戰(zhàn)略測(cè)量的未來：用人工智能增強(qiáng)關(guān)鍵績(jī)效指標(biāo)+The+Future+of+Strategic+Measurement：Enhancing+KPIs+With+AI

Incollaborationwith

BIGIDEAS

RESEARCHREPORT

MTsloan

ManagementReview

February2024

TheFuture

ofStrategic

Measurement:

Enhancing

KPIsWithAI

byMichaelSchrage,DavidKiron,Fran?oisCandelon,

ShervinKhodabandeh,andMichaelChu

AUTHORS

MichaelSchrageisaresearchfellowwiththeMITSloanSchoolofManagement’sInitiativeontheDigitalEconomy.Hisresearch,writing,andadvisoryworkfocusesonthebehavioraleconomicsofdigitalmedia,models,and

metricsasstrategicresourcesformanaginginnovationopportunityandrisk.

DavidKironistheeditorialdirector,research,ofMITSloanManagementReviewandprogramleadforits

BigIdeasresearchinitiatives

.

Fran?oisCandelonisaseniorpartnerand

managingdirectoratBostonConsulting

Group(BCG)andtheglobaldirectorofthe

BCGHendersonInstitute,wherehisresearchfocusesontheimpactoftechnologieson

businessandsociety.Hecanbecontactedat

candelon.francois@

.

ShervinKhodabandehisaseniorpartnerandmanagingdirectoratBCGandthecoleaderofBCG’sAIbusinessinNorthAmerica.HeisaleaderinBCGXandhasover20years

ofexperiencedrivingbusinessimpactfromAIanddigital.Hecanbecontactedat

shervin@

.

MichaelChuisapartnerandassociatedirectoratBCG,wherehefocusesonapplyingAIandmachinelearningtobusinessproblemsin

commercialfunctions,includingoptimizingpricing,promotions,sales,andmarketing.Hecanbereachedat

chu.michael@

.

CONTRIBUTORS

KristinBurnham,KevinFoley,CarolynAnnGeason-Beissel,WanmingHuang,GauravJha,LisaKrayer,MicheleLeeDeFilippo,AllisonRyder,BarbaraSpindel,andDavidZuluagaMartínez

TheresearchandanalysisforthisreportwasconductedunderthedirectionoftheauthorsaspartofanMITSloanManagementReviewresearchinitiativeincollaborationwithandsponsoredbyBostonConsultingGroup.

Tocitethisreport,pleaseuse:

M.Schrage,D.Kiron,F.Candelon,S.Khodabandeh,andM.Chu,“TheFutureofStrategic

Measurement:EnhancingKPIsWithAI,”MITSloanManagementReviewandBostonConsultingGroup,February2024.

SUPPORTINGSPONSORS

BCG

HENDERSON

INSTITUTE

BCG》《

Copyright?MassachusettsInstituteofTechnology,2024.Allrightsreserved.REPRINT#:65370

CONTENTS

1

Introduction

4

DevelopingSmarterKPIs

10

TransformingLegacyKPIPractices

1

1

Takeaways

1

3

Conclusion

1

4

Appendix:UsingChatGPTtoCraftNovelKPIs

ABOUTTHERESEARCH

ThisreportpresentsfindingsfromtheseventhannualglobalresearchstudyonartificialintelligenceandbusinessstrategybyMITSloanManagementReviewandBostonConsultingGroup.Inspring2023,wefieldedaglobalsurveyandsubsequentlyanalyzedrecordsfrom3,043respondentsrepresentingmorethan25indus-triesand100countries.Wealsointerviewed17executivesleadingAIinitiativesinabroadrangeofcompaniesandindustries,includ-ingfinancialservices,mediaandentertainment,retail,travelandtransportation,andlifesciences.

OurresearchexamineshowmanagersandleadersuseAItoenhancestrategicmeasurementtoadvancestrategicoutcomes.ItexploreshoworganizationshaveadaptedKPIs—andevengeneratednewones—withAItodefineanddelivermeasurablybetterperformance.

Introduction

Legacykeyperformanceindicators(KPIs)increasinglyfailtodelivertheinformationandinsightsleadersneedtosucceed.Theyfallshortintrackingprogress,aligningpeo-pleandprocesses,prioritizingresources,andadvancingaccountability.Thesefailuresbothundermineoperationalefficienciesandcompromisethepursuitofstrategicobjec-tivesandoutcomes.Sophisticatedorganizationsworld-widerecognizethattheirKPIsneedtobemeasurablysmarterandmorecapable.Theyconsequentlyinvestinalgorithmicinnovationstomaketheirperformancemet-ricsmoreintelligent,adaptive,andpredictive.SmartKPIspoweredbyartificialintelligence(AI)becomesources—notmerelymeasures—ofstrategicdifferentiationandvaluecreation.

Basedonaglobalsurveyofmorethan3,000managersandinterviewswith17executives,wefindAIbeingusedtofun-damentallyredefineperformanceaswellasenhanceit.Weseeorganizationsusingalgorithmstochallengeandimproveenterpriseassumptionsaroundperformance,profitability,andgrowth.CompaniesthatrevisetheirKPIswithAIarethreetimesmorelikelytoseegreaterfinancialbenefitthanthosethatdonot.SmarterKPIsleadtobetteroutcomes.

OnlinefurnitureretailerWayfair,forexample,usedAItoreexaminethefundamentalsbehinditslost-salesKPI.“Weusedtothinkthatifyoulostthesaleonaparticularprod-uct,likeasofa,itwasalosstothecompany,”saysCTOFionaTan.“Butwestartedlookingatthedataandrealizedthat50%to60%ofthetime,whenwelostasale,itwasbecausethecustomerboughtsomethingelseinthesameproductcategory.”

ThisinsightledWayfairtoreengineeritslost-salesKPIintoamorevaluablemetric.Whereasthecompanyprevi-ouslycalculateditem-basedlostsalesinresponsetopricechanges,itnowalsocalculatescategory-basedretentionofsalesinresponsetopricechanges.Withthenew,moreaccurateKPI,Wayfairisabletomakemoreeffectivefur-niturerecommendationsthatincorporatecustomerpref-erences—frompricepointstoshipmenttimes—whensuggestingnext-bestoffers.Logistically,theoperationsteamalignedproductplacementdecisionswithdistribu-tioncenterandwarehouseconstraintstoimprovebothcustomerandemployeeexperiences.AsmarterKPImea-surablyimprovedoutcomesforeverybody.

Thisexampleisnooutlier.Ourresearchfindsorganiza-tionsacrossindustriesusingAItorevisittheirKPIfunda-mentalsanddiscoverlatentorundervaluedperformancefeatures.Byidentifying,redefining,andcommunicatingsmartermetrics—theirstrategicmeasurementsystem—leadersincreasetheircompany’sabilitytooptimizedesiredoutcomes.1

1

TheFutureofStrategicMeasurement:EnhancingKPIsWithAI

SmartKPIscandomorethanjustunearthsourcesofvalueanddifferentiationthatwouldotherwisegoundetectedorunderexploited.Theycanalsopreventtheundesirableoutcomesthatcanresultfromafailuretoregularlyreex-amineassumptionsunderlyinglegacyKPIs.The2008globaleconomiccrisis,forexample,wastriggeredinpartbybanks’dependenceonathenwidelyusedmetric,valueatrisk,whichmeasurespotentialportfoliolossesinnor-malmarketconditionsatasinglepointintime.Financialinstitutionsdidnotadjustthismeasureasriskiersubprimemortgagesandcreditdefaultswapsbecamealargerpartoftheirportfolios.Guidedbyametricthatseverelyunder-estimatedpotentiallosses—insomecases,byordersofmagnitude—manyfinancialinstitutionswentbankruptorsufferedsignificantlosses.

ExecutiveswhorecognizethepowerandpotentialofAI-enabledKPIsacknowledgethattheircurrentKPIdesignandreviewpracticesareanachronistic.Moreandmorecompaniesarerisingtothischallengebyembed-dinggovernanceprocessesandfunctionstooverseeKPIdevelopment.KPIgovernancemechanismsthatmaintaindetailed,accurate,andrelevantperformancedataovertimeensurethatKPIqualityisexplicitlyconsideredduringtheirdesignandnotanafterthought.SchneiderElectric,forexample,ismakingsignificantfinancialandnonfinan-cialinvestmentsinlearninghowtoimprovenotonlyper-formanceagainstestablishedmetricsbuttheperformancemetricsthemselves.AccountabilityforperformanceonKPIsisincreasinglyinsufficient;companiesneedaccount-abilityfortheperformanceofKPIs,too.

Ourglobalsurveyrevealsthat60%ofmanagersbelievethattheyneedtoimprovetheirKPIs,butonlyone-third(34%)areusingAItocreatenewKPIs.Strikingly,9outof10managerswithinthat34%agreeorstronglyagreethattheirKPIshavebeenimprovedbyAI.(seefigures1and2,page3.)Whileitisstillearlydaysformostorganizations,thebenefitsofenhancingKPIswithAIarebecomingclearertomoreandmorecompanies.

Thefar-reachingimplicationsofthesefindingsaredifficulttooverstate.CompaniesthatalgorithmicallyimprovetheirKPIsarereconsideringthepurposeofperformancemea-surement,thefunctionofKPIs,andthestrategicvalueofmetrics.Ourresearchdemonstratesthatthesereconsid-erationsfrequentlyleadtonewprocessesfortransformingbothKPIdesignandhoworganizationalbehaviorsalignwithstrategicoutcomes.Strengtheningstrategicalign-mentisbothanobjectiveandanoutcomeofsmartKPIs.

Thesealgorithmicimperativesputaprovocativetwistontheoft-quotedphrase“whatgetsmeasuredgetsman-aged.”ThankstoAI,whatorganizationslearntomeasure,theymustalsolearntomanage—andhowtheymeasuremattersasmuchaswhattheymeasure.Asbetterinstru-mentation,betterdata,andbetteralgorithmschangethemeasurementprocess,opportunitiesforstrategicdifferen-tiationandvaluecreationimproveaswell.

ThisreportsynthesizesandexpandsuponkeyfindingsfromthefourarticlesinouryearlongcollaborativeresearchseriesonenhancingKPIswithAI.2WediscussanascentbutgrowingefforttocreatesmarterandmorevaluableKPIsandofferevidence-basedmanagementtakeawaysfordevelopingandusingnext-generationKPIstodrivebetterperformance.WealsodetailspecificstepsmanagerscantaketobeginrealizingthebenefitsofsmarterKPIs.

2

MITSLOANMANAGEMENTREVIEW?BCG

FIGURE1

ImprovingKPIsIsaLeadershipPriority

WhilemorethanhalfoftheleaderswesurveyedarefocusedonimprovingthequalityoftheirKPIs,onlyapproximatelyone-thirduseAItogeneratenewKPIs.

ofrespondentsassertthattheirorganizationsarestrivingto

improvethequalityoftheirKPIsfordecision-making

oforganizationsareusingAItocreatenewKPIs

60%

34%

FIGURE2

CreatingNewKPIsWithAI

Ofthe34%oforganizationssurveyedthatuseAIto

createnewKPIs,90%seeimprovements.

34%

oforganizationsuseAI

tocreatenewKPIs

AI-EnhancedKPIsPositivelyCorrelate

WithMoreBusinessBene?ts

UsingAINotUsingAI

E?ciency

54%

20%

Precisemeasurements

46%

12%

Timelymeasurements

43%

13%

Greater?nancialbene?t

33%

8%

90%

ofthoseorganizationsseetheirKPIsimprove

?AcommoncorporateobjectiveistoimproveKPIsovertime.OurresearchshowsthatusingAIisoneofthemosteffectivewaystoenhanceKPIs.

TheFutureofStrategicMeasurement:EnhancingKPIsWithAI

3

FIGURE3

ThreeTypesofSmartKPIs

EnrichedwithAI,

smartKPIsoffer

severaladvantagesovertraditionalKPIs.

DevelopingSmarterKPIs

OrganizationstypicallyuseKPIsasbenchmarkstoevalu-ateprogressonawiderangeofbusinessobjectives,suchassalesgrowth,customersatisfaction,andoperationalefficiency.Data-gatheringtechniquesandanalyticsaresub-sequentlyusedtomeasureprogressagainstthesemetrics.ThisapproachprioritizeswhatKPIsmeasureratherthanhowKPIsaremeasured.

Today,however,shiftingeconomicconditions,evolvingconsumerexpectations,anddigitaltransformationeffortscreateamandatefororganizationstoreconsider—onanongoingbasis—whatkeyperformancemeansandhowtomeasureit.Managersacknowledgetheneedforchange:Sixoutof10surveyrespondentsagreethatimprovingtheirKPIs(intermsofwhatgetsmeasuredandhowitgetsmeasured),notjustimprovingperformance,iscriticalforeffectivedecision-making.EveryexecutiveweinterviewedexplicitlydeclaredthatimprovingKPIswithAIisanenter-priseinterest,anditisanimmediatepriorityforseveral.

TherearemanypathstoimprovingKPIs.Low-costapproachesthatrelyonhumanjudgment,intuition,andexperiencecandeliverincrementalchangeswithincremen-talinsightsintoperformance.Incontrast,AI-enhancedKPIscandeliversignificantlymoredetailedandaccurateinsightsintocurrentandfutureperformance.Forcom-paniesdeployingsuch—smart—KPIs,theorganiza-tionalcostsareoutweighedbybusinessbenefitsspanningincreasedefficiency,betteralignment,andimprovedfinan-cialoutcomes.

ThreeTypesofSmartKPIs

WeseethreewaysthatAI-enrichedKPIsimproveonleg-acymetricsthatsimplytrackperformance.SmartKPIsbetterdescribeongoingandpastperformance,moreeffectivelyanticipatefutureperformance,andmakemoreusefulrecommendationstopromoteoutcomes.ThesethreetypesofsmartKPIs(descriptive,predictive,and

Example

Sanofi’s“snackableAI”toolboosts

situational

awarenessby

revealingcriticalinterdependenciesamongdifferentKPIs.

GeneralElectrichastransformeditsKPIstofocusonleadingindicatorsandis

usingAItoanalyzeorderpipelinesby

comparingorders

againsttheinstalledbaseofproductsandservices,helpingtoaccuratelyidentifyopportunitiesto

increasefutureordersanddrivestrongerrevenueandmargins.

Sanofi’ssmartKPIsalignoperations

andsalesby

recommending

adjustmentstosales

activitiesorprioritiesbasedonsupply

chainperformance.

Description

Synthesizehistoricalandcurrentdatatodeliverinsightsonwhathappenedorwhatishappening.

Anticipatefuture

performance,

producingreliable

leadingindicators.

Providevisibilityintopotentialoutcomes.Identifypatternsthatothertechniques

orhumanscannot,allowingthemto

drawonaricher

rangeofpotentiallycounterintuitive

patterns.

UseAItorecommendactionsthatoptimizeperformance.

Benefits

Provideadeeperunderstandingofperformancegapsandtheircauses.

Enablepreemptiveactionsinorder

tomitigaterisks

orleverage

opportunities.

Indicateperformancegapsandsuggest

correctivemeasures.

TypeofKPI

Smart

DescriptiveKPIs

Smart

PredictiveKPIs

Smart

PrescriptiveKPIs

4

MITSLOANMANAGEMENTREVIEW?BCG

prescriptive)maptoawell-knowndistinctionbetweendescriptive,predictive,andprescriptiveanalytics.

TheideaofprescriptiveKPIsmightbeunfamiliar,butexecutivedashboardsthatcolor-codeKPIsalreadypro-videasimplecalltoaction:Redindicatesthatperfor-manceisdown,andgreenmeansthatperformancemeetsorexceedsexpectations.SmartKPIsgofurther:Theycanmakemoredetailedrecommendationsaboutnextstepsleadersshouldtakeinresponsetothesesignalsanddiag-noseimplicationsforotherKPIs.(seefigure3,page4.)

ImprovingExistingKPIsWithAI

Goodhart’slawdeclaresthatwhenametricbecomesatarget,itceasestobeagoodmetric.Buttargetingmet-ricsthemselvesforimprovementisbothconsistentwithGoodhart’scautionandanessentialingredientforsus-tainableoperationalsuccess.ContinualKPIimprovementbecomesatargetinitsownright.

MediaconglomerateCBSoffersacompellingexampleofthissensibilityandapproach.Oneexecutivebeganexplor-ingwhetherAIcoulddiscernperformancedriversthatexecutivescouldn’tidentifythroughintuitionandexpe-riencealone.3Intheinterestofpredictingwhichprogramswouldbecomehits,theexecutivegaveherAIteams50yearsofKPIsandconsumerresearchdatatodeterminewhetherthecompanyhadtherightKPIsforpredictingthesuccessofTVpilots.

TheteamusedAItoconfirmthemeritsofexistingKPIsandidentifiedadditionalonesthathelpedrefinehowthecompanyassessespilots.“WegotbetterbygoingthroughthisAIexercise,”shenoted.“Theanalysischangedwhatwewerelookingforandhelpedimproveourperformance.”

Intentionallyusingalgorithmstorevisitandreviewkeyperformanceparametersleadstoimprovedperformance.ForTokopedia,oneofIndonesia’slargestmarketplaces,merchantreliabilityiscoretomaintainingstrongcus-tomerrelationships.Theorganizationsells1.8billionprod-uctsperdaythrough14millionmerchants—morethan85%ofwhicharenewentrepreneurs,saysCTOHermanWidjaja.Whilemerchantshavegoodproductstosell,notallofthemmanagetheirstockefficiently,whichinturnnegativelyimpactsfulfillment,customersatisfaction,and,ultimately,thevolumeoftransactionstakingplaceonthe

platform.Toaddressthisissue,thecompanybegansyn-thesizingandanalyzingmillionsofmerchantandcustomerdatapoints.

Itsalgorithmicanalysisofthedataledtoascoringsystemthatenhancedthecompany’sKPIformerchantqual-ity.Thesystemhasimprovedbothcustomerservice(byconnectingcustomerstothosemerchantsmostlikelytodeliveraqualityproduct)andmarketplaceeffective-ness(byidentifyingandhelpinglow-scoringmerchantsimprovetheirownoperations).Thescoringsystembenefitsallparties:thecustomers,themerchants,andTokopediaitself.Withit,thecompanyhasbeenabletocreateamorecredible,effectivemarketplace.

ENHANCINGKPISWITH

AIATMAERSK

Maersk,theDanishtransportation,shipping,andlogisticscompany,usedAItoreassessandredefinehowitmeasuresthroughputandtheproductivityofitsnetworkof65assetsinports,transportation,andwarehousesworldwide.Front-linemanagershadtodecidewhetherkeyperformancewasbestdefinedbyloadingandunloadingshipsortrucksasquicklyaspossibleorbymanagingtheloadingprocesssothatthetransportationcouldreliablydepartassched-uled.Toreachadecision,thecompanyusedAI-drivenmodelstorepresenteachapproachandassessitseffectsacrossthevaluechain.Thesemodelsconcludedthatreliabledeparturesthatusedlessloadingequipmentwouldpreemptbottlenecksbothattransshipmentpointsandduringconnectionsbetweendifferentmodesoftransport,suchasroadandrail.WithAI,MaerskprioritizedtherightKPI,overcomingahumanbiastomaximizespeed.Thisapproachledtomoreefficient,alignedperformanceacrosstheenterpriseand,inturn,increasedcustomersat-isfactionwithreliabledeliveries.

TheFutureofStrategicMeasurement:EnhancingKPIsWithAI

5

FIGURE4

BenefitsFrom

AI-AdjustedKPIs

OrganizationsthatcreatenewKPIs

usingAIaremore

likelytoseestrongeralignment,increasedcollaboration,moreaccurateforecasts,andmoreefficiencythanorganizationsthatdonotcreatenewKPIsusingAI.

CreatingNewKPIs

IdentifyingstrategicandoperationalKPIsisacommonfunctionofhumanleadershipteams.YetourresearchconfirmsthatAI—notonlyhumans—canproposeandevendesignnewandnovelKPIstoguideexecutivedecision-making.Weheardstoryafterstoryofleadersturningtoalgo-rithmstosourcekeyperformancemetricsthatarenotonlystrategicallyandoperationallyvaluablebut,insomecases,derivedfromdatapatternstoocomplicatedforhumanstodiscern.OursurveydataaffirmsthatcompaniesusingAItocreatenewKPIsseeabroadrangeofbusinessbenefitscom-paredwiththosecompaniesthatdon’tusethetechnologytocreatenewKPIs.(seefigure4.)OrganizationsusingAItocreatenewKPIs(34%ofallrespondents)aremorelikelytorealizebenefitsaroundalignment,collaboration,efficacy,financialbenefit,andefficiency.

TheAppendixsection

(seepage14)

describesanexperi-mentweconductedwithChatGPTthatresultedinseveralnovel,andpotentiallyuseful,KPIs.

AvinashKaushik,chiefstrategyofficeratdigitalmarketingagencyCroud,wasformerlytheseniordirectorofglobalstrategicanalyticsatGoogle,wheremachinelearninghelpedhisteamdiscoveranewmeasureofperformance.In

5×

4×

3×

3×

2×

Incentivestructureisinlinewithobjectives:5.13timesmorethanatorganizationsthatdonotuseAItocreatenewKPIs(6%?29%)

Increasedcollaborationbetweenemployees:3.97timesmorethanatorganizationsthatdonotuseAItocreatenewKPIs(11%?44%)

Moree?ectiveatpredictingfutureperformance:3.37timesmorethanatorganizationsthatdonotuseAItocreatenewKPIs(13%?45%)

Greater?nancialbene?t:

3.27timesmorethanatorganizationsthatdonotuseAItocreatenewKPIs(11%?34%)

Greatere?ciency:2.31timesmorethanatorganizationsthatdonotuseAItocreatenewKPIs(24%?56%)

onesituation,herecalls,thetechnologygianthadmadeasubstantialbutunderperformingmarketinginvestmentinaprimarydigitalchannel.“Wehadnoideawhatwaswrongwithus,”Kaushikrecalls.“Weknewwewerefailing;wejustdidn’tknowwhy,andwe’dexhaustedallthequestionswecouldask.”Google’swealthoftalent,analyticresources,anddataaccessweren’tenoughtocrackthecode.

Kaushik’steamdevelopedanalgorithmusingamachinelearningmodelcalledaclassificationtreetoidentifycon-nectionsandcorrelationstheyhadmissed.“Becausewedidn’tevenknowwhatquestionstoask,thiskindofunsu-pervisedmachinelearningalgorithmwasareallygoodapproach,”hesays.“Weletthealgorithmfindthepatterns.”

WhatthealgorithmfoundsurprisedKaushik’steam:KPIstheyhadthoughtwerethemostessentialtoopti-mizeweren’t.Themostinfluentialmetrics,theirorderofimportance,andtheiroptimalrangesofoperationwerearevelation.Asurprisingmeasurewasthepercentageofimpressionsinwhichapersonviewedandheardafullad.Ifthepercentagewasbelowacertainbenchmark,themarketingcampaignwasdoomedtofail;ifhigher,thecampaignhadachanceatsuccess.Notsixmonthsafter

6

MITSLOANMANAGEMENTREVIEW?BCG

implementingthealgorithm’srecommendations,desiredperformanceincreased30points—“aninsaneperfor-manceimprovement,”Kaushiksays.

KPIGOVERNANCEAT

SCHNEIDERELECTRIC

SchneiderElectric’schiefgovernanceofficerandsecretarygeneral,HervéCoureil,acknowledgesthatharnessingdatatounearthandgeneratenewKPIscanbecostlyandtime-consuming.But,hecontends,it’sstrategicallynecessary.“WewantourKPIstoevolveovertimebecausewedon’twanttodriveourbusinessonlegacyorvanitymetrics,”hesays.ToensurethatKPIsdevelopinaccordwithchangingobjectivesandnewmeasurementcapabilities,Coureilcospon-soredaperformancemanagementoffice(PMO)withinthecompany’sdatateamtooverseeper-formancestandards.“Welookedatperformancemeasurementasoneofourtransformationdriv-ers,”hesays.ThePMOhelpedtopmanagementevolveitsportfolioofKPIsandalignthemwiththecompany’svariousoperatingunits.Italsopushedthebusinessestobecomemoredata-drivenandanalyticallyawareofthelinksbetweenperformanceimprovements,improvedmetrics,andenhancedoutcomes.

Ourresearchfindsgrowingdemandfornew,AI-drivenKPIs.AtRegionHallandHealthSysteminSweden,forexample,researcherscreatedandtrainedanalgorithmbasedondatafromdeathcertificates,governmentrecords,andelectronicrecordstobetterpredictsuddencardiac

death(SCD)intheyearafteranelectrocardiogram(ECG)

isperformed.SCDafflicts300,000peopleintheU.S.,and

millionsglobally,everyyear.4

“Thealgorithmbecamequitegoodatpredictingwho’s

goingtosuccumbtosuddencardiacdeathintheyearafter

anECGistaken,”saysDr.ZiadObermeyer,aphysician

andprofessorattheUniversityofCalifornia,Berkeley,

whohelpedtrainthealgorithm.“Everytimesomeone

getsanECG,itgeneratesariskscorethatmeasuresthe

probabilitytheindividualwilldiefromsuddencardiac

arrest.”ThatECGscorecanfunctionlikeakeyperfor-

manceindicator:Withthescoreinhand,adoctormight

observehowapatient’sSCDriskprofilechangesafter

they’represcribedamedication,suchasabetablockeror

anACEinhibitor.

WhileusingAItodiscovereffectivepredictiveindica-

torsforSCDremainsaworkinprogress,researchersare

alreadyexcitedaboutitspotentialtocreateearlyindicators

forotherdiseases.“Havingtheabilitytoturnthesevery

complicatedbiologicalsignalsintoindicatorsisverypow-

erful,”O(jiān)bermeyersays.“Ithinkthat’sgoingtobesome-

thingwe’llseealotmoreof,notjustforsuddencardiac

deathbutfordiabeticcomplicationsandallsortsofother

preventable,high-stakesconditions.”

EstablishingNewRelationships

AmongKPIs

Becauseperformancedriversinteractincomplexways,

youdon’tjustneedbetterKPIs:Youneedabetterunder-

standingofhowKPIsarerelatedtoeachother.Wesee

executives,analysts,anddatascienceteamscollaborating

toidentifyinterdependenciesandestablishnewrelation-

shipsamongKPIs.Theserelationshipsmayrevealtacit

orsubtlelinksbetween,say,employeeproductivityand

customerengagement,profitmarginsandmarketshare,or

qualitymanufacturingoutputandreturnonassets.

Ourresearchfindsgrowingdemandfornew,AI-drivenKPIs.

TheFutureofStrategicMeasurement:EnhancingKPIsWithAI

7

8

MITSLOANMANAGEMENTREVIEW?BCG

PernodRicard,a$10billionglobalspiritscompany,usesAItodescribeanddeepentheconnectionbetweentwoofitsmostimportantKPIs:profitmarginsandmarketshare.Inthepast,theseKPIsweresiloed,eachwithitsownsetofmeasures.Thefinancefunctionfocusedonprofitability,whilesalesandmarketingfocusedonmarketshare.

ThecompanynowdeploysAItodeliverinsightsintohowcommercialandmarketinginvestmentsthatimproveprof-its—suchasmediaorin-storeactivation—alsoinfluencemarket-shareobjectivesandviceversa.InsteadofseekingtomaximizeeachindividualKPI,thespiritsmakernowseekstooptimizebothKPIsinconcertwitheachother.

“Ifyoucanimaginemovingacursorbetweenmarketshareoptimizationobjectivesandmarginoptimizationobjec-tives,youneedtoknowhowtherequiredinvestmentsvarytoreachtheseobjectives,”saysPierre-YvesCalloc’h,PernodRicard’schiefdigitalofficer.“AIisgoingtogiveyouthatinformation.WithAI,wecanbetteralignmar-ketshareKPIs,marginKPIs,andrequiredinvestmentstoreachthem.”ThiscapabilitytransformedhowPernodRicard’sleadershipallocatescapitalandbalancesitsaspi-rationsforprofitabilityandmarketshare.

BOLSTERINGKPISATWAYFAIR

“Inbusiness,there'satendencytooptimizeoper-ationsformicroKPIs.However,asingularfocusonthesemetricscansometimesleadtoa‘localmaxima’problem,wherethebenefitstoonepartofthebusinessmaycomeattheexpenseoftheoverallhealthofthebusiness,”notesFionaTan,Wayfair'sCTO.Forexample,optimizingfortheprofitabilityofcertainproductsorservices,s