BLOG Searchingfor Competitive Intelligence, Brand Image, andReputation ManagementDo you know what is being said about you in tbe blogosphere? Sbould you care? Aren't blogs just

for kids, political campaigns, and geeks, of no value in business? If you tbink tbis, you're likely

missing out on an increasingly powerful information source.

Why have blogs, which aresimply Weh pages withreverse chronologicallyarranged individual

posts, become so powerful? Accord-ing to David Sifry of Technorati[www.sifry.com/alerts/archives/000298.html]. the number of hlogshas been doubling at least every 5months. The current estimate isthat there's over 10 million blogs inthe biogosphere. Yet it's not the largenumber that creates the stir overblogs. Their power comes from thespeed of information diffusion, theirlow barrier to access, and readers'ability to scan many blogs throughXML feeds. The vast majority arenot journalistic endeavors or politi-cal advertising. Blogs are as indi-vidual and unique as their writers—

and can damage your company orbrand very quickly.

CONSEQUENCES OF BLOG POSTSTake, for example, the case of

Kaiser Permanente. A former em-ployee, Elisa D. Cooper, wbo callsherself the "Diva of Disgruntled,"posted a link to private customerhealth information, including infor-mation on lab tests, on her blog.Kaiser first learned about the postsfrom the federal Office of Civil Rightsin January 2005, although she saysshe posted the link in July 2004 andtbe site to which she linked had beenonline since 2002. A lawsuit, filed byKaiser, is in progress.

Another recent example is thecase of the easily picked Kryptonitelock. In 1992, a bicycling enthusiastmagazine. New Cyclist, published anarticle stating that circular lockssimilar to Kryptonite bike locks canbe picked witb a Bic pen. The publi-cation of tbis information bad verylittle impact and Kryptonite madeno changes to the lock design. Thenin September 2004, a video posted toa blog showed how to pick a Kryp-tonite lock in less than 30 secondsusing a Bic pen. In less than 2weeks, this video was played in allof the major media, lawsuits werebrought against the company, and amovement was started to force thecompany to recall the "faulty locks."Even though the information was

available to bike shops and bikeowners for more than 10 years, ittook blogs to precipitate tbe crisis.

In tbe mainstream media wherestory leads can originate from com-pany press releases, companies are incharge of the message. In the blogosphere, where the power is in tbe net-work, the customers control the mes-sage. A good illustration of this is inHarvard Business Review's realistic"Glove Girl" case study ("A Bloggerin Their Midst," by Halley Suitt, vol.81, no 9 [September 20031: pp.30-40). Glove Girl, who's an assis-tant foreman at a surgical glovemanufacturing plant, blogs aboutthe company's products on her per-sonal blog. She achieves more audi-ence than the official marketingcampaigns and upstages the compa-ny's CEO at a major trade show.When she says something positiveabout a line of gloves, sales increase.When she blogs her questions aboutthe company's ethics of doing busi-ness with a clinic with a poor record,the deal is in danger. Commentatorson the case study were David Wein-berger, Pamela Samuelson, RayOzzie, and Erin Motameni.

The blog cycle—the time betweenlearning about something to having iton a blog feed—is very short. A blog-ger reads something of interest, andthen posts it to his or her blog withinseconds, frequently without verifyingfacts. Feed aggregators query the site

1 6 www. onlinemag. net

The blog cycle—

the time between

learning about

something to

having it on

a blag feed—

is very short.

July/August 2005 17

about every hour, so readers all overthe world will have almost instantdesktop or mobile access to the storywithout searching for it. This ean begreat viral marketing of your prod-uct, if the post is positive. However,if the posted information is negativeor even wrong, and your company isthe last to know, damage control willbe very difficult.

telliseek offers the content analysisof posts to indicate if they are nega-tive or positive.

While providing easy solutions,the above produets may not offer thepreeision of a custom search and maybe very expensive. It is not necessaryto license a eommercial product to ef-fectively search across blogs and toset up alerts for new content. Onee

Knowing the structure of blogs and

blog posts in general will enable you

to design your search nnore effectively

COMPREHENSIVE SERVICESVERSUS CREATING YOUR OWN

Are these examples, both real andfictionalized, enough to convince youto search across blogs for informa-tion about your company, your com-petitors, and your brand? Are youconvinced you sbould create alertsso that you are up on the buzz in theblogosphere? If so, your next task isto decide how you will proceed.

As I write this, I am seeing manyannouncements of blog monitoringproducts and research from startupcompanies and established researchlabs. While this supports my pointthat it is important for companies tomonitor the blogosphere; it makes itmuch more difficult to select a solu-tion. Several companies have devel-oped independent products thatmatch new blog posts, search feeds,or search blogs for company, brand,or product information. Others part-ner with major information servicesto integrate this feature into exist-ing business and news researchproducts. For example, Moreover isadding blog information to Factiva'sreputation management products.Intelliseek, BuzzMetrics, and More-over are just a few of the companiesmarketing a monitoring product. In-

you understand the blogosphere andhow to use advanced features of gen-eral and feed or blog-specifie searchengines, you'll be able to get betterresults and will be able add newsearches or tweak existing searchesfor better precision or recall.

THE STRUCTURE OF BLOGSKnowing the structure of blogs

and blog posts in general will enableyou to design your search more ef-fectively. In addition, knowing thespecific structure of Web pages pro-duced by the most common bloggingsoftware packages will allow for ef-fective field searching. Although theabove definition of blogs is broadenough to cover a large variety ofWeb sites, the lion's share of themarket is held by a few hosted blog-ging packages such as Blogger, TypePad, and Live Journal, according toElise Bauer's Weblog Tools Mar-ket—Update, February 2005 [www.elise.com/web/a/weblog_tools_market_update_february_2005.php].

Blogs have a very low barrier toaccess. You post using e-mail, voice-mail. Web forms, or a downloadedWYSIWYG program. Hosted blogshave Preformatted template choicesthat allow varieties of layout and

color without detailed knowledge ofHTML, cascading style sheets(CSS), or XML. The content can beanything from a sparse list of linksto extensive collections of essays,videos, recordings, or pictures.Feeds are XML-formatted abbrevi-ated posts or lists of headlines thancan be gathered, displayed, and con-sumed efficiently in an aggregator.

Blogs generally have home pageswith somewhat static content and alist of the most recent posts. Blogsalso have individual post pages thatcontain a single post and may notdisplay all of the options from thehome page. Home page formats usu-ally consist of two to three columnsusing tables and CSS. The largestand frequently most prominent col-umn contains the posts that formthe actual meat of the blog. (See Fig-ure 1 on page 19.)

These individual posts are chunksof text and links arranged in reversechronological order. Fach post gen-erally has the following fields: title,date-time stamp, body, comments,traekbacks, and permalinks. Otherfront-page columns may be morestatic or contain automatically gen-erated content, such as headlinesfrom recent posts, links to the blog-ger's personal profile, links to thearchive, and a list of favorite Websites. Archives are created immedi-ately so the same post will occur onthe main page and on the archivespage. (See Figure 2 on page 19.)

URL STRUCTURESThe structure of the archives, post

pages, comment pages, and perma-links can be exploited for precisionsearching. The table on page 20provides examples of the format ofthe URLs.

GENERAL SEARCH ENGINESAlthough there are several spe-

cialized blog search engines, it's wisenot to overlook the general Websearch engines—Google, MSN,Yahoo!, and Ask Jeeves.

I'll start with Google. Onee youcreate your search in Google, youcan visit Google Alerts [www.google.com/alerts] to paste in your success-ful search string and set up imme-diate or daily e-mail alerts. For this

1 8 w\AA/i/.onlinemag.net

BLOG SEARCHING

Anatomy of the Main Blog Page

Title and description

Content (posts)

About the blog, profiles ofthe contributors

Blogroll

Sort by category

Archives by date

Source: httpV/stlq.info

Figure 1: Anatomy of the Main Blog Page

Anatomy of a Post

Date/timestamp

Permalink

Categories(click to find

all in thatcategory)

Citation

AimUDO Liuniii Uolioc Opfn lutm lourii,il al Ftsnolec

fmmittinrstat ' '1 " ' mi i ( " «»» •

"n li DM. opffl nxtu 11-̂ Ml on imv .̂ oinotogr 'i ti"thi.^mini|;

r ln OnUnt Jovfln*

liunrfi 1/ Dw Ot*rm Sawnd rf

^Mfld on • I-M •mH Wbl«hi1t

K. ** »<s csnmbMxint • ' I b( ' ivwwj In • <n)^ d*« ixntl

VtjOn Cftf')"ras 'OKI

Title (may behyperlinked)

# of comments-(click on permalinkto see them)

#of trackbacks(click on permalinkto see them)

Click to open theform to leavecomments

Source http://stlq,info/archjves/001707.htmi#001707

Figure 2: Anatomy of a Post

July/August 2005 ONUNG 19

reason and the overall success withmy trial searches, I highly recom-mend trying this search engine first.

To use Google Alerts for monitor-ing brand image, try this generalsearch string; " p r o d u c t name"( inurL:YYYY/MM1 MM2 ORinurL:YYMM YYMM OR inurL :a r c h i v e ) .

This works because Google index-es the "/" character and uses the el-lipsis range operator. Be sure to putyour product name in quotes. Selectthe year and the current few monthsor omit the eUipsis and use just themost current month. The archiveterm adds noise, in the form of non-

blog sites, but is necessary to bemore comprehensive. Of course,you'll need to update the dates fromtime to time to get new results.

The above method will also pickup most Movable Type and Word-Press blogs that are hosted on thehlogger's own site. The search string" p r o d u c t n a m e " ( i n u r L :b L o g s p o t OR i n u r L r t y p epad OR i n u r L : L i v e j o u r n a LOR i n u r L : d i a r y L a n d ORi n u r L : b L o g d r i v e ) is more re-strictive, but will pick up tbe hostedblogs with the majority of the mar-ket share. It is also more precise be-cause it avoids the noise from e-com-

merce and magazine sites that mayuse the date in the URL.

Moving to MSN, once you narrowdown your search, click on the RSSbutton at the bottom to subscribe tothe feed for the search so new re-sults will he sent to your aggregator.A good search string at MSN is" p r o d u c t n a m e " ( s i t e :b L o g s p o t . c o m OR s i t e :t y p e p a d . c o m OR s i t e r M v ej o u r n a L . c o m OR s i t e :d i a r y L a n d . c o m OR s i t e :b L o g d r i u e . c o m ) .

MSN lacks a command that al-lows you to search in the URL andit doesn't index slashes and other

URL STRUCTURES

Blogger URL

Post Page

Comments

Archives

UveJournal LJRL

Post Page

Comments

Archives

http://Name.blagspot.com (if Biogger-hosted)

http://Name.blogspot,com/YYYY/MiV!/words-of-title-hyphenated.html

http://Name,biogspot.com/ YYYY/MM/words-of-title hyphenated.html#

000000000000000000 (18-digit comment ID) (Note: Many biogger biogs user Haloscan or

other plug-in comment services that store the comments on nonbiogger servers.)

http://Name,biogspot.com/YYYY_MM_DD_Name_archive.html

(The date is the first date of the archive, may be weekly or monthly archiving.)

http://viflww.livejournal.coni

http://www.livejournal.com/users/Name/00000.html \

http://www.livejourna!.com/users/Name/00000,html?thread=11111 # tn 111

http://www,tivejournai.com/users/Name/YYYY/MM/DD (Month and day are optional.)

Typepad

Diaryland

Blogdrive

Post Page

Comments

Archives

URL

Post Page

Comments

Archives

Post Page

Comments

Archives

http://Name.typepad.com/Name2/ (Name2 may be similar or the same as Name;

it may also be omitted,)

http://Name,typepad.com/Name2/YYYY/MM/vi/ords_ofJitle,htm!

http://Name.typepad.com/Name2/YYYY/MM/words^ofJitle.html#comments

http://Name.typepad.com/Name2/YYYY/MM/index.html

http://Name.diaryiand.com/index.htmi

http://Name.diaryland.com/YYMMDD_00,htmln/a

http://Name.diaryland.com/oider.html

http://Name.blogdrive.com

http://Name,blogdrive.com/archive/OOOO.htmi (numbers refer to a post number)

http://Name.biogdrive,com/comments?id:=0000 (same post number)

http://Name.blogdrive.com/archive/cm-0_cy-0000_m-2_d-00_y-00^o-0,html

KEY: Y - year, where YY is a two digit year (05) and YYYY is a four digit year (2005); M - month, where MM is a twodigit month (03); D - day; N - Name is the name of the blog of account name; Title - indicates the t i t le of the post.

2 0 www.onlinemag.net

BLOG SEARCHING

symbols. The search string above isvery narrow, but since feeds can beset up, you can easily establish quitea few alerts.

In Yahoo!, you can get alerts de-livered to your cell phone or in e-mail when tbe new item is in tbenews, but not new items for the Web.Thus, you will have to manually re-run the search periodically. Here'sthe search string I prefer: "produ c t name" C inu rL ibLogspot OR i n u r L : t y p e p a d ORinurL:LivejournaL OR inurL:d i a ryLand OR inurL:bLogd r i v e ) .

Click directly on the advancesearch page to input all the host-names into the "any of these" box,then click the drop-down menu for"in the URL of the page." Yahoo!does not index symbols and does nothave a range operator. Therefore, amore restrictive search is preferablein this search engine. Because thereis no alert for the search, it might beuseful to limit the search to the last3 months. You may also look for themonth and year: "product name"inurL:YYYY inurL:MM.

Ask Jeeves owns Teoma and usesthe same search technology. I wasunable to get consistently good re-sults using either site. Alerts are notavailable. I do not recommend usingeither for this task.

BLOG-SPECIFICMany so-called blog search en-

gines really only index feeds. Feedsare XML versions of the blogs thatcan be read using an aggregator orincorporated into portals or Webpages. Many feeds are not full text.In fact, some just include the title ofthe post. Additionally, many schol-arly journals, mainstream mediaoutlets, and government agenciesoffer feeds. Searches in these en-gines provide many more relevantposts higher in the results list, butreturn fewer results overall. Beloware several of the blog specificsearch engines that I have found tobe most useful. In each of these, noextraordinary search skills are re-quired, because the assumption isthat all of the results will come fromhlogs, forums, or feeds.

Bloglines fwww.hloglines.com]is a Web-based feed aggregator nowowned by Ask Jeeves. The search en-gine allows you to search all of thefeeds to which Bloglines memberssubscribe. Once your search is com-plete, you can set up an alert withinyour account on Bloglines. For com-panies wanting to have the informa-tion delivered to their intranet, thismay be unsatisfactory.

Feedster [www.feedster.com] hasa very powerful search engine withmany fields you can exploit to pro-vide a precise search. Use the blog-only portion of the database[http://hlogs.feedster.com/] to ex-clude news sources. Once you'vetweaked the search the way youwant it, you can subscribe to a feedto get new results. It is worth re-viewing the advanced help page[www.feedster.com/help/advanced.php] to see what is available.

PubSub lwww.pubsuh.com] is notreally a search engine because itmatches your query against new blogposts from millions of feeds it moni-tors—it does not run the searchagainst stored information fromcrawled or spidered sites. In thissense, it fits into the "comprehensiveservices" section above, but it is freefor now and does not provide visuali-zation or analysis functions. Set upyour search using the box on the homepage and subscribe to the feed or hook-mark the digest page provides. It au-tomatically ORs terms, so use quotes,ANDs, and parentheses. I highly rec-ommend this source, as it monitorsthe largest collection of blogs.

Technorati [www.technorati.com] developed as a service to allowhloggers to track conversations onthe Web. It tracks over 5 millionblogs and reports when another bloglinks to one of your posts. More re-cently, Technorati has developedtags that allow hloggers to catego-rize their posts. See, for example,the tag "Bluetooth" [www.technorati.com/tag/BluetoothJ. Resultswith the same tags are importedfrom Flickr, Furl, and Del.icio.us—Web-based bookmark and photomanagers. Create an account andconduct a simple search. You willthen be given the option to "make

this a watchlist." This creates anRSS feed that will alert you whennew posts match your search.

Waypath [www.waypath.com]covers fewer blogs than Technoratior PubSub, but it allows two types ofRSS feeds of results: newest or best.It is also one of the few search en-gines that indexes the entire blog,not just the feed.

MANAGING RESULTSI've shown you many ways to get

early-warning information on thebuzz in the blogosphere—almost toomany. How do you go about settingup all of these alerts? What do you dowith all of this information? First, getan aggregator so that you can scanincoming feeds easily. If you use Blog-lines or another Web-based aggrega-tor, your subscriptions do not have tobe linked to your company (or name).In addition, you can create temporarye-mails within Bloglines to use forGoogle and other e-mail alerts. Evenif you do use a desktop aggregator,sign up for alerts using a Web-basedfree e-mail account.

Second, try each of the searchesabove with your actual company orproduct name. For testing purposes,I used a fairly new mobile phone of-fering, the "Motorola A630." Your re-sults may vary from mine so keepthe best general search engine, Pub-Sub, and the best blog search foryou. Third, set aside time in themorning to haul in your nets. It isvery tempting to check and re-checkyour feeds. Once the feeds are es-tablished, let them work!

Now that you know what is beingblogged, it's time to turn the tables,take the initiative, and create a mar-keting blog for your company or or-ganization. Your blog can counteradverse opinions in real time and inthe medium bloggers understand.This may seem like quite a bit ofeffort, but your company and yourcustomers will thank you for it!

Christina Pikas lcchnKtina.pikas@jhuapl.eduj is a liht-arian at The Johns Hopkins Uni-versity Applied Physics Laboratory.

Comments'? E-mail letters to the editor tomarydee@xmission.com.

July/August 2005 CDNUNG 21