Transcription
Predictive AnalyticsExamples of Real Use CasesNicolas Plocharski, SAPMay 30, 2018PUBLIC
Competing in today’s marketplaceArtificial intelligence“Early evidence suggests that AI can deliver real value to serious adopters and canbe a powerful force for disruption. Early adopters are already creating competitiveadvantages, and the gap with the laggards looks set to grow.”- McKinsey Global InstituteMachine Learning“The risk of investing too late in smart machines is likely greater thanthe risk of investing too soon.”- GartnerData Insights-Driven Business“Insights-Driven Businesses Will Steal 1.2 Trillion Annually By 2020.”- Forrester 2018 SAP SE or an SAP affiliate company. All rights reserved. ǀ PUBLIC2
Enterprises need to become Data-DrivenLeveraging alltypes of dataApplying Machine Learningacross the enterpriseAutomatingdecision makingWhy now? Massive computing power (in-memory & distributedcomputing) Big data available for training models Declining hardware and software costs IoT/device connectivity 2018 SAP SE or an SAP affiliate company. All rights reserved. ǀ PUBLIC3
What makes being Data-Driven so difficult?Massive Amountof DataAnalyticalSkill GapConversationsShortage of 140K to 190K deepanalytical skillsBelief That“It’s Too Hard” We don’t have enoughData Scientists We don’t know where to begin1.5 million managers analystswho know how to use big data tomake effective decisions will beneeded We believe that it is toocomplex and challenging toeven get startedSource: McKinsey Global InstituteTransactionsMachines 2018 SAP SE or an SAP affiliate company. All rights reserved. ǀ PUBLIC4
Machine Learning AutomationiThe fastest way to becomea Data-Driven business 2018 SAP SE or an SAP affiliate company. All rights reserved. ǀ PUBLIC5
What is Machine Learning?Machine Learning consist in extracting information from existingdatasets describing past events to determine patterns, predictfuture outcomes, trends or detect anomalies.Predict - Forecast - Classify - ClusterThe objective of SAP Predictive Solutions is to make business decision-making processesmore performing and more efficient for more business value 2018 SAP SE or an SAP affiliate company. All rights reserved. ǀ PUBLIC6
SAP Predictive Analytics enables Machine LearningautomationAutomating theend-to-end processEnabling everyoneto build modelsEnsuring the outputs areready to be consumed bybusiness users 2018 SAP SE or an SAP affiliate company. All rights reserved. ǀ PUBLIC7
SAP Predictive Analytics customers are winningCox Communications: Businessusers are creating predictivemodels to supercharge customerRelationshipsReducing Driver Turnover andCreating a Safer Workforce byautomating end to end process14%80%15%More productsper customerhouseholdReduction inmodel creationtimeReduction indriver turnover infirst year28%42X 6Reduction incustomer churnrateGreaterthroughput forcentral analysts 2018 SAP SE or an SAP affiliate company. All rights reserved. ǀ PUBLICMonths to ROIpayoffDelivering personalized bankingexperiences for 5.5 million people withinsights in the hands of businessusers400% 250%Higher hit rate for nonmortgage loansLarger hit rate forsavings products200%Increased hit rate forinsurance products8
How SAP Customers are solving business problemsSALES MARKETING Churn ReductionCustomer AcquisitionLead stomerSegmentationNext BestOffer/ActionFRAUD RISKOPERATIONS Predictive MaintenanceLoad mmendationManufacturing ProcessOpt.Quality ManagementYield Management Fraud and AbuseDetectionClaim AnalysisCollection andDelinquencyCredit ScoringOperational RiskModelingCrime ThreatRevenue and LossAnalysisFINANCE HR Cash Flow and ForecastingBudgeting SimulationProfitability and MarginAnalysisFinancial Risk ModelingEmployee RetentionModelingSuccession Planning25 IndustriesAerospace &DefenseAutomotiveBankingChemicalsHigher Educationand ResearchHigh r ProductsDefense & SecurityEngineering,Constructions &OperationsHealthcareLife SciencesMediaMill ProductsMiningOil & GasProfessionalServicesPublic SectorRetailSports andEntertainmentTelecommunicationTransportation &LogisticsUtilitiesWholesalesDistribution 2018 SAP SE or an SAP affiliate company. All rights reserved. ǀ PUBLIC9
Exciting Story, but we want some real life examples 2018 SAP SE or an SAP affiliate company. All rights reserved. ǀ PUBLIC10
KXEN Helped Driving a Smarter Obama 2nd Campaign Needed speed to scale to huge and rapidly changingonline and offline data Allowed OFA to run campaigns through email and socialmedia sites (FB, Twitter) Optimized fundraising campaigns for highercontributions Identified “persuadable” segments of votersObama For America(OFA) CampaignThe timely and accurate insights provided by KXEN led to more effectiveand quicker targeting and in the end, more votes.- Rayid GhaniChief Scientist at Obama for America 2018 SAP SE or an SAP affiliate company. All rights reserved. ǀ PUBLIC11
The Goal: Target Voters and VolunteersVisits of VolunteersTV AdvertisingFundraising emailsSocial Campaign 2018 SAP SE or an SAP affiliate company. All rights reserved. ǀ PUBLIC12
State of Indiana: Using SAP HANA to CombatInfant Mortality and Save Hoosier LivesOrganizationState of IndianaHeadquartersIndianapolis, IndianaIndustryPublic sectorProducts and ServicesLegislation, regulation,and citizen servicesEmployees28,000RevenueUS 30 billion bienniallyWeb Sitewww.in.govPartnerKSM Consultingwww.ksmconsulting.comObjectives Use data to drive program and budget decisions and improveoutcomes Better understand the state’s higher-than-average infant mortality Find new ways to analyze data to combat Indiana’s most pressingissuesWhy SAP World-class in-memory and predictive analytics technology Partnership with KSM Consulting, experts in government efficiency,Big Data architecture, and predictive data science, for immediate andlasting successResolution Used the SAP HANA platform with SAP Predictive Analytics andSAP Lumira software as part of the Management and PerformanceHub (MPH) initiative to examine infant mortality in new ways Developed sophisticated algorithms to identify at-risk subpopulationsand provide actionable insights for policymakersFuture plans Continue researching infant mortality risk to drive more positive birthoutcomes and develop a mobile app to assist mothers Use the MPH with other state agencies to combat fraud, criminalrecidivism, child abuse, and more9 billionRows of data analyzed for theinfant mortality use case15 data setsCombined for integrated analysis3 key findingsIdentified, including thepopulations at greatest risk forinfant morbidity, enabling targetedmarketing campaigns based onthese high-risk subpopulations 13.5 millionIn new budget proposed forapplication development andnew programs to combat infantmortality“SAP software and KSM Consulting have allowed us to pinpoint the problem bylocation and by subpopulations so we can get resources to the women who needthem. We’ve brought together disparate agencies and their data in this collaborative,innovative environment to develop a targeted solution.”Sara Marshall, Director of Business Intelligence and Analytics, State of Indiana 2018 SAP SE or an SAP affiliate company. All rights reserved. ǀ PUBLIC36975 (15/04) This content is approved by the customer and may not be altered under any circumstances.13
Predictive Projects at French PoliceBusiness ObjectivesPredictive Maintenance: with a growing-old vehicle fleet and shrinking budget, it’scrucial to avoid useless maintenance, to reduce failures and breakdowns whilemaximizing the use of each vehicleRH Turnover: being part of the National Gendarmerie could become tough from time totime, therefore the National Gendarmerie wanted to identify resignations’ causes toprevent them and take actions to reduce the officers’ pain pointsOperational deployment: by identifying criminality’s causes, officers are able to obtainmapping for different crimes’ types, depending on location, hour, etc. They also canoptimize force allocationCurrent ProjectsBusiness Intelligence : « Visualization of Police activity statistics»Predictive Analytics : « Optimizing the allocation of vehicles toterritories »Predictive Analytics : « Anticipation of crime evolution »Predictive Analytics : « Predictions for car thefts in the Oise province »Predictive Analytics : « Forecasting the early departures for military and civilianemployees »Innovation : « Sentiment analysis for Gendarmerie Nationale e-reputation tracking » 2018 SAP SE or an SAP affiliate company. All rights reserved. ǀ PUBLIC14
Machine Learning in Banking – a long story400%Increased response rate of Marketing campaignsBank of Montreal7x increase of response Lloyds Bank reducedIncrease of response rateby 160% and purchasesrate compared to control modelling effort fromby 35%groupdays and weeks to hours 2018 SAP SE or an SAP affiliate company. All rights reserved. ǀ PUBLIC15
10 X Improvement in Modeling ProductivityGoals Improve modeling efficiency Meet modeling needs with no additional employees Learn if new data sources improve modelperformance (up to 50,000) Respond rapidly to market changesSolution – SAP PA Implementation Allows analysts to consider tens of thousands of variablesReduces manual mistakes like dropping a variable too earlyin the processAllows expert analysts to focus on value-added tasks likedeveloping new variables & segmenting customersdifferently 2018 SAP SE or an SAP affiliate company. All rights reserved. ǀ PUBLIC16
10 X Improvement in Modeling ProductivityResults 10X improvement in productivity without compromising model accuracyEnd to end model creation has decreased from 6 months to less than 1 monthUsing traditional statistical methods it took 10 staff days to identify the best 10 variables.Using SAP PA it takes 1 staff dayResults can be used as-is or fed into further analytic effortsModel kindNum.variablesKS, traditionaltoolsKS, SAP 13001110marketing13003540 2018 SAP SE or an SAP affiliate company. All rights reserved. ǀ PUBLIC17
POCKET CARD: Doubling Revenue from DirectMail Campaigns with SAP Predictive AnalyticsCompanyPOCKET CARD Co., Ltd.HeadquartersTokyo, JapanTop objectives Increase revenue by promoting cash advance services and revolving creditand by streamlining marketing calls for affiliated insurance services Reactivate nonactive existing cardholders and find a new customer basesuitable for specific services180%Boost in the sales conversionrate of telemarketing calls forinsurance servicesIndustryBankingProducts and ServicesFinancial services, includingcredit card, loan, and insuranceagency servicesEmployees350Revenue 34,174 million(US 312.8 million)Web Sitewww.pocketcard.co.jpResolution Adopted the SAP InfiniteInsight solution for predictive modeling and dataanalysis to gain customer insight Enabled data-driven customer segment targeting decisions for eachcampaign, rather than relying solely on marketer experience Enabled the analysis of enormous amounts of data such as monthly creditcard statements, outstanding loans, and repaymentsKey benefits Allows analysts to simulate infinite scenarios and select the optimal models fortheir business needs Enables the building of predictive models from millions of customer recordsand tens of millions of historical account transactions within a few hours200%Increase in total revenue fortargeted direct mail, comparedwith mailing to lists of nonactivecardholders compiled withoutmodels400%Higher conversion forpromotional campaigns for cashadvance and revolving credit withpreferential interest rates to newcustomers“The total turnover for targeted direct mail campaigns doubled, thanks to SAPInfiniteInsight.”Kuniharu Takenaka, Manager, Sales Planning, POCKET CARD Co., Ltd. 2018 SAP SE or an SAP affiliate company. All rights reserved. ǀ PUBLIC33863 (14/11) This content is approved by the customer and may not be altered under any circumstances.18
mBank: Delivering a Personalized Banking Experience for5.5 Million Customers with SAP Predictive AnalyticsCompanymBank S.A.HeadquartersWarsaw, PolandIndustryBankingObjectives Respond to customer needs as quickly as possible Gain insights into customer preferences to support a customer-centric banking experienceacross all channels Optimize the company’s discount program (mDeals) by providing better service to partnersand more targeted offers to customers Improve the performance of marketing campaigns by better understanding customerbehavior and anticipating future demandWhy SAPProducts and Ability of SAP Predictive Analytics software to drive a close, personal connection to clientsServicesby providing a data and analytical modeling toolRetail and corporate Pragmatic, user-friendly software, allowing quick user adoption and rolloutbanking products and Easy integration with the existing IT infrastructureservices, wealthmanagementEmployees6,318Web Sitewww.mbank.plResolution Predictive modeling based on transactional and demographic data Precise segmentation, advanced reporting, and execution of real-time, multichannelmarketing campaigns Context-specific offers across all channels based on customer profilesRapidIncrease in responserates to marketingcampaigns400%Higher hit rate fornonmortgage loans200%Increased hit rate forinsurance products250%Higher hit rate forsavings productsBenefits Reduce churn and grow revenue by automating campaigns and offer activation Increase sales efficiency and reduce the cost of sales Minimize the effort to train employees and identify targeted customers Empower mBank’s discount program by personalizing discounts for each customer based onpredictive insights 2
Competing in today’s marketplace “The risk of investing too late in smart machines is likely greater than the risk of investing too soon.” - Gartner “Insights-Driven Businesses Will Steal 1.2 Trillion Annually By 2020.” - Forrester “Early evidence suggests that AI can deliver real value to serious adopters and can be a powerful force for disruption. Early adopters are already .
