Building a Business on Data: Challenges and Rewards Naras Eechambadi - PowerPoint PPT Presentation

Building a Business on Data: Challenges and Rewards Naras Eechambadi and Kurt Newman March 27, 2019

Today’s Presentation • Business Opportunity • The Data • Markets & Use Cases • Challenges • Solution • Results

Business Opportunity New Revenue Large Market Rich Data A business is born

Today’s Presentation • Business Opportunity • The Data • Markets & Use Cases • Challenges • Solution • Results

ADP is one of the leaders in Payroll and HCM solutions

Paychecks are the source of ADP data EMPLOYER ADDRESS EMPLOYEE ADDRESS

Paycheck: Data Anonymized & Aggregated

Reported by Geographic Location Layers Census Blocks ~11.2M Block Groups ~211K Census Tracts ~74K Counties ~3200 States 50 Divisions 9 Regions 4 Nation 1 Zip Codes & MSAs

Today’s Presentation • Business Opportunity • The Data • Markets & Use Cases • Challenges • Solution • Results

Market Problem & Opportunities Use C e Cases es Verticals Macr cro E o Economic S c Strategy: Capital Markets Research Macro view of the US in context of net migration, employment income, demographics, industries and job types. Micr cro Econ onom omic S c Strategy: Real Estate Validates site selections. Identifies emerging and distressed areas. Multiple Verticals Provides ability to identify, segment and target populations at a local level. Retail B l Bank nking ing: C Competit itiv ive Environment Banking Ability to view “direct deposits” via share of wallet, determined by pay check deposits.

Today’s Presentation • Business Opportunity • The Data • Markets & Use Cases • Challenges • Solution • Results

Problem Statement We n e needed ed a an automated proces ess for data set et c crea eati tion, v validation a and d deliver ery to to c clien ents. The p e process w was r req equired to t to support: t: • Rapid i iter erati tive f file e prep eparati tion f for c clien ents t that n t need eed t to e evaluate m multi tiple d data format mats. • Sched eduled ed deliver ery o of files c clients ts n need e each m month/week eek.

Today’s Presentation • Business Opportunity • The Data • Markets & Use Cases • Challenges • Solution • Results

Automated and Scalable Data Environment Bui uilt a a hi highl hly a aut utomated a d and nd scalable V Ventures Data Environment t to ens nsure effic icient hi high h qua quality da data de delivery Launched infrastructure that scales to ensure efficient sales fulfillment • Increased data security (limited access) Current Ventures Data State • Automation increases efficiency for data set preparation and New Ventures delivery Environment Data Cloud Production • Scalability to support growth • Incorporation of 3 rd Party data Implementation of a highly automated and scalable environment for • Predictable, rapid responses to client needs quality control, automation and scalability launched Q2-FY19 • Automation reduced process time and the probability of errors • Dedicated Ventures Reporting & Analytics environment • Many data files can be created, fully validated and delivered to • Secure ADP instance of AWS-hosted most clients in about 90 minutes environment (with Quaero CDP Platform)

Quaero CDP Architecture • CDP automates data processing and creates data assets • Data assets are used in client extracts, analytics and Explorer • CDP auto scales compute and storage based on data volume and processing need • Role based permission is enabled within applications and database layer

Data Monetization is a multi-step process Adjustment, analysis etc Data Received, Data Cloud Ventures team Commonly requested Client uses Sources Discovered, team models Collects, Analyzes, aggregations created for Data, Revenue provide data Analyzed by and publishes and Validates “Standard Files” Booked Client Iteration and refined data requests Client Specific Data Set Requirements Include: Frequency Filters Aggregation Fixed Panel Distribution

Data Processing and Client Extract Process 4 1 2 3 5 6 Yes Data Aggregate Automated Automated passed the Tables for Data Delivery Transformation Validation of Data Assets defined “standard via MFT Incoming Data thresholds files” extract No Process Stops 9 11 8 10 7 Data Yes Automated File Automated Configure Automated Extract passed the delivery to Clients notification to Client Extracts File Validation defined via MFT and ADP team thresholds No Process Stops

Today’s Presentation • Business Opportunity • The Data • Markets & Use Cases • Challenges • Solution • Results

Identify Work & Residence Populations Age Where Employees Come From Where Employees Go To Work Tenure Industry 9% 14% 11% 5% Commute 23% 27% 21% Profession 8% 8% Income 32% 18% Change

Share of Wallet View c compe petitiv ive de depo posit l lands ndscape a across t the he U US us using ng relia iabl ble ADP pa pay c che heck de depo posit it a and nd pa payroll ll da data $HARE OF WALLET Deposits tracked monthly: • Financial Institutions: Banks, Credit Unions • Total deposits (dollars & paychecks) Demographics tracked monthly: • Income, age, gender & generation type Data extract visualization using Tableau Data Aggregated: • State, County, City & Zip Code Share of Wallet Measurement & Trends: • Total dollar deposits • Total paycheck deposits • Top Five Banks Application tool for data analysis

Data Grain Avail ilable at the he level of de depth h requ quir ired a and nd on a n a monthly ly ba basis 1 Nation United States States MSAs Counties ZIP Codes Census Blocks * (380) (3,200) (32K) (11.2M) (50) bctcb2010: 10125001008 boro_code: 1 boro_name: Manhattan cb2010: 1008 ct2010: 012500 share_area: 31542.5183224 share_leng: 769.081961398 New York Metropolitan New York County 10036 1008 New York Area/Tri-State Area (Manhattan) (New York, NY) (Times Square) * For statistical purposes and graphical representation, the Census Bureau’s ZIP Code Tabulation Areas are used. ** ADP requires a minimum number of employees and employers to populate data for the next geo-layer. *** Census blocks hierarchy includes census blocks, block groups, and census tracts.

Predict Case Shiller Change Over the Next 12 Months • An initial model using ADP data to predict Case Shiller Index changes over the next 12 months. • The observed values (blue dots) and predicted values (orange dots) are shown for Case Shiller MSAs.

Summary/Conclusion Slide Data collected for operational purposes can have potential value outside the initial domain Realizing this potential requires significant transformation Having the right tools can accelerate this process The rewards are lower costs, faster and higher value realization

Thank You Naras Eechambadi naras@quaero.com Kurt Newman kurt.newman@adp.com