[Webinar Slides] How to Increase Your Profits by Improving Your Data Accuracy

Underwri(enby: Presentedby:

#AIIMInforma(onIsYourMostImportantAsset.LearntheSkillstoManageIt.

Underwri(enby: Presentedby:

HowtoIncreaseYourProfitsbyImprovingYourDataAccuracy

PresentedMarch7,2017

Underwri(enby: Presentedby:

SethMaislinPrincipalConsultantEarleyInforma(on

Science

GregCouncilVP,Marke4ng&

ProductManagementParascript

Host:TheresaResek,CIPDirectorAIIM

Today’sSpeakers

Underwri(enby: Presentedby:

SethMaislin

PrincipalConsultant

EarleyInforma(onScience

IntroducingourSpeaker

Underwri(enby: Presentedby:

Qualityvs.Profitability:AnObserva(on

Features:•  poordata=businessloss•  diminishingreturns•  “goodenough”exists•  culturema(ers

Underwri(enby: Presentedby:

Correla(onorCausa(on?

§  MostorganizaKonsacknowledgethecorrelaKonbetweendataqualityandbusinessvalue,butfewcansaywhichmetricsofdataqualitymaRermost.

§  Withdata,“goodenough”isastrategicchoice.

Ifyoucanextract… Thenyoucanimprove…

zipcodes geographicalanalysis

namesofdependents frauddetecKon

dollaramounts arithmeKcvalidaKon,workflow

everythingoverKme UnitedStatestaxandfinancialsystems

anything savings,speed,capability,growth,…

U.S.FederalTaxForm

Underwri(enby: Presentedby:

GoverningData“ARrac(veness”

§  DataqualityismulKdimensional.§  Whichfeaturesma(ermost?Why?(Howdoyousettargets?)§  It’snearlyimpossibletoopKmizealldimensionsatonce.

•  accurate•  available•  certain/precise•  clean•  complete•  consistentacrosssources

•  integrity

•  formaRed(conformity)

•  reasonable/logical•  relevant/valid•  reliable•  (mely/current•  traceable/audited

Underwri(enby: Presentedby:

Underwri(enby: Presentedby:

DataIsanAsset

•  somethingyoucansell•  directinputforabusinessprocess•  contextualinputformakingabusinessdecision•  workflowtrigger•  macro-levelself-awareness(governance)

Toknowifyourdataaregoodenough,youdon’tmeasurethedata.Youmeasureitseffects.

DatagovernancerequiresanintegratedsystemofmetricsandKPIs.

Underwri(enby: Presentedby:

ProcessesenableobjecKves

LINKAGE

EnterpriseStrategy RevenueGrowth

Contentsupportsprocesses

ObjecKvesalignwithstrategy

CEO:“Howwillthisincreaserevenue?”

BusinessProcessesMeasuringhere

(processindicators) Sessions SearchRelevance

Leads ConversionRatesProcessScorecards

BusinessUnitObjecKves

Measuringhere(businessoutcomes)

NewBusinessOpportuni(es

AverageOrderSize TotalAccountRevenue

OutcomeScorecards

DigitalTeam:“HowdoIknowtaxonomy/content/searchisworking?”

DigitalContent

Working&Measuringhere(content,IA,taxonomy,search,productdata,etc.) Site

Search

DigitalContent&SEO

WebAnaly(cs

ContentScorecards

CTR DataFill Content etc.

Metrics&KPIsFocusGovernance

Underwri(enby: Presentedby:

GedngDataQualitytoPayOff

EnterpriseStrategy

BusinessProcesses

BusinessUnitObjecKves

DigitalContent

•  deliverables quality•  goodreturnonITspending cost/profit•  increasedthroughput growth•  prioriKzedteambehaviors produc4vity•  errorreducKon/compliance liability

LINKAGE

•  brandpercepKon,reputaKon quality•  revenuedollars cost/profit•  customers,marketshare growth•  businessstrategyrealizaKon produc4vity•  marketawareness liability

Underwri(enby: Presentedby:

UseCases:HowtoLeverageFinancialData

•  accurateandKmelyaccountsmanagement•  be(erspendalignmentwithbusinessstrategy;innovaKonrecogniKon•  improveddecisionmakingfromdata-drivenROI;smarterR&D•  acKveassetmanagement(human,technology,physical,digital)•  smartbusinessculturedesign•  intelligentassetandpartneracquisiKon,compeKKveprocurement•  strongriskmanagement•  improvedcompliance•  customerandmarketplaceintelligence(correlaKons,shocks,trends,gaps)•  streamlinedoperaKons

Underwri(enby: Presentedby:

Buildingthelinkageisthehardestpart.

1.  Establishasystemofcompany-widegovernance.§  IdenKfyandunderstandthecompany’sdatadomains§  DocumentprocessesandassignaccountabiliKes§  EstablishaCenterofDataExcellence,andsocialize§  Implementmetricstomeasuregovernanceitself

2.  Decidewhatma(ersatyourlevel.§  KnowthestrategicKPIsforeachpartofthebusiness§  IdenKfygapsbetweenexisKngandnecessarymetrics§  Inventoryandprofileyourdata;calculatebaselines§  SettargetscollaboraKvelyandfindsynergies

3.  Buildyourdataprojectsroadmap.

LINKAGE

Underwri(enby: Presentedby:

OtherResourcesandContact

§  TheEvolvingRoleoftheChiefDataOfficer:IncreasingInfluenceintheC-suite(on-demandwebcast)

§  DigitalGovernance:It’saNumbersGame(Metrics,KPI’s,andROI,thatis)(arKcle)

[email protected]

781-775-0684@sethmaislin

www.earley.com

Underwri(enby: Presentedby:

GregCouncil

VP,Marke(ng&ProductManagement

Parascript

IntroducingourSpeaker

Underwri(enby: Presentedby:

Func(onalandPerformanceAbili(es

FUNCTIONALABILITYPERFORMANCEABILITY

SystemDataFlow

Underwri(enby: Presentedby:

Observa(onsandQues(ons

PERFORMANCEABILITY

ManyorganizaKonsfocustoomuchonfunc8onalcapabiliKesofthesystem,andnottheperformanceofthesystem.

§  Whatistheerrorrateofyourstaff?§  Whatistheerrorrateofyoursystems?§  Whatistheactualuplio/costcreatedbyyoursystems?

Isitwrongtoask“howaccurateisyoursystem”?(Forexample,OCRaccuracy?)

§  YesandNo–Knowwhattoaskfor

Whatisthebestwaytomeasureperformance,andhowcanhighperformancebeachieved?

SystemDataFlow

Underwri(enby: Presentedby:

Func(onal:HowtoAssignDocumentstoTypes

UsingVisualElements

TextstaKsKcallyorbasedonNLP

BusinessRulesManualvsAutomated

FUNCTIONALCAPABILITIES

Underwri(enby: Presentedby:

Func(onal:RangeofDataSupported

Languagesthatcanbesupported

Typesofdatasupported

Formats:text,constrainedhandprint,unconstrainedhandwriKng,electronic,image-based

Structure:standardized,parKallystandardized,

nostandard

RANGEOFDATASUPPORTED

Underwri(enby: Presentedby:

Self-DrivingCars&ManualVerifica(on

…AndcompaniessKllmanuallyverifytheirrecogniKon

results?

Underwri(enby: Presentedby:

WRONGWAY-ManualDataEntry

RECOGNITION

INPUTIMAGES

MANUALDATAVERIFICATION&MANUALENTRY

Underwri(enby: Presentedby:

RIGHTWAY:FullAutoma(on

ACCEPTED

99.5%CORRECT

INPUTIMAGES

Error

RECOGNITION

Underwri(enby: Presentedby:Underwri(enby: Presentedby:

Discussing“Accuracy”

§  IsOCRaccuracyrelevant?Yesbut….•  Itdoesn’ttellyoudata-levelerrorofasystem;typicallyonlycharacter-levelaccuracy.•  Itdoesn’tinformyouofthecostsavingsorproducKvitygains.

o  Forexample,“HowmanydataelementsdonotrequireanyintervenKonorreview?”

§  YourdataaccuracyisnotthesameastheOCRaccuracy.

§  Most(orallof)OCRaccuracyismeasuredatthecharacterlevel.

§  Yourformdataisimportantatawordorfieldlevel.

IfOCRoutputis99%accurateatthecharacterlevel,thenconsideringapagewith500wordswithanaverageof6le(ersperword,30characterswillbeincorrect.Those30errorscanbedistributedacrossmulKplewordsleadingtoahigheractualerrorrate.Aword-levelerrorratecouldbe6%,not1%.Ifpartofthe6%

errorincludesanaccountnumberofSSN,that’sabigproblem.

Underwri(enby: Presentedby:

GedngatActualDataAccuracy

Howmuchdataislocatedandwhatistheerrorratefield-by-field?WiththisinformaKon,youcanhaveasolidunderstandingofgains.

§  SSNhasareadrateof85%anderrorof.5%translatesto:Only15%ofdatarequiresanyreviewatall.

§  Theremaindercangostraightthrough.

READ

RAT

E

ERRORRATE

100%

100%

Confidencethreshold=15

Confidencethreshold=50

FuncKonal-onlySystem

SSN OtherData

Underwri(enby: Presentedby:

WhenItComestoPerformance

PERFORMANCEABILITY

Classifica(on:§  Recallandprecision.AdocumentclassifierwillnotcorrectlyidenKfy

alldocumentsandperformanceiskey.•  Precision(Relevance).IfthesystemidenKfied8invoicesinabatchof20

documents,where5arecorrectand3areincorrect,precisionis5/8or62.5%.

•  Recall(Completeness).Iftherewereactually15invoicesinthebatchof20,recallis5/15or33.3%

DataLoca(on/Extrac(on:§  ReadRate.Thepercentageofdatathatislocatedandextracted.§  ErrorRate.Thepercentageofthatlocated/extracteddatathatisnot

correct.§  ConfidenceThreshold.ThestaKsKcally-significantpointthatallowsfor

datatomovestraightthroughthesystem.§  Allatthedataelementlevel

SystemDataFlow

Underwri(enby: Presentedby:

TechnologyKnowhow&StaffExper(se

SignificantuniqueexperiencewithAIandpa(ernrecogniKonexperKsethatsetstheteamapart.

AppliedExperKse ExperKseLevel DomainExperKseY

AbilitytoperformsophisKcatedanalysisofdocuments,datasets,anddevelopsoluKonswithtunedperformance.

In-depthexperiencewithdataextracKondeploymentsandtrackrecordofdeliveringhighqualityresultson-Kme.

TypicalsoowarevendorsuseinformaKonsystemsandsoowaredevelopmentstaffwithgeneralsoowareskillstobuildsoluKons What to Look For:

Underwri(enby: Presentedby:

ParascriptEnterpriseEnablementProgram

EnterpriseEnablement

IdenKfydiscrepanciesbetweendataqualityexpectaKonsandrealresults

Implementconsistentmeasurementsthroughouttheworkflowprocessforongoingdataqualityassessments

Fillthegapsinthecurrentdeliverymodel,opKmizeprocessestoensurehigh-qualitydataresultsandincreaseefficiencyofoperaKons

MonitorResults

Underwri(enby: Presentedby:

ParascriptAcceleratorProgramforBPOsMonitorResults

SignificantlyreducecoststhroughautomaKonofdataentry-intensiveprocessesusingprecisiontechnologyandaprovenconsultaKveapproachthatidenKfiesbaselineperformanceandconcrete,acKonableareasforperformanceimprovement.

Herearethephasesoftheprogram:

1.  Onsitediscovery2.  Baselineperformanceanalysis3.  CostreducKonopportunityidenKficaKon

4.  Businessacceleratoranalysisreportdelivery5.  Comprehensivein-personbriefingoftheresults

Underwri(enby: Presentedby:

ThankYou!

GregCouncilVicePresident

Marke(ng&ProductManagement

[email protected]

888-225-0169

Underwri(enby: Presentedby:

WanttoLearnMore?

www.parascript.com

[email protected]

(888)225-0169

YOURSOLUTIONSTRANSFORMBUSINESSES.LETOURSOLUTIONSDELIVERYOURDATA.

Underwri(enby: Presentedby:

TakeyourskillstothenextlevelbylearningbestpracKcesandtechnologiesfordigitalimagingwithAIIM’sCapture&Imagingtrainingcourse.

Visit:AIIM.org/CaptureTraining

Underwri(enby: Presentedby:

AIIMistheCommunityforInforma(onProfessionals

AIIMbelievesthatinforma(onisyourmostimportantasset.Learntheskillstomanageit.

OurmissionistoimproveorganizaKonal

performancebyempoweringacommunityofleaderscommi(edto

informaKon-driveninnovaKon.

Learnmoreatwww.aiim.org