关闭

关闭

关闭

封号提示

内容

首页 Making Advanced Analytics Work for You - Harvard…

Making Advanced Analytics Work for You - Harvard Business Review.pdf

Making Advanced Analytics Work f…

still_Crazy_after_all 2013-09-20 评分 0 浏览量 0 0 0 0 暂无简介 简介 举报

简介:本文档为《Making Advanced Analytics Work for You - Harvard Business Reviewpdf》,可适用于高等教育领域,主题内容包含MakingAdvancedAnalyticsWorkforYoubyDominicBartonandDavidCourtArtwork:Tamar符等。

MakingAdvancedAnalyticsWorkforYoubyDominicBartonandDavidCourtArtwork:TamarCohen,TheBigQuick,,silkscreencollageonvintagebookpages,"x"BigdataandanalyticshaverocketedtothetopofthecorporateagendaExecutiveslookwithadmirationathowGoogle,Amazon,andothershaveeclipsedcompetitorswithpowerfulnewbusinessmodelsthatderivefromanabilitytoexploitdataTheyalsoseethatbigdataisattractingseriousinvestmentfromtechnologyleaderssuchasIBMandHewlettPackardMeanwhile,thetideofprivateequityandventurecapitalinvestmentsinbigdatacontinuestoswellThetrendisgeneratingplentyofhype,butwebelievethatseniorleadersarerighttopayattentionBigdatacouldtransformthewaycompaniesdobusiness,deliveringthekindofperformancegainslastseeninthes,whenorganizationsredesignedtheircoreprocessesAsdatadrivenstrategiestakehold,theywillbecomeanincreasinglyimportantpointofcompetitivedifferentiationAccordingtoresearchbyAndrewMcAfeeandErikBrynjolfsson,ofMIT,companiesthatinjectbigdataandanalyticsintotheiroperationsshowproductivityratesandprofitabilitythataretohigherthanthoseoftheirpeers(see“BigData:TheManagementRevolution”inthisissue)Evenso,ourexperiencerevealsthatmostcompaniesareunsurehowtoproceedLeadersareunderstandablyleeryofmakingsubstantialinvestmentsinbigdataandadvancedanalyticsThey’reconvincedthattheirorganizationssimplyaren’treadyAfterall,companiesmaynotfullyunderstandthedatatheyalreadyhave,orperhapsthey’velostpilesofmoneyondatawarehousingprogramsthatnevermeshedwithbusinessprocesses,ormaybetheircurrentanalyticsprogramsaretoocomplicatedordon’tyieldinsightsthatcanbeputtouseOralloftheaboveNowonderskepticismaboundsManyCEOs,too,recalltheirexperienceswithcustomerrelationshipmanagementinthemids,whennewCRMsoftwareproductsoftenpromptedgreatenthusiasmExpertsdescendedonboardroomspromisingimpressiveresultsifnewITsystemswerebuilttocollectmassiveamountsofcustomerdataItdidn’tturnoutthatwayToomanyCsuiteswereblindtothepracticalimplicationsofnewCRMtechnologiesnamely,thattocapitalizeonthem,organizationswouldhavetomakecomplexprocesschangesandbuildemployees’skillsThepromisedgainsinperformancewereoftenslowincoming,becausethesystemsremainedstubbornlydisconnectedfromhowcompaniesandfrontlinemanagersactuallymadedecisions,andnewdemandsfordatamanagementaddedcomplexitytooperationsTobefair,mostcompanieseventuallymanagedtogettheirCRMprogramsontrack,butnotbeforesomehadsufferedsizablelossesandseveralCRMchampionshadlostcareermomentumGiventhishistory,weempathizewithexecutiveswhoarecautiousaboutbigdataNevertheless,webelievethatthetimehascometodefineapragmaticapproachtobigdataandadvancedanalyticsonetightlyfocusedonhowtousethedatatomakebetterdecisionsInourworkwithdozensofcompaniesinsixdatarichindustries,wehavefoundthatfullyexploitingdataandanalyticsMakingAdvancedAnalyticsWorkforYouHarvardBusinessReviewhttp:hbrorgmakingadvancedanalyticsworkforyouarprof:PMtcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlightrequiresthreemutuallysupportivecapabilities(Seetheexhibit“HowtoBenefitfromBigData”)First,companiesmustbeabletoidentify,combine,andmanagemultiplesourcesofdataSecond,theyneedthecapabilitytobuildadvancedanalyticsmodelsforpredictingandoptimizingoutcomesThird,andmostcritical,managementmustpossessthemuscletotransformtheorganizationsothatthedataandmodelsactuallyyieldbetterdecisionsTwoimportantfeaturesunderpinthoseactivities:aclearstrategyforhowtousedataandanalyticstocompete,anddeploymentoftherighttechnologyarchitectureandcapabilitiesHowtoBenefitfromBigDataToimproveperformancewithadvancedanalytics,companiesneedtodevelopstrengthsinthreeareasMultipleDataSourcesCreativelysourceinternalandexternaldataUpgradeITarchitectureandinfrastructureforeasymergingofdataPredictionandOptimizationModelsFocusonthebiggestdriversofperformanceBuildmodelsthatbalancecomplexitywitheaseofuseOrganizationalTransformationCreatesimple,understandabletoolsforpeopleonthefrontlinesUpdateprocessesanddevelopcapabilitiestoenabletooluseEquallyimportant,thedesiredbusinessimpactmustdriveanintegratedapproachtodatasourcing,modelbuilding,andorganizationaltransformationThat’showyouavoidthecommontrapofstartingwiththedataandsimplyaskingwhatitcandoforyouLeadersshouldinvestsufficienttimeandenergyinaligningmanagersacrosstheorganizationinsupportofthemissionChoosetheRightDataTheuniverseofdataandmodelinghaschangedvastlyoverthepastfewyearsThesheervolumeofinformation,particularlyfromnewsourcessuchassocialmediaandmachinesensors,isgrowingrapidlyTheopportunitytoexpandinsightsbycombiningdataisalsoaccelerating,asmorepowerful,lesscostlysoftwareaboundsandinformationcanbeaccessedfromalmostanywhereatanytimeBiggerandbetterdatagivecompaniesbothmorepanoramicandmoregranularviewsoftheirbusinessenvironmentTheabilitytoseewhatwaspreviouslyinvisibleimprovesoperations,customerexperiences,andstrategyButmasteringthatenvironmentmeansuppingyourgame,findingdeliberateandcreativewaystoidentifyusabledatayoualreadyhave,andexploringsurprisingsourcesofinformationSourcedatacreativelyOftencompaniesalreadyhavethedatatheyneedtotacklebusinessproblems,butmanagerssimplydon’tknowhowtheinformationcanbeusedforkeydecisionsOperationsexecutives,forinstance,mightnotgraspthepotentialvalueofthedailyorhourlyfactoryandcustomerservicedatatheypossessCompaniescanimpelamorecomprehensivelookatinformationsourcesbybeingspecificaboutbusinessproblemstheywanttosolveoropportunitiestheyhopetoexploitForexample,abankingteamthatneededtoimprovetheefficiencyofitscustomerserviceoperationscreatedadegreeviewbycombininginformationfromATMtransactions,onlinequeries,customercomplaints,andsoonThatallowedduplicativeinteractionstobeidentified,therebyreducingcostsandstreamliningthecustomerexperienceManagersalsoneedtogetcreativeaboutthepotentialofexternalandnewsourcesofdataSocialmediaaregeneratingterabytesofnontraditional,unstructureddataintheformofconversations,photos,andvideoAddtothatthestreamsofdataflowinginfromsensors,monitoringprocesses,andexternalsourcesthatrangefromlocaldemographicstoweatherforecastsOnewaytopromptbroaderthinkingaboutpotentialdataistoask,“Whatdecisionscouldwemakeifwehadalltheinformationweneed”Usingthatlogic,oneshippingcompanyimprovedtheontimeperformanceofitsfleetbytappingspecializedweatherforecastdataandliveinformationaboutportavailabilitythatithadn’trealizedwereavailableSeniorexecutivescantaketheleadhereTheCEOofonemajorpackagedgoodscompanytoldusthatheviewsdataasastrategicassetwhosevaluehetakesintoaccountwhenassessingpotentialacquisitionsButleadersatalllevelsmustalsoMakingAdvancedAnalyticsWorkforYouHarvardBusinessReviewhttp:hbrorgmakingadvancedanalyticsworkforyouarprof:PMtcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlightbeattunedtonovelapproachestogatheringandhusbandinginformationAsbusinesspracticesintheinterneteracontinuetoevolve,inspirationcanoftenarisefromascanoftheexternalenvironmentAcorporatefinanceexecutive,forinstance,mightlooktoacompanysuchasKabbage,astartupthatsuppliesworkingcapitaltoonlinebusinessesToslashthetimerequiredtounderwriteloans,Kabbageasksmerchantstooptintosharingtheircustomerfeedbackratings,Facebookinteractions,andelectronicshippingrecordsThosewiththestrongestfeedbackandhighestbusinessvolumereceivegreaterfinancingGetthenecessaryITsupportLegacyITstructuresmayhindernewtypesofdatasourcing,storage,andanalysisExistingITarchitecturemaypreventtheintegrationofsiloedinformation,andmanagingunstructureddataoftenremainsbeyondtraditionalITcapabilitiesManylegacysystemswerebuilttodeliverdatainbatches,sotheycan’tfurnishcontinuousflowsofinformationforrealtimedecisionsFullyresolvingtheseissuesoftentakesyearsHowever,businessleaderscanaddressshorttermbigdataneedsbyworkingwithCIOstoprioritizerequirementsThismeansquicklyidentifyingandconnectingthemostimportantdataforuseinanalytics,followedbyacleanupoperationtosynchronizeandmergeoverlappingdataandthentoworkaroundmissinginformationSuchshorttermtacticsmayleadcompaniestovendorsthatfocusonanalyticsservicesoremergingsoftwareNewcloudbasedtechnologiesmayalsoofferwaystoscalecomputingpowerupordowntomeetbigdatademandscosteffectivelyTogetherthoseapproachesestablishanITinfrastructurethatpropelsinnovationbyfacilitatingcollaboration,rapidanalysis,andexperimentationBuildModelsThatPredictandOptimizeBusinessOutcomesDataareessential,butperformanceimprovementsandcompetitiveadvantagearisefromanalyticsmodelsthatallowmanagerstopredictandoptimizeoutcomesMoreimportant,themosteffectiveapproachtobuildingamodelrarelystartswiththedatainsteaditoriginateswithidentifyingthebusinessopportunityanddetermininghowthemodelcanimproveperformanceUnfortunately,notallmodelbuildingfollowsthiscourseOneapproachthatgetsinconsistentresults,forinstance,issimpledataminingCorrallinghugedatasetsallowscompaniestorundozensofstatisticalteststoidentifysubmergedpatterns,butthatprovideslittlebenefitifmanagerscan’teffectivelyusethecorrelationstoenhancebusinessperformanceApuredataminingapproachoftenleadstoanendlesssearchforwhatthedatareallysayOnecompanyfollowedamoretargetedstrategytooptimizecomplexproductpricingAtitscorewasamodelbasedonthehistoricalpriceelasticityofitsproducts,salesdata,competitors’responses,andothervariablesToimproveitschancesofsuccess,thecompanybeganthemodelingprocessbypositingwhichfactorsaffectedsalesvolumes(forinstance,competitors’pricingandpromotions)andthenaskedwhatdataandwhichmodelwouldbestdeliverinsightsthatwereusefulformakingbusinessdecisionsWehavefoundthatsuchhypothesisledmodelinggeneratesfasteroutcomesandalsorootsmodelsinpracticaldatarelationshipsthataremorebroadlyunderstoodbymanagersRemember,too,thatanymodelingexercisehasinherentriskAlthoughadvancedstatisticalmethodsindisputablymakeforbettermodels,statisticsexpertssometimesdesignmodelsthataretoocomplextobepracticalForexample,apredictivemodelwithvariablesmayexplainhistoricaldatawithhighaccuracy,butmanagingsomanyvariableswillexhaustmostorganizations’capabilitiesCompaniesshouldrepeatedlyask,“What’stheleastcomplexmodelthatwouldimproveourperformance”TransformYourCompany’sCapabilitiesTheleadconcernexpressedtousbyseniorexecutivesisthattheirmanagersdon’tunderstandortrustbigdata–basedmodelsOnelargeretailerintendeditsmodeltooptimizereturnsonadvertisingspending,butdespiteconsiderableinvestment,itwasn’tbeingusedThereasonsoonbecameevident:Thefrontlinemarketerswhomadekeydecisionsonadspendingdidn’tbelievethemodel’sresultsandhadlittlefamiliaritywithhowitworkedManycompaniesgrapplewithsuchproblems,oftenbecauseofamismatchbetweentheorganization’sexistingcultureandcapabilitiesandtheemergingtacticstoexploitanalyticssuccessfullyInshort,thenewapproachesdon’talignwithhowcompaniesactuallyarriveatdecisions,ortheyfailtoprovideaclearblueprintforrealizingbusinessgoalsToolsseemtobedesignedforexpertsinmodelingratherthanforpeopleonthefrontlines,andfewmanagersfindthemodelsengagingenoughtochampiontheiruseakeyfailingifcompanieswantthenewmethodstopermeatetheorganizationBottomline:Usingbigdatarequiresthoughtfulorganizationalchange,andthreeareasofactioncangetyouthereDevelopbusinessrelevantanalyticsthatcanbeputtouseLikeearlyCRMmisadventures,manyinitialimplementationsofbigMakingAdvancedAnalyticsWorkforYouHarvardBusinessReviewhttp:hbrorgmakingadvancedanalyticsworkforyouarprof:PMtcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlightdataandanalyticsfailsimplybecausetheyaren’tinsyncwiththecompany’sdaytodayprocessesanddecisionmakingnormsTheaforementionedcaseofacompanythataimedtooptimizepricesillustrateshowtoavoidthosecommonpainpointsThecompanystartedwithananalyticstaskforcethatconvenedaseriesofmeetingswithpricingandpromotionsmanagerstobetterunderstandthetypesofdecisionstheymadewhensettingpricesandhowthosechoicesultimatelyaffectedrevenueandcustomerretentionModeldesignersalsoinquiredaboutthetypesofbusinessjudgmentsthatmanagersmaketoaligntheiractionswithbroadercompanygoalsTheseconversationsensuredthatbothpricinganalyticsandresultingscenariotoolswouldcomplementexistingdecisionprocessesThemodelingallowedthecompanytoreachitsultimategoal:moreeffectivemanagementofpriceandvolumetradeoffsasproductlaunchesproliferatedEmbedanalyticsintosimpletoolsforthefrontlinesManagersneedtransparentmethodsforusingthenewmodelsandalgorithmsonadailybasisBynecessity,terabytesofdataandsophisticatedmodelingarerequiredtosharpenmarketing,riskmanagement,andoperationsThekeyistoseparatethestatisticsexpertsandsoftwaredevelopersfromthemanagerswhousethedatadriveninsightsOnelargeindustrialcompany,forinstance,soughttobetterforecastworkforceneedstoreflectlocalmarketvariationsHistorically,asthecompanyhadtriedtokeeplaborcostslow,ithadoftenfounditselfshortstaffedinsomemarkets,leadingtosignificantovertimecostsandservicesnafusToremedytheproblem,thecompanyconvenedasmallworkinggroupofanalystsandITprogrammerswhodevelopedaseriesofpredictivemodelsthatforecastworkforceavailabilityonthebasisoffactorssuchasvacationtime,absenteeism,andworkrulesinlaborcontractsThemodelsincorporatedmillionsofnewdatapointsonthousandsofemployeesacrossdozensoflocationsButratherthanprovidingmanagerswithreamsofdataandcomplexmodels,theycreatedasimplevisualinterfacethathighlightedprojectedworkforceneedsandnecessaryactionsUltimately,thatapproachofusingasimpletooltodelivercomplexanalyticssubstantiallyimprovedworkforceplanningandreducedtheneedfornewhiresandovertimeDevelopcapabilitiestoexploitbigdataEvenwithsimpleandusablemodels,mostorganizationswillneedtoupgradetheiranalyticalskillsandliteracyManagersmustcometoviewanalyticsascentraltosolvingproblemsandidentifyingopportunitiestomakeitpartofthefabricofdailyoperationsEffortswillvarydependingonacompany’sgoalsanddesiredtimelineAdultlearnersoftenbenefitfroma“fieldandforum”approach,wherebytheyparticipateinrealworld,analyticsbasedworkplacedecisionsthatallowthemtolearnbydoingAtoneindustrialservicescompany,themissionwastogetbasicanalyticstoolsintothehandsofitsroughlysalesmanagersTrainingbeganwithaninfieldassignmenttoreadabriefdocumentandcollectbasicfactsaboutthemarketNextmanagersmetincentralized,collaborativetrainingsessionsduringwhichtheyfiguredouthowtousethetoolsandmarketfactstoimprovesalesperformanceTheythenreturnedtothefieldtoapplywhattheyhadlearnedand,severalweekslater,reconvenedtoreviewprogress,receivecoaching,andlearnaboutsecondorderanalysisoftheirdataThisprocessenabledafourpersonteamtoeventuallybuildcapabilitiesacrosstheentiresalesmanagementorganizationAdjustingcultureandmindsetstypicallyrequiresamultifacetedapproachthatincludestraining,rolemodelingbyleaders,andincentivesandmetricstoreinforcebehaviorOnelargeconsumerproductscompanyappliedsuchanapproachsuccessfullyItcreatedasophisticatedprogramtoimprovetheprofitabilityofpromotionalspendingwithitsretailersThelaunchincludedtrainingledbycompanymanagementandanewpromotionsanalysistoolforsalesrepresentativesHowever,afteraninitialwhirlwindofactivity,theprogramanduseofthetoolfizzledTheobstaclewasthatcompanyincentivesandreportingprotocolsforsalesmanagerstrackedsalesandsalesgrowth,notprofitsAsaresult,themanagersconsideredtheprofitfocusedprogramtobebureaucraticoverheadthatwasunrelatedtotheirkeysalesgoalsAfteraseriesofdiscussionswiththemanagers,thecompanyrelaunchedtheprogram,offerednewincentivesforimprovingprofits,andtailoredreportstoprofitrelateddataAlthoughongoingtrainingandcoachingwasnecessary,theeffortsgraduallyproducedashiftinmindsetsuchthatthepowerofpromotionsanalyticsisnowusedtofurtherthecommongoalofincreasingprofitabilityTheeraofbigdataisevolvingrapidly,andourexperiencesuggeststhatmostcompaniesshouldactnowButratherthanundertakingmassiveoverhaulsoftheircompanies,executivesshouldconcentrateontargetedeffortstosourcedata,buildmodels,andtransformtheorganizationalcultureSucheffortswillplayapartinmaintainingflexibilityThatnimblenessisessential,giventhattheinformationitselfalongwiththetechnologyformanagingandanalyzingitwillcontinuetogrowandchange,yieldingaconstantstreamofopportunitiesAsmorecompanieslearnthecoreskillsofusingbigdata,buildingsuperiorcapabilitiesmaysoonbecomeadecisivecompetitiveassetMakingAdvancedAnalyticsWorkforYouHarvardBusinessReviewhttp:hbrorgmakingadvancedanalyticsworkforyouarprof:PMtcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlighttcoxHighlight

用户评论(0)

0/200

精彩专题

上传我的资料

每篇奖励 +1积分

资料评分:

/4
1下载券 下载 加入VIP, 送下载券

意见
反馈

立即扫码关注

爱问共享资料微信公众号

返回
顶部

举报
资料