建立 GenAI 商業案例.pdf

《建立 GenAI 商業案例.pdf》由會員分享，可在線閱讀，更多相關《建立 GenAI 商業案例.pdf（43頁珍藏版）》請在三個皮匠報告上搜索。

1、Establishing the Business Case for GenAIData Summit-May 20252025 ExlService Holdings,Inc.All rights reserved.2Dave CroleneEXL VP of Data,Analytics&AIDecades of experience in Data,Analytics and AIGoal:Leverage technology to drive business results and innovationFocus:AI and Data StrategyB ExlService H

2、oldings,Inc.All rights reserved.3AgendaSectionSection DurationDurationPart 1-AI Intro 5 min History,Impact&Sample Use CasesPart 2-Investment Types 5 minData at the Center“Foundational”and“Impact”InvestmentsPart 3-Make the Case 10 minMeasure AnythingBuilding a Business CasePart 4 Example(s)15 minPart

3、 5-Closing Thoughts/Q&A 10 min2025 ExlService Holdings,Inc.All rights reserved.4The Impact of AIWhy AI:Economic Impact:Generative AI is projected to add$4.4 trillion annually to the global economy*.Business Transformation:From automating content creation to improving customer service,Generative AI e

4、nables efficiency,creativity,and innovation across industries.Competitiveness:Organizations leveraging Generative AI gain a competitive edge,reducing time-to-market and enhancing decision-making.*The economic potential of generative AI:The next productivity frontier(McKinsey,2023)Customer Operations

5、Marketing and SalesSoftware EngineeringResearch and Development2025 ExlService Holdings,Inc.All rights reserved.5Artificial Intelligence An EvolutionDawn of AI(1950-61)Expert Systems(1961-2010)AI Renaissance(2010-2025)Contemporary(2025+)Machine Learning:Customer Service Automation Fraud Detection Na

6、tural Language Processing:Intelligent Document Processing Financial Reporting AutomationRobotic Process Automation:Claims Processing Agent AssistGenerative AI:Document Verification Real-time Intelligent Agent Assist Data MigrationDeployed Use CasesArtificial Intelligence(AI):Machines acting in ways

7、that seem intelligentRule-Based Systems:Make decisions using if-then logicExpert Systems:Mimic expert decision-makingNeural Networks:Brain-inspired pattern recognitionAI Winter(1980-2010)Generative AI:Creates content like text or imagesLarge Language Model(LLM):Advanced AI models that understand and

8、 generate languageMulti-modal:Combines various data types(text,images,audio)Retrieval-Augmented Generation(RAG):Merge retrieval and generative models for improved contextMachine Learning(ML):Find patterns to predict outcomes Natural Language Processing(NLP):Interprets and generates human languageRob

9、otic Process Automation(RPA):Automates routine tasks with AIDeep Learning:Neural Networks with many layers for complex data analysisRetrieval-Augmented Generation(RLHF):Train AI by rewarding good behavior with human guidance2025 ExlService Holdings,Inc.All rights reserved.6The art of the Possible AI

10、 Impact FrameworkInternalExternalOptimizeInnovateCode GenerationData ClassificationData Cleansing and PrepAutomated Testing/Synthetic DataEnhanced Customer SupportDynamic Marketing PersonalizationSmart Contract GenerationAutomated Transaction SummaryInnovative Product DesignScenario Planning&Strateg

11、yVirtual Prototypes/SimulationsTraining ScenariosCustomized User ExperienceGenerative Content for EngagementAI-Driven R&D SolutionsGenerative Design2025 ExlService Holdings,Inc.All rights reserved.7Example Use Cases for ConsultingHow Can Generative AI Revolutionize the Way Consultants Deliver Value

12、to Clients?Covered consulting use cases:1.Onboarding/Client Research 2.Brainstorming3.Risk identification4.Content Creation and Refinement5.Client Engagement6.Learning7.Interview Preparation8.Notes ExtractionInvestment Types2025 ExlService Holdings,Inc.All rights reserved.9Data is the foundation tha

13、ts our focus todaySpeed to insightsSpeed to outcomesDataDataAnalyticsAnalyticsDigital Digital operationsoperationsDigital Digital and AIand AIDataData-led led valuevaluecreation creation Embed intelligence in the workflowExecute operationsEnable better business decisions2025 ExlService Holdings,Inc.

14、All rights reserved.10Accelerate Your BusinessAccelerate Your BusinessAccelerate Your DataAccelerate Your DataEngineeringTestingQualityAnalyticsInnovationGrowthEngagementPartnershipsStrategy&GovernanceClassificationOptimizationThe AI Virtuous Cycle-The end and the meansAI2025 ExlService Holdings,Inc

15、.All rights reserved.11Hardware/Software/Services to store,process,and manage dataData Platform implementation to support the broader organization and future AI use casesData Governance capabilities to ensure data quality,security and complianceMetadata management Operating model transformation to b

16、etter support data initiatives and workflowsTraining and enablementCustomer Segmentation AnalyticsPredictive Analytics Smart Agent AssistFraud DetectionProcess Optimization AnalyticsReal-time data feeds for time-sensitive processes or decisionsProduct AnalyticsVendor AnalyticsFoundationalImpactDirec

17、tly contributes to strategic business value:Revenue,Margin,Market Position,RiskDrives immediate value through process optimization,but more importantly enable Impact OpportunitiesExample“Foundational”and“Impact”opportunities2025 ExlService Holdings,Inc.All rights reserved.12Good news,most“Foundation

18、al”use cases reside hereInternalExternalOptimizeInnovateSelf-healing ETL AgentAuto-merge CI AgentSchema-drift Detection AgentResource-scaler AgentDependency-updater AgentData AnnotationRegulatory-Reporting AgentVendor/Partner-Synch AgentAPI-Tuning AgentCustomer Data Onboarding AgentSLA Monitoring Ag

19、ent Pipeline synthesizer AgentData Classifier AgentCost-Performance Experimentation Anomaly Rule Discovery AgentGovernance Policy Generator AgentData CurationDaas Product Creation AgentPersonalized Pipeline AgentMarketplace Curation AgentPartnership Recommendation AgentExternal Support Chatbot Agent

20、2025 ExlService Holdings,Inc.All rights reserved.13ConsumeConsumeLooking Forward Intelligent OrchestrationCaptureCaptureCurateCurateReporting&AnalyticsReporting&AnalyticsApplications&External FeedsApplications&External FeedsStructuredStructuredSemiSemi-StructuredStructuredUnstructuredUnstructuredDat

21、aDataLakeLakeODSODSDWDWFabricLakehouseMeshData ProductsData PlatformData PlatformData Engineering|Governance|DataOpsData Engineering|Governance|DataOpsAI-driven ClassificationConversational BI monitoringData Platform auto-provisioningAuto-adjusting ingestion pipelinesAgent-initiated pipeline restart

22、sData archivalPipeline monitoring/orchestrationData Quality monitoring&managementSupervisory AI AgentReal-time fraud monitoringMetadata maintenanceAI Data AnnotationData curation2025 ExlService Holdings,Inc.All rights reserved.14 and without a proper foundation,AI is at best,riskyTechnicalOperationa

23、l RisksStrategic RisksAnalytic CapabilitiesTeam RisksCode MaintainabilityIntegration&Scalability IssuesLicense and Compute CostsCross-compatibilitySystem DowntimeRegulatory ChangesCompliance RiskDataOpsSilos and In-fightingIT/Business alignmentStrategic Goal AlignmentCompetitive PressuresOpportunity

24、 Costs of Fighting FiresBrand EquityLack of Metadata and Self-serviceInconsistent Enterprise AnalyticsData QualitySkill ImbalanceResource BurnoutKey Person DependencyUncoordinated effortsMake the CaseFirst you must measure it2025 ExlService Holdings,Inc.All rights reserved.17Why measure it?“If it ma

25、tters,it can be measured”D.HubbardHubbard,Douglas(2014).Hubbard,Douglas(2014).How to Measure Anything.Finding the Value of“Intangibles”in BusinessHow to Measure Anything.Finding the Value of“Intangibles”in BusinessInformed DecisionInformed Decision-MakingMakingMeasurement reduces Measurement reduces

26、 uncertaintyuncertaintyInvestment Investment JustificationJustificationQuantified benefits Quantified benefits support funding support funding approvalapprovalStakeholder Stakeholder CommunicationCommunicationClear metrics convey Clear metrics convey cost,value and risks cost,value and risks effecti

27、velyeffectively okay fine,but how?2025 ExlService Holdings,Inc.All rights reserved.18Concept 1 DecompositionPrincipleBreak down complex projects into measurable componentsBreak down complex projects into measurable componentsApplicationSeparate costs,benefits,and various risk types Separate costs,be

28、nefits,and various risk types Further break down each into smaller measurable areasFurther break down each into smaller measurable areasExamplesBenefits:cost savings like reduced support hours,performance gains,Benefits:cost savings like reduced support hours,performance gains,scalability scalabilit

29、y Costs:project,software,compute,training,supportCosts:project,software,compute,training,supportRisks:compatibility issues,compliance audit,personnel availabilityRisks:compatibility issues,compliance audit,personnel availabilityBreaking Down Complex Benefits,Costs,and Risks2025 ExlService Holdings,I

30、nc.All rights reserved.19Concept 2 Calibration and Confidence IntervalsPrincipleUse 90%confidence intervals to express estimates,providing a Use 90%confidence intervals to express estimates,providing a realistic rangerealistic rangeApplicationUse this to estimate benefits,costs,and risks.Use this to

31、 estimate benefits,costs,and risks.Make sure you ask the person with the highest expertise for the Make sure you ask the person with the highest expertise for the specific estimate.specific estimate.ExamplesEstimating with a 90%confidence,improvement to project Estimating with a 90%confidence,improv

32、ement to project completion pletion time.Support Cost Reduction:Estimate a 25%Support Cost Reduction:Estimate a 25%40%decrease,equating to 40%decrease,equating to$50k$50k$80k annually.$80k annually.Data Wrangling Reduction:Data Science and Analytic team Data Wrangling Reduction:Data Science and Anal

33、ytic team performance increased by 20performance increased by 20-40%40%Estimating with Confidence2025 ExlService Holdings,Inc.All rights reserved.20Concept 3Value of Information(VOI)PrincipleFocus on measuring variables where additional information will Focus on measuring variables where additional

34、information will significantly impact the decisionsignificantly impact the decisionApplicationIdentify which costs,benefits and risks have the highest uncertainty Identify which costs,benefits and risks have the highest uncertainty Identify which costs,benefits and risks have the highest potential I

35、dentify which costs,benefits and risks have the highest potential impact on ROIimpact on ROIDecide on conducting a pilot programs or if additional information Decide on conducting a pilot programs or if additional information should be collectedshould be collectedExamplesDetermine if further data co

36、llection on productivity improvements is Determine if further data collection on productivity improvements is neededneededAvoid measuring irrelevant variables to streamline data collection Avoid measuring irrelevant variables to streamline data collection processesprocessesPrioritizing Measurements

37、That Matter2025 ExlService Holdings,Inc.All rights reserved.21Concept 4Using Analogies for EstimationPrinciple of Using AnalogiesEstimate benefits by comparing with similar projectsEstimate benefits by comparing with similar projectsLeverage experiences from other organizationsLeverage experiences f

38、rom other organizationsApplicationBenchmarking:Compare expected outcomes based on industry data or case Benchmarking:Compare expected outcomes based on industry data or case studies.studies.Identify projects with similar scope and outcomesIdentify projects with similar scope and outcomesAnalyze bene

39、fits achieved in those projectsAnalyze benefits achieved in those projectsProvides relatable examples and strengthens the argumentProvides relatable examples and strengthens the argumentExamplesA published article from a competing company indicates they saved$14M A published article from a competing

40、 company indicates they saved$14M over a period of 3 years by deploying AI capability Xover a period of 3 years by deploying AI capability XGartner reports an average product discovery velocity of 30%for Gartner reports an average product discovery velocity of 30%for organizations that deploy AI cap

41、ability Yorganizations that deploy AI capability YLeveraging Similar Cases2025 ExlService Holdings,Inc.All rights reserved.22Summary Remember Remember Anything Can Be MeasuredAnything Can Be Measured:Systematically analyze:Systematically analyze and estimate costs,benefits and risks directly or indi

42、rectlyand estimate costs,benefits and risks directly or indirectly Apply Apply Decomposition Decomposition technique:Break down complex technique:Break down complex benefits into measurable componentsbenefits into measurable components Use Use Calibration and ConfidenceCalibration and Confidence:Use

43、 calibration techniques:Use calibration techniques and Provide realistic estimate ranges.90%and Provide realistic estimate ranges.90%Focus on Focus on HighHigh-Impact Areas with VOIImpact Areas with VOI:Use VOI to prioritize:Use VOI to prioritize measurements,and Identify areas that most affect the

44、measurements,and Identify areas that most affect the business casebusiness case Enhance Credibility with Enhance Credibility with AnalogiesAnalogies:Support estimates with:Support estimates with comparable cases and use analogies to make arguments more comparable cases and use analogies to make argu

45、ments more persuasivepersuasiveThen,you can build the case2025 ExlService Holdings,Inc.All rights reserved.246.Present the Business Case with Risks and Mitigation Strategies5.Deep Dive Analysis of Top Opportunities4.Initial Evaluation of Opportunities3.Differentiate“Foundational”from“Impact”Opportun

46、ities2.Research the“Art of the Possible”1.Align with Corporate Goals and Stakeholder CriteriaDeveloping a data-driven business caseA proven method to evaluate a broad range of potential data-driven opportunities,categorize and prioritize them,deep-dive and fully understand the business impact for th

47、e top opportunities and present the case to key stakeholders and the sponsor2025 ExlService Holdings,Inc.All rights reserved.251,2,3 The basicsAlign on corporate and stakeholder objectives,potential opportunities,and understand the prerequisites1.Align 2.Research 3.Differentiate GoalsGoalsExampleExa

48、mpleCorporateRevenue,Margin,NPSStakeholderCustomer Cross-SellApprovalCost,TCO,ROI,NPV,IRRDetermine:Internal Use CasesIndustry Use Cases“The Art of the Possible”Competitive LandscapeEstablish:“Foundational”Use Cases-prerequisites“Impact”Use Cases Value to the businessOpportunities to phase Foundation

49、al2025 ExlService Holdings,Inc.All rights reserved.26Articulate Opportunities Long description(sentences)Long description(sentences)“Foundational”or“Impact”“Foundational”or“Impact”Identify Associated Inputs Required dataRequired data Impacted business processesImpacted business processesIdentify Pot

50、ential Risks TechnicalTechnical FunctionalFunctionalEstablish Evaluation Criteria ValueValue FeasibilityFeasibilityScore and Prioritize Multiple perspectivesMultiple perspectives Business and ITBusiness and IT4.Initial evaluation of opportunitiesFind and rank potential opportunities leveraging AI to

51、 produce valueOpp#OpportunityDescriptionName 1 Name 2 Avg(Company)Variance%Name 3 Name 4 Name5 AVG(EXL)Variance%Avg(Company+EXL)Adjusted to 0-9 Scale1 EDW AssessmentEvaluate EDW for data quality,ETL,and automation,and recommend preservation,upgrading,replication,or replacement.282929-3%283230-5%295.

52、82 Fund Valuations EngineOptimize the valuations engine for scalability and performance to support retail and future investor products.36313416%343635-4%348.13 Fund Valuations OrchestrationCross-Business Unit Data and Analytics for Fund financial close.Data platform,metadata,and governance,data line

53、age.Target:spotting discrepancies and reducing risks.Start with IS to support Retail.Key person dependencies.4 people in CS assigned-2.5-3 day window-intense to find discrepancies-24 hour days.SEC footnotes are result of issues-monthly,quaerterly and yearly.Cost savings,but bigger is risk.283230-13%

54、313131-3%316.34 Vendor Expense AnalyticsDevelop a data pipeline from PeopleSoft for fund management expenses and analytics.33293114%28272813%295.85 Partner Experience DataPartner Experience from a data perspective.Component for analytics and metadata to support.Cross-Business Unit data integration.R

55、obust QC.Investors discover at least 1 issue per month and sometimes 2-3 per month in the portal.Time consuming and costly for one-off analytic requests from high-value Partners(weeks or months-plus Business Unit 37303423%263135319%327.16 Hardent APartnerinvest DataStrengthen APartnerinvest investor

56、 data storage in Snowflake for GIS through DE and automation processes.242625-8%272627-6%264.17 Accounting Close DiscoveryCapture metadata for fund and corporate accounting close,including lineage,data catalog,and classification to support manual close steps.293030-3%2529279%285.38 Retail PlatformEs

57、tablish data platform and analytics framework for data capture and reporting.3130313%323333-6%326.89 Financial Data FlowEvaluate,document,and store data flow details between Investran,OneStream,and Peoplesoft.202724-26%2023229%232.6Score(0-10)OpportunityDescriptionFoundational?Stakeholders ImpactedS

58、trategic AlignmentBreadth of UtilityProcess OptimizationScalabillity/PerformanceTotalAdjusted to 9 Scale 2 Person MonthsThe number of stakeholders that identified issues addressed by opportunitySupports AUM,Margin,Investor Alignment,Organizational AgilityCross-Business Unit,executives,process owners

59、,Business Unit leadersReduced Effort,cost,risk,timelineCan grow to support new capabilities,transparency,accessability,time to valueEDW AssessmentEvaluate EDW for data quality,ETL,and automation,and recommend preservation,upgrading,replication,or replacement.Small1667762682025 ExlService Holdings,In

60、c.All rights reserved.271.Research2.Strategic View3.Functional View4.Opportunity Definition5.Deep dive:Business Value FrameworkThe Business Value Framework is often necessary for“Foundational”opportunities.8.Impact&Assumptions9.TCO&ROI10.Business Value Definition10123678549Outputs Detailed Interview

61、 Notes Conceptual Bus Process Model Categorized Opportunity List:Business Impactful(3-6)Secondary Options Metrics,baseline values and data benefit Quantified ROI for Business Impactful Opportunities Candidate Project Scoping Assumptions Value Proposition Implementation Options Implementation Plan/Ro

62、admap Identified RisksDiscovery(1Discovery(1-3)3)OpportunityOpportunityQuantification(4Quantification(4-9)9)Business Value Business Value Definition(10)Definition(10)5.Business Process&Data Sources6.Metric Identification7.Metric Baseline2025 ExlService Holdings,Inc.All rights reserved.28Deep dive ex

63、ampleCorporate Goal AlignmentRevenue:Enhance decision making with higher data quality and consistency for analyticsMargin:Reduce manual processing costs to free up data-engineer capacityInnovate:Accelerate time-to-insight to support faster,data-driven decisionsInvestment Metrics Tooling:$150K Integr

64、ation&Training:$350K Total:$500 KSavings Metrics Manual annotation:4,000 h$75/h*5075%Range:$150K-225K/year Error remediation:$120 K*3050%avoidance Range:$40K-60K/year Annual Improvement:15%3-Year Savings:$1.52M 2.45M Net Benefit:$1.02M 1.95M3-Year Savings&ROI:ROI(3-Year):204%390%Deploy an AI annotat

65、ion engine that automatically extracts and maps fields from unstructured documents(e.g.,emails,PDFs,logs)into the enterprise data modeleliminating manual tagging and reconciliation.2025 ExlService Holdings,Inc.All rights reserved.296.Core components of a Business caseMake it simple and impactful One

66、 slide is often enough as long as the supporting evidence is at handOverview of the opportunity and challenges that it addressesIntroductionHow the opportunity maps into the organizational goals such as:Financial,Operational,Market,Customer,EmployeesCompetitive positioningStrategic Alignment Require

67、d financial information such as Investment,ROI,Break-evenQuantitative Benefits(e.g.,30%reduction in data wrangling,$500K annual software license reduction)Qualitative Benefits(e.g.,improved sales performance visibility,enhance customer experience)Financial Highlights Relevant risks to the program su

68、ch as technical,security/compliance,change managementRisk Overview Final recommended actionProposed next stepsRecommendation&Next Steps Some More Examples2025 ExlService Holdings,Inc.All rights reserved.31Example 1:Foundational and ImpactAn Insurance firm looks to drive a Data Product strategyA majo

69、r insurance company wants to justifies enterprise investment with Direct and Ancillary impacts:1.Currently funding all data and AI investments via Business Unit use case investments2.Little to no enterprise(foundational)investment has left the data extremely siloed for some Bus and across Bus and Fu

70、nctions3.The organization struggles to find operational efficiencies,understand their customers across data sources,scale the product roadmap with constrained data processes,and provide enterprise-wide analytics without one-off,error-prone and slow data wrangling efforts2025 ExlService Holdings,Inc.

71、All rights reserved.33Example 1:Expected Benefits across the Strategy PillarsEnd to End Data Governance Role-Trust&Confidence in DataData DemocratizationEfficient BAU data product deliveryData CatalogingEfficiency on Data Analysis expressed as a%savings of timeYr 1:$0.7 MYr 2:$1.4 M(+/-10%)Resource

72、ProductivityData Quality Metrics(Defect Velocity,Resolution Rate etc.)Trust&Confidence in DataRegulatory Compliance and Third-Party Data Contractual FinesStrategic Cross-functional team for Data Product build-out%Time savings fromBI/Analyst Efficiency LiftPipeline ReuseRework ReductionIncremental Re

73、use of modelsYr 1:$1.5 MYr 2:$2.1 M(+/-10%)Avg.Project Timeline&CostThroughput(backlog Items)Foundational step for Advanced analytics/AI adoption and SolutioningFaster TAT for Business Analytical InsightsEnterprise Collaboration:Business and IT alignment Promotion of Unified analytical products(vs s

74、iloed functional reporting)Analyst empowerment:Self-service enablementPlatform Cost OptimizationAbility to mine from unstructured dataNear Real-time Data%of Rationalized Licenses(Inactive users and reduction in usage of Visualization toolStorage Cost Savings(using modern architecture)Annual:$0.7M(+/

75、-10%)Software/Licenses CostsStorage CostsSingle source of truth vs Siloed dataUsing unstructured data for AI driven business Use cases(Call Center Ops etc.)Operating ModelGovernance and QualityProcessArchitecture and TechnologyKEY ITEMSEXPECTED BENEFITS:SAVINGS EXPLANATIONEXPECTED BENEFITSMEASURABLE

76、 METRICSANCILLARY BENEFITS2025 ExlService Holdings,Inc.All rights reserved.34AI and Data Governance CommitteeAI and Data PoliciesAI Risk ManagementEnhance Data Management PracticesTraining&Culture ShiftEnhanced Compliance MonitoringExample 2:FoundationalImplement AI and Data governance at Apex Finan

77、cialChallenge:Apex aims to leverage AI to enhance operations and customer service through initiatives like:Smart Agent AssistDocument Fraud DetectionPersonalized Investment RecommendationsHowever,they have concerns over:Need for AI Governance to safely enable AI initiativesRegulatory Compliance to a

78、void fines($1-$3M)Risk Management to reduce AI project failuresOperational EfficienciesAI and Data Governance FrameworkApproach:Implement an AI and Data Governance Framework to manage risks and support AI initiatives safely and efficiently.2025 ExlService Holdings,Inc.All rights reserved.36Example 2

79、:Cost and benefit analysis(Over 3 years)Costs:Initial Investment(Year 1):$700K Policies and Governance Structure:$250K Governance Tools:$100K Data Mgmt.Enhancements:$150K Implementation&Integration:$100K Initial Training:$100KAnnual Ongoing Costs(Years 2&3):$45K/year Technology Maintenance:$25K Recu

80、rring Training:$20KTotal Costs Over 3 Years:$790KQuantified Benefits:Reduced Regulatory Fines:$0.5-1.5MIncreased AI Implementation Efficiency(20-30%):$1.1M-2.2MReduced AI Project Failures(20):$1.8MReduced Operational Expenses(10-30%):$0.5M 1.5MTotal Benefits Over 3 Years:Minimum:$3.9MMaximum:$7M2025

81、 ExlService Holdings,Inc.All rights reserved.37Example 2:Ancillary benefit analysis(Over 3 years)Benefits from Dependent Impact Opportunities:Increased Efficiency Smart Agent Assist Reduced handling time(15Smart Agent Assist Reduced handling time(15-25%):$600K 25%):$600K$1.2M$1.2MDocument Fraud Dete

82、ction:Document Fraud Detection:Reduced Fraud Loss(30Reduced Fraud Loss(30-50%):$600K50%):$600K-$1M$1MReallocate Manual Verification Staff(5 employees):$350KReallocate Manual Verification Staff(5 employees):$350KTotal:$1.35 Total:$1.35-$2.55M$2.55MRevenue Growth from Personalized Investment Recommend

83、ationsClient Retention(2%Client Retention(2%-4%):$10M 4%):$10M-$20M$20MUpselling(1%Upselling(1%-2%):$5M 2%):$5M-$10M$10MTotal Ancillary Benefits Over 3 Years:MinimumMinimum:$48M$48MMaximumMaximum:$96M$96MThese can be identified as bullets,or for Extra Credit,estimate the financial impact2025 ExlServ

84、ice Holdings,Inc.All rights reserved.38Example 2:ConclusionTotal Costs Over 3 Years:$790KNet Benefits Over 3 Years:Minimum:$3.1MMaximum:$6.2MQualitative Benefits:Enable AI Initiatives:Potential benefits in the range of($48-$96M)for just 3 use casesEnhanced Reputation:Increased client trustEmployee C

85、onfidence:Improved morale,productivity,and confidence in data and AIRecommendation:Proceed with the AI and Data governance initiative to achieve significant financial and strategic gainsFoundationalImpact2025 ExlService Holdings,Inc.All rights reserved.39Example 3:ImpactA bank wants to improve the b

86、ottom lineA bank is considering ways to leverage data and analytics to improve margin:1.One potential area is identified as reducing default rates of home loans.2.If they can create a more accurate and predictive default model,losses may be mitigated to improve the bottom line3.However,to achieve th

87、is goal,not only is a better prediction required,but also the corresponding business process to leverage this better model.2025 ExlService Holdings,Inc.All rights reserved.40Example 3:Identify Impact and ROI1.Understand Functional Model/Business Processes40BankLine of Business FunctionsDevelop Produ

88、ctsDevelop MembershipExtend CreditOriginate LoanPurchase LoansService LoansProcess PaymentsMitigate LossIdentify Loan RiskFormulate Action PlanExecute Action PlanRestructure LoanDrop PMICredit ConsolidationRefinanceManage LossShort SellRepossess Collateral PropertyForeclose Collateral PropertySecuri

89、tize LoansManage CollateralEvaluate Property ConditionDispose CollateralMaintain PropertyRehabilitate CollateralRent PropertyManage InvestmentsSupport FunctionsManage Human ResourcesManage Financial ResourcesSr.ExecutiveInterview FocusDirector/Manager Interview Focus2025 ExlService Holdings,Inc.All

90、rights reserved.41Extend CreditService LoansMitigate LossIdentify RiskFormulate Action PlanExecute Action PlanRestructure LoanManage LossRisk score:Propensity to defaultDelinquencyWhat to do when score exceeds thresholdTeam executes plan when threshold exceededTeam executes plan when threshold excee

91、dedAvoid if possible because impact is negative and predictableExample 3:Identify Impact and ROI2.Identify how data opportunities tie into business processes a)a)Analytics:Analytics:Create an Aggregate Risk Factor that predicts default using Advanced Analytics.Consider previously ignored but relevan

92、t risk inputs such as:industry health,economic indicators,management changes,market share,legal issues,loan terms,M&A,Analyst input,etc.b)b)Business Process:Business Process:Establish corresponding action plan(business process change)depending on scenarioExample:Example:high risk loan is delinquent.

93、Contact borrower to verify information and when necessary,consider ways to remediate issues e.g.,restructure credit2025 ExlService Holdings,Inc.All rights reserved.42Example 3:Identify Impact and ROIBuild the case:a.Metric to track:Total Annual Lossb.Baseline value:450$Mc.Estimate data impact range:

94、LowHigh10%40%90%accuracy range by Bus Process Ownerd.Monetize:45 180 Directional$range of impacta.Cost of data effort:1.2$M-Initial and Supportb.Cost of BP change:0.3$M-Training,infrastructure,OCMc.Total cost:1.5$MROI2,900%-11,900%within 90%confidence range3)Value4)Cost5)Cost-BenefitClosing Thoughts

95、 Q&A2025 ExlService Holdings,Inc.All rights reserved.44Closing ThoughtsData is at the core of most initiatives including GenAI use casesAnything can be measuredValue Is the language of business,often framed in financial terms.AI isnt just a shiny object;how can it impact value to your organization?Are there any projects or foundational investments that you might be struggling to justify?RecapQuestions/ScenariosDave CroleneVice President Data,Analytics and AI