Fox,

2011),DVD.

2. Darren Davis, “The Secret History of Kanban by Darren Davis [Guest

Post],” Northwest Cadence, February 19, 2015,

http://blog.nwcadence.com/the-secret-history-of-kanban-by-darren-

davis/.

3. KateMurphy, “No Time to Think,”TheNew York Times, July 25, 2014,

http://www.nytimes.com/2014/07/27/sunday-review/no-time-to-

think.html.

4. Niklas Modig and Pär Åhlström, This is Lean: Resolving the Efficiency

Paradox,(Stockholm:RheologicaPublishing,2016),Introduction.

5. W.EdwardsDeming,asquotedinJohnHunter,“ABadSystemWillBeat

a Good Person Every Time,” The W. Edwards Deming Institute Blog,

February 26, 2015, https://blog.deming.org/2015/02/a-bad-system-will-

beat-a-good-person-every-time/.

PartI

1.1

1. Vanessa Bohns, “Why Is It So Hard to Say No?,” interview by Jeremy

Hobson, Here and Now, March 31, 2014,

www.wbur.org/hereandnow/2014/03/31/saying-no-psychology.

2. Todd Watts, “Addressing the Detrimental Effects of Context Switching

withDevOps,”DevOpsBlog, SoftwareEngineering InstituteatCarnegie

Mellon University, March 5, 2015,

https://insights.sei.cmu.edu/devops/2015/03/addressing-the-

detrimental-effects-of-context-switching-with-devops.html.

3. “Context Switching,” OSDev.org, last modified December 29, 2015,

http://wiki.osdev.org/Context_Switching.

4. HarryF.Harlow,asquotedinDanielH.Pink,Drive:TheSurprisingTruth

aboutWhatMotivatesUs,(NewYork:RiverheadBooks,2011),3.

5. “The Hounds of Baskerville,” Sherlock, directed by Paul McGuigan,

writtenbyMarkGatiss,airedonJanuary8,2012,onBBC.

6. DavidRock,YourBrainatWork:Strategies forOvercomingDistraction,

RegainingFocus,andWorkingSmarterAllDayLong,(NewYork:Harper

Business,2009),47.

7. EdwardR.Tufte,EnvisioningInformation,(Cheshire,CT:GraphicsPress,

2013),50.

8. DanWeatbrook,personalconversationwithauthor,2015.

1.2

1. Troy Magennis, “Entangled: Solving the Hairy Problem of Team

Dependencies,”AgileAllianceconferencevideo,1:15:15,August5,2015,

https://www.agilealliance.org/resources/videos/entangled-solving-the-

hairy-problem-of-team-dependencies/.

2. Maura Thomas, “Your Team’s Time Management Problem Might Be a

Focus Problem,” Harvard Business Review, February 28, 2017,

https://hbr.org/2017/02/your-teams-time-management-problem-might-

be-a-focus-problem.

1.3

1. 2016 State of DevOps Report, (Portland, OR: Puppet Labs, 2016) 26,

https://puppet.com/resources/whitepaper/2016-state-of-devops-report.

1.4

1. RossGarber,asquotedinGaryKeller,TheONEThing:TheSurprisingly

SimpleTruthBehindExtraordinaryResults(London:JohnMurray,2013),

19.

1.5

1. MichaelFeathers,WorkingEffectivelywithLegacyCode, (UpperSaddle

River,NJ:PrenticeHall,2004),xvi.

2. Donald G. Reinertsen, The Principles of Product Development Flow:

Second Generation Lean Product Development (Redondo Beach:

Celeritas,2009),152.

3. Reinertsen,ThePrinciplesofProductDevelopmentFlow,47.

PartII

1. Colin Ware, Information Visualization: Perception for Design, (San

Francisco:MorganKaufman,2000),2.

2. Ware,InformationVisualization,2.

3. Linda Kreger Silverman, Upside-Down Brilliance: The Visual-Spatial

Learner. (Denver: DeLeon Gifted Development Center, 1999),

www.gifteddevelopment.com.

2.1

1. Philippe Kruchten,What Colour is Your Backlog, presentation, July 7,

2011, https://pkruchten.files.wordpress.com/2012/07/kruchten-110707-

what-colours-is-your-backlog-2up.pdf.

2. Silverman,Upside-DownBrilliance.

2.3

1. CorneliaDavis,personalconversationwithauthor,April2017.

2.4

1. Wikipedia, “Pomodoro Technique,” last modified April 10, 2017,

https://en.wikipedia.org/wiki/Pomodoro_Technique.

2. LangdonMorris,HighPerformanceOrganizations in aWickedProblem

World (Walnut Creek, CA: Innovation Labs, 2004),

http://www.innovationlabs.com/high_performance.pdf.

2.5

1. DanielKahneman,ThinkingFastandSlow(NewYork.Farrar,Strausand

Giroux,2015),4.

2. “Cost of Delay,” BlackSwanFarming.com, accessed April 22, 2017,

http://blackswanfarming.com/cost-of-delay/.

2.7

1. Julia Wester, “Visualizing More than Just Work with Kanban Boards,”

EverydayKanban.com, March 9, 2016,

http://www.everydaykanban.com/2016/03/09/visualizing-more-than-

just-work-with-kanban-boards/#housemove.

2. Jim Benson and Tonianne DeMaria Barry, Personal Kanban: Mapping

Work,NavigatingLife(Seattle,WA:ModusCooperandiPress,2011),158–

159.

PartIII

3.1

1. Wikipedia, “Hofstadter’s law,” last modified February 12, 2017,

https://en.wikipedia.org/wiki/Hofstadter%27s_law.

2. Daniel S. Vacanti, Actionable Agile Metrics for Predictability: An

Introduction(Victoria,BC:Leanpub,2015),51–53.

3. Reinertsen,ThePrinciplesofProductDevelopmentFlow,59.

4. Kaomi Goetz, “How 3M Gave Everyone Days Off and Created an

InnovationDynamo,”Co.Design,February1,2011.

5. Goetz,“How3MGaveEveryoneDaysOff.”

3.4

1. “Lean Coffee Lives Here,” Lean Coffee, accessed May 29, 2017,

http://leancoffee.org/

2. AdamYuret,HowtoHaveGreatMeetings:ALeanCoffeeBook (Seattle,

WA:ContextDrivenAgilityPress,2016).

3.5

1. David McCandless, “What Makes a Good Data Visualization,”

InformationIsBeautiful.net, accessed July 2017,

http://www.informationisbeautiful.net/2015/what-makes-a-good-data-

visualization.

CONCLUSION

1. JaredM.Spool,“TheQuietDeathoftheMajorReLaunch,”UIE,August7,

2006,https://articles.uie.com/death_of_relaunch/.

2. Klaus Leopold and Siegfried Kaltenecker, Kanban Change Leadership:

Creating a Culture of Continuous Improvement, (Hoboken, NJ: Wiley,

2015),277.

ACKNOWLEDGMENTS

I could not have written Making Work Visible without the support of ToddSattersten.ToddbelievedintheallureofthisbookwhenIwasn’tsosure.Helaidthefoundationandputthewheelsinmotiontomakeitallpossible.Toddhaskeeninstinctsandaknackforattendingtothemostcrucialaspectsofbookpublishing.

IthasbeenmyextremepleasuretoworkwiththeremarkableAnnaNoak.Annahas got to be the most supportive and dedicated editorial director in thebusiness. She coachedme through the entire writing and publishing processduringweekends,holidays, andvacationswith the spiritofanastutemaster.As a novice illustrator, I was extremely lucky to have designer Joy Stauber’sguidance on color and image design. I also appreciate how generous SylviaCottrell,SarahHeilman,andLeahBrownwerewiththeireditorialtalents.

WritingMakingWorkVisiblehasbeenawayformetoconsolidateandreflectonthethingsI’velearnedwhileteachingandcoachingteamsonLean,kanban,andflow—any colleagues and friends have influencedme over the course of mycareerwhile learningmy tradewhomI’d like to thank, especially the teamatCorbis for all the collaboration during the early experimental years: LarryCohen,DarrenDavis, SandyThompson, EliHurst, Jay Freer, CalvinNguyen,Dwayne Johnson, Debbie Earle, Suzanne Bagdon, Jason Birklid, and RickGarber.

I had numerous teachers who helped me to learn Agile and Lean practiceswhom I want to thank, including David J. Anderson, Daniel Vacanti, ChrisHefley, Ian Carroll, Mattias Jansson, Arne Roock, Liz Keogh, TorbjörnGyllebring,MattiasSkarin,YuvalYuret,KarlScotland,DavidJoyce,KatherineKirk,KlausLeopold,MarkusAndrezak,PawelBrodzinski,HakanForss,JoakimSundén, Eric Willeke, Don Reinertsen, Mike Burrows, Jim Benson, Russell

Healy, SteveHolt,Michael Cheveldave, Jeff Anderson, Jason Yip, Jon Terry,GaetanoMazzanti,andJoshuaArnold.

TroyMagennis,whoseworkondependenciesIprofileinsections1.2and2.3,hasbecomeanextraordinarymentorandfriend.ImetTroyatCorbiswhereweshared an office together and where I experienced first hand his influentialwritingandcoaching.ItwasinthatofficewhereIfirstdreamedthatmaybeoneday,Itoomightwriteabook.Iamproudtobeoneofhismanydevoteesandfriends.

At DevOpsDays Seattle 2016, Pauly Comtois unveiled one of his first hand-drawnpresentationswhichinspiredmetodothesame.Laterthatyear,DavidO’Nealsharedhistalenteddrawingtechniquewithme.Manythankstothemboth—mypresentationshaveneverbeenthesamesince.

I am especially grateful for the kindness, encouragement, and inclusivenessduringmyfirstinteractionswiththeDevOpscommunity,whoinspiredmein2011/2012, and still do to this day, topracticemy learnings: PatrickDebois,John Vincent, Damon Edwards, Andrew Clay Shafer, Stephen Nelson-Smith,JohnWillis, Jez Humble, Ben Rockwood,MandiWalls, Jennifer Davis, TomSulston,MichaelRembetsy,MariusDucea,AdrianCockcroft,KarthikGaekwad,JamesWickett,ErnestMueller,andSaschaBates.

Ithasbeenmyabsolutepleasure toworkwith the talentedChrisHefley andJuliaWester.Theirtremendoussupportandfeedback,duringtheearlydraftofthebook,andthisbookbenefitedtremendouslyfromtheirreviewandinput.

ThearticulatefeedbackandenthusiasmIreceivedfromtheoneandonlyGeneKimkeptmeincheckandhelpedmeperseveretotheend.

I owe a special debt to Torianne DeMaria who penned the foreword for me

whileonvacationsailing.Herwritingbringsmepurejoywiththemagnificentwordsandexpertisesheshedsonusalltolearn.

And to my beloved husband, Joseph, who is the most generous person onearth. During the writing of this book, he took over household and gardendutiesonmanyaweekendandholidaysoIcouldwrite.Hemademetea,mademefood,andmademesmile.

ANOTEONTHETYPE

ThetextofMakingWorkVisiblewassetinChaparralProandDIN2014.

ChaparralProwascreatedby the legendaryandawardwinning typedesignerCarol Twombly. Released in 1997, Chaparral was Twombly’s last typefacebeforeretiring.Itcombinesthelegibilityofslabserifdesignspopularizedinthe19thcenturywiththegraceof16th-centuryromanbooklettering.ApartoftheAdobeOriginals familyof typefaces,Chaparralhas varying letterproportionswiththickandthinstrokes,creatingafriendlyappearanceandwasusedinthebodytexttogivealovelyvisualfinishtotheillustrationsthroughoutthebook.

DIN 2014 is a contemporary version of the well-known DIN typeface. DINstands for Deutsches Institut für Normung (German Institute forStandardization.) TheDIN1451 typefacewas designed in 1936 for road andrailwaysignage.ThetypefacewasdesignedbyVasilyBiryukovandreleasedbyParatype in2015.TheRegular ismostoftenused in long text settings,whileLightandBoldfacesareextremelywellproportionedatlargesizes,whichcanbeseen in thestrikingpart, chapter,andsectionheaders in thepagesof thisbook.

…

PrintedandboundbyWorzalla,StevensPoint,Wisconsin