Saving IT’s Soul: Human-Centered Information...

8
Nc~-~ ~ ~ _________________________ _____ ~ ~ ‘~, t~c, ~ ~ A new approach to information promises business benefits that fewinanagers could conceive of when focusing strictly on technology. Saving IT’s Soul: H u ma n-Centered Information Management by Thomas 1-L Davenport information technology has a polarizing effect on managers~ it either bedazzles or frightens. Those who are afraid of it shun it, while bedazzled IT de- partments frequently become prisoners of their own fascination, constructing elaborate technology architectures and enterprise information models to guide systems devel- opment. Senior executives who buy into this view promote technology as the key catalyst of business change. But such techno- cratic solutions often spec- ify the minutiae of ma- chinery while disregarding how people in organiza- tions actually go about ac- quiring, sharing, and mak- ing use of information. In short, they glorify infor- niation technology and ig- nore human psychology. It shouldn’t surprise ~nyone that human na- cure, good and bad, can throw a wrench into the best-laidiT plans, yet tech- nocrats are constantly caught of I guard by the %rratiorial” behavior of “end users.” In fact, people who are afraid of infor- ination technology may have good reason to feel that way. Companies that ballyhoo their latest management information systems or groupware usually spend little time training employees to use them. Even thosç who like computers can fi~d them- selves hobbled by the rig- id structure and rules of many IT shops. Obviously, people han- dle information in any number of ways, horn ba- sic data processing to gen- Thomas H. Davenport is a partner and director of ze- search at Ernst & Young’s Center/or information Tech- nology and Strategy in .Bos- tori and cm adjunct profes- sor at Boston Unfversitys School of Management. He is the author of Process Inno- vation: Rcengincering Work Through Information Tcch- nology and two previous HBR articks. People handle i.ifozuiat~on in myriad ways—from data processing to exchanging E-mail worldwide. DgAWU4GS B? PAUL M~tSEL Ill,

Transcript of Saving IT’s Soul: Human-Centered Information...

Nc~-~~ ~

_________________________ _____ ~ ~ ‘~, t~c,~ ~

Anewapproachto informationpromisesbusinessbenefitsthatfewinanagerscouldconceiveofwhenfocusingstrictly

on technology.

Saving IT’s Soul:Hu ma n-Centered

InformationManagement

by Thomas1-L Davenport

informationtechnologyhasa polarizingeffectonmanagers~it either bedazzlesor frightens.Thosewhoareafraidof it shunit, while bedazzledIT de-partmentsfrequently becomeprisonersof theirownfascination,constructingelaboratetechnologyarchitecturesandenterpriseinformationmodelsto guidesystemsdevel-opment.Seniorexecutiveswho buy into this viewpromotetechnologyas thekey catalystof businesschange.But suchtechno-craticsolutionsoftenspec-ify the minutiae of ma-chinerywhiledisregardinghow peoplein organiza-tionsactuallygo aboutac-quiring,sharing,andmak-ing useof information.Inshort,theyglorify infor-niation technologyandig-norehumanpsychology.

It shouldn’t surprise~nyone that humanna-cure,goodandbad,canthrow awrench into thebest-laidiT plans,yettech-nocratsare constantly

caughtof I guard by the %rratiorial” behaviorof“end users.”In fact, peoplewho areafraidof infor-ination technologymayhavegoodreasonto feelthat way. Companiesthatballyhootheir latestmanagementinformationsystemsor groupwareusuallyspendlittle timetrainingemployeesto use

them.Eventhosçwholikecomputerscanfi~dthem-selveshobbledby the rig-id structureandrules ofmanyIT shops.

Obviously,peoplehan-dle information in anynumberof ways,hornba-sic dataprocessingto gen-

ThomasH. Davenportis apartnerand directorof ze-searchat Ernst & Young’sCenter/orinformationTech-nologyandStrategyin .Bos-tori and cm adjunctprofes-sor at Boston UnfversitysSchoolof Management.Heis theauthorofProcessInno-vation: RcenginceringWorkThroughInformationTcch-nology and two previousHBRarticks.

Peoplehandlei.ifozuiat~onin myriadways—fromdataprocessingto exchangingE-mailworldwide.

DgAWU4GSB? PAUL M~tSEL Ill,

INFORMATION MANAGEMENT

cniting sophistic;ucdaccountingdocumentsto cx-changing informal E-mail messagesaroundtheworld. For the manydivcts~informationusersinlargeorganizations,only onethingis certain:eifec~tive information managementmust begin bythinking.abouthow peopleuseinformation—notwith howpeopleusemachines.While it’s impossi-ble to accountfor all the unforeseenconsequencesof information expansionandusein today’scom-panies,thefollowing threeobservationsexemplifyhow ahuman-centeredapproachto informationmanagementcontrastswith the standardIT view:o Information evolvesin manydirections,takirig

on rrwltiple meanings.While IT specialistsaredrawn to commondefinitionsof termslike cus-tomeror product, most informationdoesn’tcon-form to such strict boundaries.Forcingemployeesto cometo onecommondefinition, assometech-nologiesrequire,only truncatesthe very conver-sationsandsharingof perspectivesthat the tech-nology is supposedto ensure.Ratherthanforcingemployeestosimplify informationsothatit will fitinto acomputer,ahuman-centeredapproachto in.formation calls for preservingthe rich complexitywepreferinourinformationdiets.oPeopledon’t shareinformationeasily.Assumingthat different departments,professionals,or lineworkerswill wantto usetechnologyto shareirif or-mation is oneof the biggestmistakesexecutivesmake.Yet it is oneof thefundamentalassumptionsmadein planninganyIT system.That is, if youbuild it, peoplewill useit.o Changingan IT systemwon’t changea corn-pany’s information culture.Thepresenceof tech-nology, in andof itself, cannotwholly transformacorporation.Changingacompany’sinformationculturerequiresalteringthe basicbehaviors,atti-tudes,values,managementexpectations,andin-centivesthat relateto information. Changingthetechnologyonly reinforcesthebehaviorsthatal-readyexist.Yetinmostcompanies,manymanagersstill believethat oncethe right technologyis inplace, theappropriateinformation-sharingbehav-ior will inevitablyfollow.

At ono largepharmaceuticalcompany,for ex-ample,IT managerstried to implement shareddatabasesandother new technologiesto speedupR&D, onlyto havetheirefforts foiledby significantcultural barriers.In this case,managersassumedthat researchersinvolved in the developmentof adrugwould passalongall information about it to•the people conductingits clinical trial; if re-searchershadfoundearlyon that,say, thedrug’s ef-fect diminishedwhentakenwith certainfoods,thenpatientsin theclinical trial couldbeinstruct-ednot to takethedrugat meals.Suchearlyreleaseof data,however,rarelyhappensat this pharma-

ceuticalcompany.Clinical studiesthereforeoften haveto be redone,delayingthe drug-approvalprocesssometimesforyears.

In this company,managementpushedthe new databasesand soft-ware, but researcherswereeitherhostile or apathetic.The IT depart-merit wassofocusedon thetechnol-ogy that they hadfailed to under-standthe rigid rules of scientific

explorationthatgovernhowscientiststhink aboutinformation.Different departmentscouldn’t agreeon whatconstituteda “drug” or a“clinical trial” —

or evenwhat font theyshouldusefor researchre-ports.In this case,the rateof technologicalchangefar outstrippedthepaceof changein thecultureasawhole.Insteadof instituting new technologies,executivesshouldhaveinstitutedaprogramof cu?-tural changeto convincehighly competitivescien-tists that theywouldn’t be penalizedfor sharingearlyandperhapsincompleteresults.

Technology,after all, is neitherthe saviornorarchdemonof the informationage.At its worst, itdistractsandmisleadsus.But at its best,new sys-temscansupporttheki~n~dof informationusethatresultsin realbusinesschange.

What’sWrongwith theView from iT?Sincethefirst businessapplicationsof computers

in themid-1950s,planningandcontrolhavedomi-natedsystemsdevelopmentin largecompanies.Inparticular,the conceptof “informationarchitec-ture”hasovershadowedahuman-centeredview ofinformation.IBM createdthefirst structuredap-proachin the 1960sandhasdefinedthefield eversince.Originally named“businesssystemsplan-

-• fling” (BSP),laterversionscametobecalled“strate-gic dataplanning” and“informationarchitecture.”

The analogyto an architecturalblueprint,inwhich the locationandusesof differentroomsare

Too many managersstill believethatoncetheright technologyisin place,appropriateinformationsharingwill follow.

120 HARVARD BUSINESSP.EVIEW March-April 1994

1.Mostoftheinformationin organizations-andmostof theinformationpeoplereallycareabout—isn’t oncouterL .

The Information Factsof Life,

6. If inforñi~atioriis pow~i ~sharciteasily -

7meWi1liofj13th:I3~formation 1orinat is didy~piôpärt~

.much-~~. Max~gerspreferto get informationfrom peopleratherthancomputers,peopleaddvaluetorawinformationby inter.~

• pxttingitzxidaddingatext.

3 Themoreconipiexanddetailed-

~ in ormation.rnanageuientap-~.~proach,.�he.1css.Iikelyit is to~dflC anyoneabehavior

~ of~l~ii~1~thdison3erisdesrrabIe - °° -

- LThemoreacosnpanyknowsand; ~lO There’scaresaboutuscorebu.sinessarea, -. mationoverlothe Ii emplOyeeswill beto it informationis powerandmoney, - reallyuseful1~agreeonacommondefinitionofit.. -peoplewon’tshareit easily

- .‘- - ..!-~ -. — ‘.

- - -

.-•—..-.

)ortant,theymaketheunrealisticassumptionthatnostof a company’sinformationcanbe organizedLccordingtoafewcommonterms.

t~NaturalMess:Multiple Informationv\eanings

While informationarchitecturecan’tcapturethereality of humanbehavior,thealternativeis hardFor traditionalmanagersto grasp.That’sbecauseahuman-centeredapproachassumesinformationiscomplex,ever.expanding,andimpossibleto con-trol completely,Thenaturalworld is a moreaptmetaphorfor the Information agethanarchitec-ture.From thisholisticperspective,all informationdoesn’thaveto be commonssomedisorderandevenredundancymaybe desirable.(Seethechart,“Human-CenteredIT ManagersPocusonHowPeo-pleUseInformationRatherthanMachines.”)

Nomatterhowsimpleorbasicaunit of in.forma-non mayseem,therecanbevalid disagreementsaboutits meaning.At Digital EquipmentCorpora-tion,forexample,a“sale” totheindirectmarketingorganizationhappenedwhena distributor or re-sellerorderedacomputer;but to directmarketing,thesaleoccurredonlywhentheendcustomertookdelivery.Evenwithin directmarketing,therewere

differencesof opinion: salespeoplerecordedasalewhenthe orderwas placed,manufacturingandlo-gisticswhentheproductwasdelivered,andfinancewhenit waspaidfor.

At AmericanAirlines, thereare severalperspec-tiveson whatan“airport” is. Somemanagersarguethat an airport is any locationto which Americanhasscheduledservice;otherscountany airportgrantedthat statusby theinternationalstandardsbody. At Union Pacific i~.a~i1road,there’slittle con-sensuson whata “train’~is. Is it a locomotive,allcarsactuallypulledfrom anorigin to adestination,oranabstractschedulingentity! EvenU.S. Depart-ment of Agriculture officials can’t agreeon themeaningof“farm.”

Th~semultiple meaningsmakethejob of infor-mationmanagementtreacherousatbest.At oneoilexplorationcompany,for example,informationar-chitectsworkedfor yearson ineffectivemodelsbecausepeopleassigneddifferentmeaningsto “oillocation.”Someusersdefinedit asthe originalgeographiccoordinatesin the ground; othersthoughtit wasthewell from whichoil spran~stillothersusedthe term to referto~theoil’s currentlo-cationin a tankfarm or pipeline.Eachdefinitionfoundits wayinto computerdatabases.As aresult,it wasdifficult to shareeventhemostbasicinfor-

KkRVARD BUSINESSREVIEW M3~ch-ApZil1994

gether.But while dual information streamsaremessyandhard to control, theyseemrealisticforthis diversecompany,

A largermanagerialbarrier, however,remains:operatingwith multiple meaningsalso requiresba-sic changesin behavior—not only for informationproviders,who categorizeand collecttheinforma-tion, butalsofor users.TheCEO who is annoyedwhentold there’sno quick answerto how manycustomers(oremployeesorproducts)thecompanyhasis justasguilty of oversimplifyinginformationasthedatabasedesignerwho insistson onedefini-tionofcustomer.

And whenit is necessaryto define commonmeanings,theprocessrequiresmuchmoreman-agementparticipationandtimethanmanyassumeorwant to allot, For instance,Xerox diddatamod-eling andadministrationfor 20 years,but in thewordsof the directorof informationmanagement,“We got nowhere.”Theseinitiativesweredrivenby iT ratherthanbyseniorbusinessmanagers;theywerealwaysabandonedin favorof specificdevelop-ment projectslike the new order-processingorblUingsystem,whichyieldedobviousbenefits.- Finally,Xerox’s IT departmentaskedseniorexec-utives to identify thekeypiecesof informationon

which theentire businessshouldbe run. The executivesdebatedtheissueon severaloccasionsbutweren’tableto reachaCOflSCflSUS.They did agree,however,thattheir main prioritieswere cus-tomer,financial, andproductin-formation— in thatorder.

Xerox’s IT departmentthentook anothertack.From aroundtheworld, 15 marketingandsalesmanagers,accompaniedby theirIT counterparts,met to agreeonthesetof commoncustomerin-formation the companywoulduse.As usual,peopledisagreedaboutwhat “customer” meant.But thesemanagerseventuallyagreedto definecustomersascor-porationsthat badalreadypur-chasedproductsor servicesfromXerox andto refer to themwith acommonworldwidenumber;theyalsoreachedconsensuson 11 oth-er customer-orientedterms,in-cluding customer-satisfactionmeasures.This coordinatedap-proachallowedcountrymanagersto thencreatecustomerinforma-

tion thattheIT departmenthasnowcombinedintoaglobaldatawarehouse.

TheTroublewith InformationSharing

In today’scompetitivebusinessenvironment,itmakessenseto give informationparticularismitsdue;but asXerox’s exj~etiencewith customerin-formationillustrates,ex~cutivesmustalsodecidewhichaspectsof acompany’sinformationareglob-al. More to the point, executivesmustdeterminehowsuchinformationis to besharedeffectively—oneof thetrickiestmanagementissuesfor today’scor~ipanies.While informationarchitecturecanspecifywho controlsinformation,suchrigid mod-els don’t accountfor theunpredictablegrowthofinformationorhumannature.

Somemanagersarequick to point out the ob-vious difficulties with informationsharing,espe-cially whenit’s driven by newtechnologieslikeelectronicmaiL If sharingmakesit easierfor a

• company’semployeesto getatcritical information,it alsoopenstheway for any interestedexternalparties— competitors,attorneys,evencomputerbackers.Giventhemanyrecentandhighly visible

No unitof informationis toobasictopreventdisagreementaboutitsmeaning:USDA officialscan’tevenagreeonwhaxafarmis.

HARVARD BUSINESSREVIEW M.2xcb-April 1994

informationtechnologiesdon’t inevitably leadtoflattenedhierarchiesandempoweredemployees.Workingout informationissuesin a companywitha monolithic culture—insteadof wrestlingwithtwo competinginformation culturesthat resultfrom a merger—ofteninvolves diggingout en-trenchedattitudestowardorganizationalcontrol.

In suchcompanies,technolo-gies that promoteinformationsharingcanendup controllingemployeesratherthanempower-ingthem.Whenlowerlevelwork-ersareorderedto “share”infor-mationwith thosehigherup thecorporateladder,acutthroatinfor-mationcultureof meddlingmicro-managementcanresult.At therefiningandmarketingdivisionofa largeoil company,for example,thedivisionpresidentdelightedinbeingableto usehis computertopeerelectronicallyovertheshoul-dersof oil traders—andoccasion-ally to overrideor initiatea deal.

On the otherhand,Xerox’s ex-ecutivesupportsystemhasbeenlimited to accessingdatatwo lev-els below the user—preciselytoavoid this type of excessivecon-trol. Suchhuman-centeredtech-nology implementationsarestillrare,but theyindicatethewaymanagersmust think abouttheissuesthat informationsharingbringsto thesurface.

Populist exhortationsto thecontrary,unlimited informationsharingdoesn’twork. In fact, in-creasedinformationsharingcaneither improveor activelyharmcompanymorale.Sharinginfor-mation aboutactualcorporateperformanceis usuallygood formorale- evenwhenperformanceispoor,sinceuninformedemploy-eesoftenassumethat it’s worsethanit really is. Sharingrumors,however,canbedenioralizirig.

An informationsystemsman-agerat a New York bank, forexample,createda LotusNotesbulletinboard that he calledthe“Rumor Mill.” The systemal-lowedemployeesin his depart-ment to sharerumorseasily~the

managercouldthenquashfalseoneson-line.Thisexperimentworkedjust fine—until rumorsWerepostedaboutthemanager’sown departurefrom thebank.Whenhe refusedto comment,employeescorrectlysurmisedit wastrue.Theybecamecyni-cal.aboutthisattemptto shareinformationthroughtechnology,sincethemanagerhadn’tcommunicat-

edwith themon this particularpieceof information.Needlesstosay,Rumor Mill was~not contiri-tied by hissuccessor.

Sharingrumorsin this fashionunderscoresthe distinctionbe-tweeninformationandnoninfor-motion. Manypeoplesufferfromfar too muchnoninformation—

which companiesseemto gener-atewith easeandattheexpenseofuseful information—ratherthanthe “informationoverload”theycomplainabout.Any heavyB-mail usercantestify to thejunkmail problem.Right now I havemorethan160 messagesin myelectronicmailbox,someofwhichinform me that onecolleaguelost his appointmentbookorthatanotherwantedto be includedin lastThursday’spizza run. Ishouldneverhavereceivedthem,andnow I don’t-havt the timetodeletethem.

Technologistsareworking onpersonalizedfilters or “agents”thatcanseparaterealinformationfrom junk.Butit’s likely thatgoodmarketersof electronicinforma-tionwill find waysto circumventfilterst_~justasdirect mail nowlookslike ataxrefundorpersonalcheck.In fact,somecomniuiiica-tion technologiesjust exacerbatethisproblem.

At TandemComputers,for ex-ample,a combinationE-mail/bul-letin boardallows field-servicepersonnelto senda “hasanyoneseenthisproblem!”messagetoalltechnicalpeoplein the company.Theservicetechnicianmaygetananswer,but is it really necessaryfor everyoneto readthismessage!As Insomanyothercases,simplyimplementinganelectronic-mailsystem— without any guidelines

WhenChemicalBatikandManufacturersHanovermerged,

two informationculturesclashed.

HARVABD BUSINESSREVIEW Maxc~-Apd11994

INFORMATION MkNAGEMENT

I~A1, #‘ ~_I__ fi l~S ‘C~ OT anTormc2tlon

.~rkEflVh~~~~M~ N

~E)

stem thOr provides

toots,andto increase

Ii~kLr1~e~ecppkotions~hefollowing catogories:

inForm0tb0~~

0~dto°~

0nd toolS

and tools

~0~iotl/suPpor, tools

1oetno~i0fl

~,FHONE repre5~tmajor

,0orkel inf0~0~~cnd ore~ det~dj~iterin this section:

M~’rkChfl9PerspectiveICQM$1

DocismentDatabCs~(MDO~on Center (NSC)

Dccumdt~tpa~obo5e(PDOC)Accout~Support(SIA$tjpp1

line:

WHATt$0

pphCOti0~1name

0fferifl95

— Key HONEREF.~1GU1DEfr0m~

HO1’~E~~ifl menu.— on4inedem0n5~0tb0t~5

0F s.etected 0~pheotions~r funcb~On theMO~Emain menu.

AnuW1~eroI~Oflt0tn

User Guides, whichmaY be viewed bykeying GUIDE onoppflc0~’0’~menU.

EasternTime

128 HARVARDBUSINESSREVIEW Msrch.April 1994

INFORMATION MANAGEMENT

were afraid that answeringquestionsabout infor-mationwould be too time-consuming.In practice,however,theextratime involved hasn’t really in-terferedwith anyone’sjob. Manyoftheseinforma-tion ownersnowsaythey learnfrom thequestionsandcommentsof others.More important,~M hassavedmillions by avoidingduplicationIn thepur-chaseof externalmarketinformation.

in formationGuides.Along withmaps,informa-tion usersneedpeopleto guide themto therightkind of informationin the first place.Librarianshaveoftenperformedthis role in the past.Butwhile informationownersat IBM cananswerspe-cific questions,few companieshavegeneralguidesto the vastinformation resourcesavailablethroughoutanorganization.Onceagain,includingnewkinds of humansupport for technologycanhelpchangeacompany’sinformationcuhu.re.

In1991,HallmarkCards’sMIS managersrealizedthat the company’s informationuserswerecon-fusedabouthowtoaccessnecessarydata.Theprob-lem wasboth technicalandbehavioral,Financial,customer,supplier,product,and otherdatawereburiedin manydifferentdatabases.In addition,ex-istingapplicationswerehardtouseandprovidednoinformationabouthowthedatawerecreated.

Hailtuark’sMIS managersthereforeestablishedin eachbusinessunit a newfull-time position: the“information guide.”Theseindivid-uals are theprimary point of con-tactfor anyoneat Hallmark seekingcomputer-basedinformation.TheytranslatebetweenuserinformationrequestsandtheIT staff who canquerydatabasesandgetthecomput-erizedinformationthat usersneed.Hallmark’s inforniation guideshavehelpedimprovedataaccesssomuchthat therearenow 10 guidesaroundthecompany.Theyhavesubstantiallyreducedthetime it takesfor employeesto find the right infor-mation andto compareinformationacrossbusi-nessunits.

BusinessDocuments.Theform in which infor-mationis presentedis also critical to its under-standinganduse.Alterall, rawdataisnot inforrna-tion; and accumulatingdatais not the sameasinterpretingit andputting it in a usableform.Com-panyB’s emphasison documentationandpresenta-tion,demonstrateshowsuchanattitudeshapestheoverall informationculture. In that case,promo-tionsandotherfinancialincentivesweretiedtothekindsof documentsprofessionalsproduced.

Ingeneral,businessdocumentsprovideorganiza-tionandcontext,andtheyexcludeenoughinforma-

tionsothatwhatremainsisdigestible.Focusingonwhichdocumentsanorganizationnee~lsoftenleadsto amorefruitful discussionthanlookingatbroadinformationrequirementsor trying to pin downatermlike “customer.”

Severalcompanieshavebeguntoidentify criticalinformationneedsin theform of documents..AiDeanWitter, for instance,informationmanagers,particularlythosein thecentrallibrary, werefrus-tratedby their inability to addressbrokers’in.for-marionneedsefficiently. They advocatedhiringmorelibrarians,but financialexecutiveswerexc-luctantto takeon additionalworkers.

With thehelpof a consultant,financemanagerstalkedto brokersaboutwhat informationtheyneeded.Insteadofphrasingtheirquestionsin termsof informationandsystems,theyaskedwhichkeydocumentsbrokersrequired.As it turnedout, al-mostall usedthesamedocumentsover andover.Theirneedswere categorizedinto asetof “coredocuments,”mostofwhichwereregulatoryandre-portingdocumentsfrom U.S.companies.

By separatingthe documentsinto threeor fourIndustrygroups,90%ofthe informationneededbya typical brokerfit on oneCD-ROM disk. DeanWitter thencreateda“perfectinformationplatter,”which wasupdatedmonthly and kepton a localareanetworkserver.By definingcommoninforma-

Hallmark hasestablished

usersand the IT staff.

tionalneedsandimplementingtechnologyto sup-portwhatbrokerswerealreadydoing,DeanWitterwasableto reduceits library staff—ratherthanin-creasingit asoriginallysuggested—whilegreatlyLa..cilitatinginformationuse.

Groupware.Groupwarelike LotusNotes,NCR’SCooperation,andDigital Equipment’sTeam.Linksare excellentexamplesof lessstructuredinforma-tion-sharingtechnologies.Thisnewtechnologyal-lows teamsin different locationsto sharedocu-mentselectronically,to discussissueson-line,andto captureanddistributekeyinformationeasily.

Even so,companieswill fail totakeadvantageofgroupwareif theydon’t alsoprovideadequatetrain-ing andhumansupport.Indeed,groupwareimple-mentationstandsor falls on a company’sinforma-

“information guides”—

translators betweeninformation

130 HARVARDBUS~S5REVIEW Maxch-Ap~i11994