Transcript
Page 1: Supply chain forecasting – Collaborative forecasting supports supply chain management

BPMJ6,5

392

Business Process ManagementJournal, Vol. 6 No. 5, 2000,pp. 392-407. # MCB UniversityPress, 1463-7154

Supply chain forecastingCollaborative forecasting supports

supply chain managementMarilyn M. Helms

Dalton State College, Dalton, Georgia, USA

Lawrence P. EttkinUniversity of Tennessee at Chattanooga, Chattanooga, Tennessee, USA, and

Sharon ChapmanBrach and Brock Confections, Chattanooga, Tennessee, USA

Keywords Supply chain, Forecasting, Supply-chain management, Demand

Abstract Supply chain management is built on the principles of partnerships and thedevelopment and use of the connections that exist between the links of the chain to provideinformation that will increase the efficiency of all members in the chain. Success stories abounddescribing lower costs, shorter lead times and increased customer service. Collaborativeforecasting applies supply chain management concepts to the forecasting function and usesavailable information and technology to force a shift from independent, forecasted demand todependent, known demand. Eventually, the future of forecasting may evolve to the point whereforecasting is not even necessary. Demand information will be supplied completely by supply chainpartners and the need to predict demand will be eliminated.

IntroductionSupply chain management has become the mantra of many companies seekinga way to meet the competitive challenges of today's business environment.Supply chain management is a broader perspective of the businessenvironment, as compared with more traditional approaches. Instead ofmanaging a business as a group of virtually separate functions, supply chainmanagement views these functions as closely connected links of a chain. Thechain extends beyond the boundaries of the organization to include suppliersand customers. Supply chain management involves the entire flow of a productfrom the purchase of raw materials from the supplier, all the way to thepurchase made by the final consumer.

The concept of supply chain management is based on several key tenets.The key principle is that all strategy, decisions and measurements are madeconsidering their effect on the entire supply chain, not just separate functionsor organizations. This broader approach is based on partnerships and thesharing of information between the links in the chain.

The goal of supply chain management is to meet the needs of the finalconsumer by supplying the right product at the right place, time and price. Thesupply chain management approach allows companies to meet this goal whilealso achieving competitive advantages. There are countless success stories inwhich companies have used supply chain management to `̀ replace inventory

The current issue and full text archive of this journal is available athttp://www.emerald-library.com

Page 2: Supply chain forecasting – Collaborative forecasting supports supply chain management

Supply chainforecasting

393

with information'', squeeze out costs, and improve efficiency and customerservice. The results of these improvements have also translated into healthierbottom lines. Success stories, coupled with the push of an ever more demandingcompetitive environment, and recent technological advances, are the drivingforce behind the rapidly growing number of companies implementing thesupply chain management concept.

As companies implement supply chain management, they must look at waysto bring the concept to each of the functional areas in their organization. Thisrequires making the cultural and process changes that support the concept ofsupply chain management which will ultimately lead them to the savings,efficiency and customer service levels that they seek. The forecasting functionis one area that must receive priority in these functional reviews and processchanges. The demand from the final customer is the force that drives theactivities in the supply chain. Each of the links in the supply chain operates inreaction to actual or anticipated demand from the consumer at the end of thechain. The level of accuracy and efficiency with which this demand iscommunicated up and down the chain is directly connected to inventory andcustomer service levels. Forecasting and demand planning are therefore a keyfactor in the successful implementation of a supply chain managementstrategy.

The complexity and uncertainty that exist in the supply chain make theconcept of accurate and effective forecasting an elusive target. Manycompanies are, however, making significant improvements by using anapproach that supports and facilitates the concept of supply chainmanagement. Collaborative forecasting is a way in which the entire supplychain is a participant in decisions about the demand that will drive theiractivity. Collaborative forecasting reaches internally and externally to gatherinformation that allows for the best and most timely predictions of demand.Recent technological advances are used to collect and bring the informationtogether as well as to transmit forecasts back to chain members. The number ofsuccess stories are increasing and include companies like Eastman Chemical,Reynolds Aluminium, Wal Mart, Ocean Spray and Heineken. They have beenable to dramatically reduce inventories and lead times while increasingcustomer service and forecasting accuracy. Experts now envision a day whenforecasting, as we now know it, will not be necessary. As the process improves,experts predict the need for forecasts will lessen and be replaced by themanagement of actual demand information.

Problems with the traditional forecasting processForecasting is often the most maligned department in any company regardlessof whether it is the responsibility of finance, marketing, sales or logistics. Mostcompanies know that their forecasts are inexact, but don't know what to doabout it and therefore ignore the issue, hoping the problem with solve itself.The idea that forecasts are always inaccurate and that there is nothing than canbe done about it forces companies to find ways to compensate for the

Page 3: Supply chain forecasting – Collaborative forecasting supports supply chain management

BPMJ6,5

394

uncertainty. The most often used method of dealing with uncertainty is bybuilding inventory. Departments buffer against their lack of confidence in theforecast by building safety stocks. As each link in the chain creates its ownbuffers, inventories skyrocket. Jack Stack, president of SpringfieldRemanufacturing Corp. in Springfield, Missouri, describes the situation likethis: `̀ Hate to say it, but there's a good chance that right now you're creating amajority of the problems you'll be dealing with next year. How? By puttingtogether an annual sales forecast based on gut feelings and wishful thinking''(Stack, 1997, p. 7). The results of a recent study of 1,380 manufacturing andretail companies, done by KPMG Peat Marwick, said `̀ many companiesworldwide don't know the exact amount of goods they need to manufacture,transport and store, yet they are very sluggish to adopt the very techniques andtechnologies that would solve the problem'' (Chain Store Age, 1998, p. 97).

Overcoming the challenges involved in creating a credible forecast is anincredibly daunting task for most forecasters. The marketplace is constantlychanging. There is a constant flow of new products, promotions and changingchannels of distribution. Unfortunately, forecasters often attempt to predictsuch a volatile future using only historical data. The problem with using onlyhistorical data to predict the future is that it requires the assumption that thepatterns that have occurred in the past will occur again in the future. In today'sever changing market this may or may not be a valid assumption. Forecasterstoday need more information. Ed Wodarski, director of Global Consulting withLPA Software in Rochester, New York says, `̀ using a forecast based on historymakes as much sense as driving a car by looking in the rear-view mirror. Theyneed to be looking ahead, not back'' (Fulcher, 1998, p. 89).

Companies also often have to deal with the problem of the existence of manyforecasts in their organization. This problem seems to be related to the lack ofconfidence in the sales forecast and the differing needs of each forecast's users.Marketing may use forecasts with an emphasis on trends occurring in themarketplace, while finance needs a forecast with an emphasis on budgeting.Sales may adapt a forecast based on sales quotas and production may createwhat they consider to be a better forecast based on their experience andproduction capacity and efficiency. Purchasing may also adjust a forecast toreflect their viewpoints and experience and create what they consider to be abetter forecast. While in theory all these forecasts would roll up the samenumber, this is rarely the case.

The problem that is created with the existence of so many forecasts is that,as a result, no one operates from exactly the same plan. Drs Mentzer and Moonof the University of Tennessee describe the problem as `̀ islands of analysis''.`̀ Without a clear communication of how the process and systems interact,islands of analysis will develop within the forecasting process. Islands ofanalysis are systems phenomena where one individual or group develops asales forecast based upon their own information and needs, and does not sharethat information or forecast with others in the company. The resultant salesforecasts may be significantly different than forecasts developed elsewhere in

Page 4: Supply chain forecasting – Collaborative forecasting supports supply chain management

Supply chainforecasting

395

the company (other islands) and these differences lead to conflicting plans''(Mentzer et al., 1997, p. 12). In describing the results of detailed study of thestate of forecasting in the USA, Drs Mentzer and Kahn describe the fact thatusing multiple forecasts has turned into a backward planning process. `̀ Inmany responding companies (34 percent) the planning process is backward,i.e. the business plan is used to develop the forecast instead of the other wayaround. Apparently, management in these companies is more concerned withthe business plan than the sales forecast, even though the latter should drivethe former'' (Mentzer and Kahn, 1997, p. 12). Whether the reason is lack ofconfidence in the sales forecast or the need to develop a forecast that meetstheir particular needs, the use of many forecasts creates a moving target thatis difficult, if not impossible, to hit. Operating from many plans is a handicapthat creates misinformation, excess inventory, lack of accountability andresults in the creation of a cycle of poor forecast credibility that is difficult tobreak.

The benefits of collaborative forecastingConsidering that `̀ in 1996 about $700 billion of the $2.3 trillion retail supplychain was in safety stock'' or, put another way, `̀ almost 30 percent was tied updue to waste and inefficiency'', solutions are needed to overcome the issuessurrounding the forecasting process (Lewis, 1998, p. 17). Collaborativeforecasting is one of the ways that many companies have found to overcomesome of the inherent problems with traditional forecasting and at the sametime support the supply chain management initiative of their companies.Collaborative forecasting is a method in which the knowledge andinformation that exists internally and externally is brought together into asingle, more accurate, forecast that has the support of the entire supply chain.While specifics may differ between companies, it is an approach that seeksand uses the available information and expertise of the entire supply chain.The process of collaborative forecasting coordinates the gathering ofexpertise and information from diverse sources and the consensus buildingthat turns these inputs into a more accurate, effective forecast used by theentire supply chain.

Collaborative forecasting is an approach that breaks down the functionalsilos or `̀ islands of analysis'' and opens the supply chain's information flow tothe benefit of the entire chain. Inventory is replaced with information andpartnerships are formed internally and externally that support supply chainmanagement objectives. Issues of complexity are overcome by informationsharing and cooperative solution seeking. In a collaborative forecastingprocess, all supply chain members contribute their particular expertise to thedevelopment of the best possible forecast, as opposed to the former many, lessthan optimal, forecasts. Sales brings information from customers fresh from thesales field, as well as information about current promotions. Marketingcontributes the latest market trends, new products and product changes.Production and manufacturing bring their expertise about manufacturing

Page 5: Supply chain forecasting – Collaborative forecasting supports supply chain management

BPMJ6,5

396

capacity and efficiency. Purchasing brings information from the suppliers.Forecasting brings the information from historical patterns and statisticalanalysis and ties all the pieces together.

One benefit of a collaborative forecasting approach is that it reduces acompany's reliance on historical records. The baseline of historical informationis supplemented by current knowledge about specific trends, events and otheritems than often undo the key assumption that history will repeat itself.Collaborative forecasting decreases the necessity to rely exclusively on history.Historical information remains a vital piece of the forecast as the baseline andis especially important in the longer term view where specific information maystill not be available. Eric Stellwagen, vice president of Business ForecastSystems, says, `̀ historical information is still a valuable asset. While long-termforecasting does have a basis in historical records, it also involves variableslike changing market share, budgets, and promotional schedules. That's why aconsensus forecast is valuable; it relies on judgement founded on specificknowledge'' (Fulcher, 1998, p. 92).

Using a collaborative or consensus forecast also helps to ensure that allfunctions are operating together as one supply chain using a single plan. Manycompanies are realizing the importance of having one forecast and are usingthe collaborative approach to achieve that objective. Terry Hisey, a principal inthe manufacturing segment of Unisys says, `̀ many companies are starting torealize the importance of combining the disparate parts of their forecastingculture. Now manufacturers are moving toward developing a single forecast inan effort to improve inventory management, increase accuracy, speed ordercycles and meet service level commitments'' (Dilger, 1998, p. 44).

Companies that are using some type of collaborative forecasting approachare impressed with the results. The study conducted by Drs Mentzer and Kahnfound that over half the companies that they surveyed (478), were trying sometype of negotiated or consensus based forecasting approach and that thesecompanies were more satisfied with their results (61 to 69.8 percent describedthemselves as satisfied with their forecasting approach) (Mentzer and Kahn,1997, p. 11). Wal Mart, Sears, Sara Lee and Warner Lambert are a few examplesof companies satisfied with their forecasting results. Through internal andexternal efforts they have been able to reduce time, cost and slack from theirsupply chains and are now better able to coordinate inventory levels withchanging demand. Heineken USA has used collaborative forecasting andassociated technology to cut order cycle time from 12 weeks to four or fiveweeks (Verity, 1997, p. 14). That translates to fresher product and happiercustomers, the goal of supply chain management. Reynolds Metals, makers ofReynolds Wrap, has found `̀ that even a 1 percent improvement in forecastingcan translate into millions of dollars in savings'' (Fryer, 1997, p. 140). Thesesavings come from all affected functions or each of the links of the supplychain. The company-wide combination of these benefits will result in improvedcustomer service, lower supply chain costs, and reduced risk for the company.Although the first of these is obvious, the supply chain cost savings come in the

Page 6: Supply chain forecasting – Collaborative forecasting supports supply chain management

Supply chainforecasting

397

form of lower inventory levels, lower production costs, lower incidence of trans-shipment (which occurs when product is originally shipped to one location anddemanded at another), and lower incidence of product obsolescence (becausethe shelf-life expired) (Mentzer et al., 1997, p. 8).

The process of collaborative forecastingLike many specific supply chain strategies there is no single correct method forcollaborative forecasting. The collaborative forecasting process is developed tomeet the specific needs of each individual company. Some of the mostinfluential factors affecting a company's specific approach are how fullydeveloped the company's supply chain management approach is, its particularbusiness environment, the available technology, and their existing internal andexternal relationships. However, there are some key points that apply to allcompanies who are interested in moving to a collaborative forecastingapproach.

The collaborative forecasting process usually begins with the departmentcharged with responsibility for the forecast or those most greatly impacted byits effects. Deciding who or what area will be responsible for the collaborativeforecasting process is a critical first step. During the process of supply chainmanagement implementation, many companies decide to move their salesforecasting departments into their supply chain or logistics organizationalstructure, and out of more traditional locations like sales, marketing or finance.This move supports the premise that it is demand that drives the activities ofthe supply chain and therefore is more closely linked with those functions. Inthe collaborative forecasting approach the forecasting department willcoordinate the collaboration of all incoming information, facilitate theconsensus building and disseminate the resulting forecast. Because of thiscritical role in the process companies should first take steps to ensure theresponsibility for forecasting is organizationally located in a way that supportsits key role in the supply chain.

To implement a successful collaborative forecasting approach onedepartment and often one person in that department must lead the charge. Thisperson initially will be someone who is willing to present the costs of forecasterror and the benefits of a collaborative approach. Forecasting is and must beviewed as a critical business process and the forecasting department and itsleadership must support and defend that fact. This role has been described asthe `̀ forecasting champion'' (Mentzer et al., 1997, p. 3). The forecastingchampion, be it a single person or a department, is critical to successfulcollaborative forecasting for several reasons. Many functional areas use theforecast and it cannot be created without considering its effect on the entireorganization. The forecast champion understands this and can effectivelycommunicate and lead the organization to share this understanding. Theforecast champion is knowledgeable in forecasting methods and technologiesand has the ability to train others involved in the process. A `̀ forecastingchampion'' or leader is critical to the collaborative forecasting process as the

Page 7: Supply chain forecasting – Collaborative forecasting supports supply chain management

BPMJ6,5

398

person or group who can emphasize the critical nature of the process andfacilitate the cross-functional efforts required to achieve the rewards availablefrom improved forecasting.

With the forecasting leadership and organization in place, the collaborativeforecasting process moves to the next step of forming the forecast collaborationgroup. The membership of this group is, again, specific to each organizationand its needs. Its members should represent a variety of functional areas thatall have a stake in the resulting forecast and its impact on the company. Theserepresentatives will usually include members from sales, marketing, logistics/operations, finance and information systems, but can also include membersfrom external partners like suppliers and customers. Including participantsfrom internal and external sources helps ensure that the most recent and bestpossible information is included in the final forecast and also ensures that theforecast addresses the changing needs and environments facing businesstoday. Every member contributes different yet vital information to the process.Elliot Sipos, president of Adapta Solutions, a supply chain planning softwarevendor, says, `̀ People in marketing may be more optimistic than people inproduction. In the past, marketing folks would put together a forecast, butproduction personnel would put together what they considered to be a moreaccurate forecast. With a consensus forecast everybody in the organization ison the same page. And when there is only one forecast number to shoot for, itsimplifies reporting and performance evaluation. More importantly, ascompetition increases, manufacturers must synchronize operations. Both overor under stocking inventory cause problems and decrease a manufacturer'scompetitiveness. Having an unified forecast helps eliminate those problemsbecause it uses enterprise-wide knowledge and is more accurate than atraditional forecast'' (Fulcher, 1998, p. 88).

The first tasks of the group will be foundational. As a group they will decideon the goals, objectives and immediate needs of the collaborative forecastprocess. They will identify the factors, processes, and technologies that impactthe forecast, as well as identify the relevant sources of information available.These factors and conditions will include those that are internal as well asexternal in the competitive environment of the market and economy at large.Also to be addressed are the informational needs of all forecast users andensuring that information can be accessed at all necessary levels. This means,if deemed appropriate, a specific salesperson or territory should be able to drilldown to see the forecast at their specific level of detail, while marketing orfinance are also able to see higher levels of forecast information. All of thesefoundational issues and first steps should be done based on the current state ofthe company's forecasting processes and systems. Their current and mostimmediate needs and goals should be addressed first, turning to long-termgoals and improvements as the process grows and develops.

A large and important part of the collaborative group's initial work will bethe identification and review of the types of information available to theprocess. How easy or difficult this information is to obtain and share, as well as

Page 8: Supply chain forecasting – Collaborative forecasting supports supply chain management

Supply chainforecasting

399

how it will flow through the process must be examined. Each functionalmember will usually have access to specific areas of information that may berelevant to the forecast process. Sales is the source of very important pieces ofthe information mix. They usually have the most recent and accurateinformation affecting the close in time periods. Their information comes freshfrom the field and includes customer based details on promotions, pricing,distribution gains/losses, store openings/closings, plan-o-gram changes, etc.Marketing brings information on new or changing items, trends in themarketplace, competitive information, market share changes and IRI typeinformation. Operations/logistics brings information concerning production,manufacturing and distribution constraints and issues. Forecasting providesthe historical baseline and associated statistical analysis. Other groups bringother relevant pieces of the information mix such as supplier and sourcingissues, relevant economic drivers and even weather information that may berelevant to certain organizations. Input from the information systems area willbe critical to decisions about how and if certain types of information will beaccessible to the group and the associated cost in dollars and time.

Once the relevant information is decided upon and available, the next step isto decide on the process by which the various pieces of information will bebrought together. Companies often choose between a monthly review meetingprocess or a Delphi type of approach. In the monthly review meeting processthere are usually at least two meetings during the month, scheduled on aregular basis. The first meeting is often for the purpose of gatheringinformation. The baseline forecast is presented and other members presentinformation affecting the time period or products being reviewed. Theinformation may be incorporated into the forecast at that meeting or the partiesmay take what they have learned and prepare forecasts improved by theadditional information. The purpose of the second meeting is to bring thealternative forecasts together and work through issues to arrive at a consensus.The consensus forecast is then presented to management for their approval andsubsequent entry into the company's sales and planning systems. In the Delphitype of approach, the baseline forecast is circulated through the group andchanges or suggested changes are made by members and then passed along.The supporting reasoning for changes can be attached to provide explanation.After the forecast has cycled through the entire team or at pre-determinedpoints in the process, the team should gather to review the final product, workthrough areas of disagreement, and end up with a final consensus forecastwhich is also presented to management for approval. Another version of theprocess could include a roll up of very low level (specific account) informationthat is summed up to higher levels where adjustments can be made based onthe additional broader based information available as it flows up the chain inthis direction.

Whether either of these processes, a combination, or something completelydifferent is used, the process should become a monthly cycle. The processshould include analysis of the actual sales versus the forecast and the creation

Page 9: Supply chain forecasting – Collaborative forecasting supports supply chain management

BPMJ6,5

400

of a baseline forecast based on historical information. Of course statistical toolsare invaluable in creating this baseline forecast and should be used if available.Information must be included from all members and a calendar of the timetablefor inputs and outputs should be published and supported. The assumptionsand information used in the process should also be documented and included asa part of the process. The forecasting group becomes the keeper and organizerof the process and should be the group that assimilates the changes andassumptions together into the forecast. A critical part of the monthly cycle isthe management review and approval meeting, which results in the finalforecast that is published and entered into all relevant systems includingmanufacturing planning, sales and finance. The cycle, once established,becomes a continuous improvement cycle in which changes and improvementsare constantly made to push the standards and subsequent performance toalways-higher levels. Charlie Smart, president of Smart Software of Belmont,MA says that, `̀ Even if a long term forecast doesn't prove to be accurate, thepoint is that it can be revised when pertinent information is available. If amanufacturer uses a consensus forecast for next year, the individuals willprobably meet once a month. Each month they can combine a statisticallygenerated forecast with their knowledge of actual demand. Their forecasts forthe rest of the year should start to become more accurate because the reportedactual numbers contribute to the ongoing forecast'' (Fulcher, 1998, p. 92).

To ensure that forecast accuracy and related supply chain performanceactually do improve as a result of the collaborative process, measurement andincentives must be a part of the process. What may now be a standard clicheÂremains true: `̀ What gets measured gets rewarded and what gets rewarded getsdone.'' Merely asking for and bringing together the information to create abetter forecast is not enough. If results are not measured and the participantsgiven real incentives to provide quality and timely information the process willend up as nothing more than a collection of good intentions. Many companieshave bonus and/or management by objective performance evaluationprograms that can be used to provide real and necessary incentives to theorganization regarding the accuracy of the collaborative forecast.

Measurements should be included that demonstrate the success of thecollaborative efforts, not just at a fixed point in time, but that measure the rateof improvement over time. Measurements can vary, but should include themeasurement of the actual versus the forecast, frozen at some point in time(usually two months out), for consistency and comparability. A commonmethod is to compute the absolute error for each item (the actual minus theforecast, divided by the actual, without the sign). The cumulative error istherefore the sum of the absolute errors, divided by the sum of the actuals.Another key measure to include is a bias indicator that shows the percentage ofitems that were either over or under forecasted. The bias indicator can point outtrends and tendencies to over or under forecast certain items and can be criticalto improving the affected forecasts. Forecasting texts are good sources ofmeasurement methods. The specifics are not as critical to this discussion as the

Page 10: Supply chain forecasting – Collaborative forecasting supports supply chain management

Supply chainforecasting

401

fact that measurement is done and the results published and used to provideincentives to the organization. The resulting improvements, in turn, reinforcethe importance of the forecasting process.

The caveats of collaborative forecastingAs the process of collaborative forecasting is implemented there are issues thatmust be addressed and overcome. Collaborative forecasting requires a varietyof personalities from a variety of backgrounds to work together to achieve aforecast that can be used by the entire company. Often in group situations thereare personalities that tend to be more dominant than others. In the collaborativeforecasting process there is a risk that the opinions of such dominantpersonalities can control the resulting forecast. Setting strict rules of order formeetings and other parts of the process can control many of these personalityissues. The rules of order should include items such as limiting the length oftime any one person can control the discussion, not allowing certain types ofnegative comments and even requiring that all members participate in thediscussion. Another possible method of controlling personality issues may beto have a group membership that changes or rotates on a periodic basis.Another suggestion is to `̀ ask everyone involved to make his or her forecastsindependently and to write them down prior to a group meeting. At themeeting, compare the forecast numbers, allowing participants to ask questionsto understand the logic behind a number or to explain their own reasoning. Besure not to allow anyone to make disparaging remarks about a forecast. Onceall the participants have a clear understanding of the minority as well as themajority viewpoints, ask the group to make and write down new forecastsindividually'' (Teplitz, 1997, p. 78). Strong leadership from the group leader,earlier called the forecast champion, is also extremely important to ensuringthat the process is fair and results in a truly consensus forecast.

The collaborative approach asks some members of the organization to makesignificant changes to the ways that they have worked in the past, and as aresult, change resistance issues are also common. In some cases all the effortthat goes into creating an improved forecast process, produces no real resultsbecause participants do not actually change their behavior. Some individualsstill produce and use their own forecasts, while giving only lip service to thecollaborative process. The collaborative process does in fact require more workfor many of the participants. Many participants were not previously involvedwith the forecasting process at all, and may view the process as merely extrawork. In addition, the collaborative forecast process takes time to put in placeand results do not occur over night. Before real benefits can be realized there isa learning curve for participants, systems to develop and process decisions tobe made, all of which require time. All of these issues combine to make theprocess change to collaborative forecasting a challenge for the organization.However, the benefits that can be achieved in improved supply chainperformance and, ultimately, profitability will ultimately overcome all of theseobjections. Rex Sprietsma, marketing research manager for Babson Brothers, a

Page 11: Supply chain forecasting – Collaborative forecasting supports supply chain management

BPMJ6,5

402

manufacturer of dairy farm equipment, agrees. He says, `̀ a consensus forecastrequires extra work from people not normally involved with forecasting, suchas sales personnel, but it offers valuable benefits. It's an extra exercise for them,but it helps the company be more competitive, improves customer service andreduces production costs'' (Fulcher, 1998, p. 89). The support and involvementof senior management is critical to overcoming these change issues.Management's commitment to the process, communication of the expectedbenefits, as well as implementation of real incentives that are tied to theforecast will be key steps to helping the organization adopt a successfulcollaborative forecasting approach.

Concerns may also exist regarding whether it is truly possible to agree on asingle forecast number. The reality is of course, that it is not always possible toreach agreement on such an important facet of the business. This is a keyprinciple of the collaborative forecast: agreement on the forecast may not bepossible, and is not absolutely necessary. What is absolutely necessary isacceptance. At some point the forecast must be accepted. Some situations mayrequire that there be a person or group in upper levels of management whomake `̀ tie-breaking'' decisions in situations where agreement cannot bereached. Regardless of the method, at some point the forecast must become thenumber that is presented and accepted as the number that drives the activitiesof the company. `̀ Everyone collaborating on the forecast should recognize thatthe company is responsible for both customer service and inventory. Oneshouldn't be sacrificed in favor of the other'' (Fulcher, 1998, p. 92). Processparticipants must, at this point, move beyond personal feelings to accept andsupport the forecast. Acceptance and support are demonstrated by theprofessional conduct, commitment and superior effort that result in theachievement of the plan that has been put in place.

The technology of collaborative forecastingThe technological advances made in recent years have had a direct impact onthe ability of companies to make dramatic improvements in all areas of supplychain performance. The cost and speed of personal computers has enabled evensmall businesses to have access to information and software that are constantlyimproving the ability of organizations to manage the complexity of theirbusiness. The collaborative forecasting process has also been facilitated by thedevelopment of more sophisticated methods of gathering and processing theinformation critical to the development of an improved forecast.

Forecasters are no longer obligated to rely solely on historical information.Data for forecasting now are available from a multitude of sources includingPOS (point of sale) data directly from the consumer transaction and marketdata from sources like IRI and Nielsen. Salespeople can provide up to theminute feedback on account level sales activity. Customers, inventory positionsand promotional activities are also often available. A wealth of information isavailable but often in different systems, making the integration of this vitalinformation difficult. Decision support technologies are being developed that

Page 12: Supply chain forecasting – Collaborative forecasting supports supply chain management

Supply chainforecasting

403

are able to bring data together and make it useful information by making itmore easily accessible and useful for planning and reporting. Many newforecasting systems are now able to allow information to be viewed at differinglevels of detail depending on the requirements of the current analysis.

Massive amounts of data are stored in electronic warehouses that are nowbeing mined for pertinent information to assist with forecasting as well asmany other processes. Wal Mart, the consummate leader in supply chaintechnology, is now using data mining software to help locate and organize themasses of information stored in its vast data warehouse. The data miningsoftware uses advanced algorithms to sift through the volume of stored dataand detect patterns and relationships. These patterns are used to update WalMart's automated ordering and replenishment systems. `̀ Already data mininghas turned up very different buying patterns from store to store andthroughout the course of the year. Wal Mart expects to save millions andmillions of dollars on inventory costs by getting a better handle on seasonaland week to week sales variations,'' says Rob Fusillo, director of ReplenishmentInformation Systems at Wal Mart (Stedman, 1997, p. 64). Data mining is still afairly new concept for many businesses and the technology is still fairlyexpensive. It is a concept that is still mostly out in the future for manycompanies at this point, since many are just now working on building theirdata warehouses.

The Internet, no longer a new concept to most computer users, is now beingused as a cheaper, quicker way to exchange information and develop moreaccurate forecasts. In collaborative forecasting initial efforts may focus ongathering useful knowledge that exists inside the company, but such usefulinformation also exists outside the company. Customers and suppliers areincreasingly willing to share information to improve supply chain efficiencyand in turn make their role easier. Leveraging all the knowledge that existswithin and without an organization is a complex task, but an initiative calledcollaborative forecasting and replenishment, or CFAR, is addressing this issue.The concept was created by the firm Benchmarking Partners in Cambridge,MA, along with software vendors SAP, Manugistics, as well as retailer WalMart and manufacturer Warner Lambert (Michel, 1997, p. 10). CFARtechnology uses the Internet to coordinate the collaborative planning andforecasting efforts of external supply chain partners, or off site distributors andsales people. Using the Internet overcomes issues of platform incompatibility,time lags and avoids the cost of a significant investment in order to participatein the system. Ellen Valentine, vice president of marketing at AmericanSoftware, says, `̀ The advantages the Web offers is that it enables much morecollaboration to take place between a corporate headquarters and its remotedistributors or sales force. Not using the Web would require a sophisticatedclient/server system with software at each location'' (Dickey, 1997, p. 1).

The CFAR concept has been touted as having the potential to provide asmuch as $179 billion in savings by reducing wasted inventory. Andre Martin, aprincipal of Retail Pipeline Integration Inc., a consulting firm, says, `̀ CFAR's

Page 13: Supply chain forecasting – Collaborative forecasting supports supply chain management

BPMJ6,5

404

goal is to improve the accuracy of sales forecasts and reduce inventory. Add tothat integrated planning among stores, distribution centers, and suppliers andmore accurate inventory systems and the retail industry could eliminateuncertainty and eliminate inventory'' (Caldwell et al., 1996, p. 46). The resultsthus far have been very positive. Warner Lambert and Wal Mart participatedin a pilot of the CFAR project for Listerine products in 1996. During the pilot,replenishment lead times were cut from 12 weeks to six weeks. The sharedinformation helped to prevent the need for safety stocks and enabled WarnerLambert to more efficiently provide product to Wal Mart. Other companiessuch as Home Depot, Proctor and Gamble, Revco, Rite Aid and Target joined ina broader pilot in 1997. American Software's version of the concept, called theResource Chain Voyager, has been implemented by Heineken USA, whereorder lead times have been reduced from 12 to ten weeks to four to six weeks. Inthe future, the concept may even be applied to the transportation planning areato assist with improved efficiency in the deployment of transportationresources.

CFAR is not a complicated decision support tool. It merely facilitates thecollaboration of the cumulative intelligence of supply chain partners. WithCFAR, one partner generates a forecast and passes it to the other partner viathe Internet, where it is posted to a secure, shared Web server. The receivingpartner accepts it, may change it, or add comments. The other partner can thensee those changes and/or notes and may, in turn, accept it or make furtherchanges. The process continues until agreement is reached and both partnersaccept and use the final forecast. At Heineken, the distributors enter forecastrecommendations and replenishment orders into their own sub pages of theplanning system, called HOPS (Heineken Operational Planning System), whichroll up to a total time phased plan for replenishment orders. Distributors canmake changes to the plan as local conditions change. These changes areavailable in real time at the Heineken brewery in Europe, who can then adjustproduction and shipment schedules to correspond to real needs. As a result ofthe concept's popularity, many software vendors are now developing or havealready developed applications that use CFAR type collaboration, includingSAP, Manugistics, American Software, Logility, PeopleSoft Inc., and i2Technologies.

The future of collaborative forecastingAs technology becomes faster and smarter and as the willingness of supplychains to share information increases, the forecasting function may becomedramatically different than it is today. Replacing inventory with information isone of the fundamental principles of supply chain management. In thecollaborative forecasting environment, companies attempt to supplementstatistical, historical based information, the more traditional forecastingapproach, with information gathered from the customer, the market, the salesforce and other sources. This supplemental information replaces some of theuncertainty that exists in the forecast and therefore replaces the inventory

Page 14: Supply chain forecasting – Collaborative forecasting supports supply chain management

Supply chainforecasting

405

created to cover that uncertainty. As supply chain partnerships expand andincrease, and the supporting technology also advances, the supplementalinformation will also increase in its quantity and quality. Forecasting, as weknow it, will no longer exist. The forecasting function will likely evolve into ademand management role where there is no real prediction of demand. Theforecaster's role will instead be to coordinate and manage the informationreceived so that the supply chain knows exactly what to produce, when toproduce it and where it will be delivered. Many companies are already well ontheir way to what has been called a `̀ zero forecasting'' environment.

This concept is essentially a shift from independent demand to dependentdemand. Traditionally, independent demands are thought of as the demand forfinal product from a company's external customers. Independent demand mustbe forecasted or predicted because of the associated uncertainty. Dependentdemand is a calculation. Companies have pre-determined bills of materials thatestablish exactly what is necessary to build each product. Every bicyclerequires two wheels, two pedals, one seat and one set of handlebars. Dependentdemands are calculated based on the demand at the next highest level ofdemand. Each link of the supply chain shares information with the next linkand passes on their requirements. Dependent demand is not forecasted becausethe requirements are certain. As the forecasting process evolves, moreindependent demand will be replaced with dependent demand on all levels inthe supply chain. Walter Goddard, president and CEO of Oliver WightInternational, explains the advantages this way. `̀ With dependent demands,each level in the supply chain shares its needs with the next lower level, such ascustomers sharing future plans with their suppliers, or a manufacturer passingproduct plans to their components. By linking each level in the supply chain,dependent demands produce better planning. The results are that material,labor, equipment, tooling, engineering specifications, space, transportation andmoney all can be synchronized, ensuring that the flow of products through thesupply chain moves swiftly and economically'' (Goddard, 1998, p. 1).

ConclusionSupply chain management is built on the principles of partnerships and thedevelopment and use of the connections that exist between the links of thechain to provide information that will increase the efficiency of all members inthe chain. Success stories abound describing lower costs, shorter lead times andincreased customer service. Collaborative forecasting applies supply chainmanagement concepts to the forecasting function and uses availableinformation and technology to force a shift from independent, forecasteddemand to dependent, known demand. Eventually, the future of forecastingmay evolve to the point where forecasting is not even necessary. Demandinformation will be supplied completely by supply chain partners and the needto predict demand will be eliminated.

The focus of collaborative forecasting is not merely on the improvement offorecast accuracy. The foundation of the concept of collaborative forecasting,

Page 15: Supply chain forecasting – Collaborative forecasting supports supply chain management

BPMJ6,5

406

as with the entire concept of supply chain management, is to leverage theinformation that exists internally and externally to improve the performance ofthe supply chain. George Palmatier, also of Oliver Wight, provides fivequestions that assist in the development of an effective forecasting process thatcan lead to a better performing supply chain:

(1) Does the demand planning process lead to a commitment to sell andmanufacture the plan?

(2) Does it lead to an improved understanding of the customers andmarketplace?

(3) Does it lead to improved relationships with customers?

(4) Does it result in improved communications between all departments inthe company that, in turn, enhance the performance of each department?

(5) Does the process enable the company to operate with one set of numberseliminating second-guessing (Palmatier, 1998, p. 10)?

Information, cooperation and relationships throughout the supply chain drivethe collaborative forecasting process and provide the means to answer thesefive questions positively.

Achieving the benefits of collaborative or supply chain forecasting will notbe without its difficulties. It requires changes to traditional behaviors and waysof thinking about the forecast process. Goddard says, `̀ Specifically it requires aspirit of partnership among customers and suppliers, one that will lead to anexchange of believable, timely, accurate information. The easy part is thetechnical side such as communicating data via EDI. The tougher challenge is toestablish a trusting relationship with both parties striving to help the other.Because the rewards are great, it's inevitable that this will happen. The sametalented people who are responsible for forecasting should also be responsiblefor aggressively reducing it. Instead of being constantly blamed for inaccurateforecasts, they will gain great satisfaction by contributing higher qualityinformation to all of the other managers of their business'' (Goddard, 1998, p. 3).

References

Caldwell, B., Stein, T. and McGee, M.K. (1996), `̀ Uncertainty: a thing of the past?'', InformationWeek, 9 December, p. 130.

Dickey, S. (1997), `̀ Forecasting and ordering system rides the Net'', Midrange Systems,17 January, p. 1.

Dilger, K.A. (1998), `̀ Predictive prowess'', Manufacturing Systems, January, pp. 40-50.

Chain Store Age (1998), `̀ Don't downplay forecasting, KPMG warns'', March, p. 97.

Fryer, B. (1997), `̀ Forecasting for dollars'', Information Week, 16 June, pp. 140-2.

Fulcher, J. (1998), `̀ A common vision'', Manufacturing Systems, February, pp. 88-94.

Goddard, W.E. (1998), `̀ The forecast for the future zero forecasting'', http://ollie.com/ZForecast.html,accessed 6 August, pp. 1-3.

Lewis, T.G. (1998), `̀ Electronic warehouses'', Datamation, December/January, p. 17.

Page 16: Supply chain forecasting – Collaborative forecasting supports supply chain management

Supply chainforecasting

407

Mentzer, J. and Kahn, K.B. (1997), `̀ State of sales forecasting systems in corporate America'', TheJournal of Business Forecasting, Spring, pp. 6-13.

Mentzer, J.T., Moon, M.A., Kent, J.L. and Smith, C.D. (1997), `̀ The need for a forecastingchampion'', The Journal of Business Forecasting, Fall, pp. 3-8.

Michel, R. (1997), `̀ Best practices make perfect'', Manufacturing Systems, April, pp. 10-12.

Palmatier, G. (1998), `̀ Forecast measurement and evaluation'', http://ollie.com/Forecast.html,accessed 6 August, pp. 1-10.

Stedman, C. (1997), `̀ Wal-Mart mines for forecasts'', Computerworld, May 26, pp. 63-4.

Teplitz, P. (1997), `̀ Conjuring up a profitable forecast'', Catalog Age, November, pp. 76-8.

Verity, J. (1997), `̀ Collaborative forecasting: vision quest'', Computerworld, November, pp. 12-14.


Recommended