BI Transformation & Innovation Jeroen Dijkxhoorn MSc Global Technology Practice SAS Ir. Edwin Peters Solution Director SAS Netherlands
Agenda Definitions of BI Drivers of BI Transformation and Innovation 5 evolutionary BI Models Examples of application How to control the evolution
What is Business Intelligence in this context? Generic Insight into the operations of a business/organization A system which turns data into information Three software market categorizations End-user Query and Reporting tools (EUQR) Suite of EUQR tools + data integration Above + data management + advanced analytics = Enterprise Intelligence
Drivers of BI Transformation and Innovation: The CIO as business strategist The CIO role is changing CIO s are now expected to deliver the solutions that make the enterprise different in a way that matters to company performance and customer satisfaction Mark McDonald, VP Gartner Executive Program group This is changing the influence of CIO s in enterprise software decisions as well as required skills and external relationships Future State CIO Model. Source: C level Competencies Report, CIO.com
Drivers of BI Transformation and Innovation: IT as enabler of Improvement & Innovation Alignment of Organization Strategy and IT Strategy Reflected in budget Multi-year planning Strong adoption of new Technology Monitoring use of IT in the Industry Source: McKinsey Quarterly
Drivers of BI Transformation and Innovation: CIO 2008 Priorities Top 10 Business Priorities Ranking Top 10 Technology Priorities Ranking 1. Business process improvement 2. Attracting and retaining new customers 3. Creating new products and services (innovation) 4. Expanding into new markets or geographies 5. Reducing enterprise costs 6. Improving enterprise workforce effectiveness 7. Expanding current customer relationships 8. Increasing the use of information and analytics 9. Targeting customers and markets more effectively 10. Acquiring new companies and capabilities 1. Business intelligence applications 2. Enterprise applications (ERP, CRM, etc) 3. Servers and storage technologies 4. Legacy modernization, upgrade or enhancement 5. Technical infrastructure 6. Security technologies 7. Networking, voice and data 8. Collaboration technologies 9. Document management 10. Service-oriented architecture (SOA) and service-oriented business applications (SOBA) Source: Gartner EXP (January 2008)
Drivers of BI Transformation and Innovation: Changing IT function From Technology to Process oriented IT innovation From De-central to Central IT coordination 4 quadrants of IT Function evolution with distinctive challenges: 1. Modernization 2. Integration 3. Consolidation/Standardization 4. Partnering Technology 3 Central/ Integrated De-central/ Specialized 4 1 2 IT Function Evolution Quadrants Process
Agenda Definitions of BI Drivers of BI Transformation and Innovation 5 evolutionary BI Models Examples of application How to control the evolution
1. Classic BI Architecture Data Flows managed by ETL tool 1 (------) Data Flows managed by ETL tool 2 (solid red) Analytical Data Flows Results Flows Consumers are Humans Useful for Monitoring Processes Provides Consistency and Management
Rijkswaterstaat Netwerkmanagement informatie Systeem (NIS) Rijkswaterstaat beheert Hoofdwegennet, Hoofdvaarwegennet en Watersystemen Uitdaging: Corporate denken & doen Sneller beslissen Continuïteit, integriteit, transparantie Oplossing: SAS Enterprise Intelligence platform / GIS systeem Resultaat: Inzicht in de kwantiteit, kwaliteit en prestatie arealen
2. Classic BI Architecture with Data Quality Data Quality added Various Possibilities in Combination Real Time DQ in Operational Applications DQ added in Flows to EDW or to Marts One Set of Rules Advanced Analytics Engine Model Repository DQ Managed Analytical Data Marts Reporting & Analytics Engine Report Repository Data Warehouse Operational Data Store Human Receiver DQ DQ DQ Operational Application Databases DQ Operational Applications
HDFC Bank India HDFC Bank, one of the leading private sector Banks in India with a customer base of 6.5 M and awarded Best Retail Bank of India 2007. Challenge Incompleteness and discrepancies in customer data Could not support a more scientific marketing approach Better data quality needed for credit checks Increased compliance on customer data accuracy Solution Both online (point of sale) and offline (data warehouse) data quality Results Able to effectively manage customer campaigns Reducing marketing spend by better targeting Enable online credit worthiness verification Improved (speed of) customer service
3. BI Architecture with Feedback Results fed back into Operational Systems Integration at the Data level Advanced Analytics Engine Model Repository Reporting & Analytics Engine Report Repository Human Receiver Useful for Cyclical Applications eg Planning Managed Analytical Data Marts DQ Results of Forecasts, Cost Models or Optimisations DQ Data Warehouse Operational Data Store Users see same workflow but with better recommendations DQ Operational Application Databases DQ DQ Operational Applications
BI Feedback Loops Toekomstmuziek of realiteit?
Werkverdeling SRB (Interpolis) Probleem: Iedere dag handmatig ± 400 zaken over 200 mensen verdelen 5 rechtsgebieden op 4 locaties 40 manuur per dag om handmatig te verdelen Onvrede over kwaliteit eindresultaat, bv behandelaar die niet aanwezig is, krijgt toch een zaak toegewezen Ontevreden klant!
Werkverdeling SRB (Interpolis) Functionele inventarisatie: Indeling 1200 soorten zaken (fijn mazige indeling) DWH gebruikt om doorlooptijd / bestede tijd te meten per zaaksoort DWH gebruikt om normering zaken vast te stellen Alle factoren die van belang zijn in kaart gebracht en vastgelegd: - kennis / ervaring medewerkers - hoeveelheid openstaande zaken per medewerker - aanwezigheid (werktijden, vakantie, afspraken, ziekte) - automatische uitbesteding externe partijen als intern geen capaciteit meer is (contract gegevens geparametriseerd)
Werkverdeling SRB (Interpolis) IT Oplossing: Specificaties uitwerken Interface gebouwd in operationeel systeem Optimalisatie engine in SAS, omdat: correctheid, onderhoudbaarheid, performance (!!) Omgeven door integriteit controles Resultaten: Verdeling nu ± 2 uur per dag (was 40 uur) Match medewerkers (rekening houdend met aanwezigheid, kennis, werkdruk) zaken vrijwel optimaal Onderzoek gedaan door externe expert IT oplossing optimaal
Voorbeeld van BI met Feedback Loops Ontwikkeld door Maarten Zaanen van PZvK http://www.pzvk.com/
4. Real Time BI Architecture Integration at a Service Level between Operational Application and Model Case Manage ment System Advanced Analytics Engine Model Repository Reporting & Analytics Engine Report Repository Human Receiver Real Time Scoring or Decisioning Managed Analytical Data Marts Must Satisfy Throughput Requirements Services Data Warehouse DQ Useful when Score on Demand is required DQ DQ Operational Data Store DQ May need to marshall data from operational system Operational Application Databases Services DQ Operational Applications
Voorbeeld Realtime BI Retail Bank met Credit Cards Programma: Kostenreductie door fraude detectie at swipe time Mainframe CICS systeem voor Credit Card transacties Resultaat moet snel beschikbaar zijn enkele Milliseconden Uitkomst Ja, Nee of Refer (doorsturen naar case systeem voor beoordeling door individu) Toepassing moet ook recente data gebruiken die nog niet in het DWH zit
In a Fraction of a Second Shop System Bank System Shop System Bank System
5. BI with Business Activity Monitoring Events flow out of Operational Systems Combinations of Events are Analysed Alerts Generated after Analysis (Rules) Alerts Return to Operational System Alerts also to Human Receiver Case Manage ment System DQ Services Advanced Analytics Engine Model Repository Alerts Engine DQ Reporting & Analytics Engine Report Repository Rules Repository Managed Analytical Data Marts Data Warehouse Operational Data Store Operational Application Databases Human Receiver DQ DQ Services Services DQ Operational Applications
Mid region US Bank Leadership Dashboard A. Something going on in originations B. Standard Alert Button for SLA s A. Loans are not closing as expected. B. Pipeline stronger than expected not pressing panic button - yet C. Abandoned sales costing us more than expected D. Idle Capacity is being managed E. Production delays are costing us money Alerts will tell us more. F. Production delays and larger pipeline are putting pressures on standard costs A A B C D E F B
Mid region US Bank Leadership Dashboard Standard Costs are unfavorable Drilling into Sample Credit Check area
Mid region US Bank Operations Dashboard Tailored to Operations Managers Top Levels with Drilldown X in Pre-Approval indicates near real-time problem Visual Process View
Mid region US Bank Dashboard Drilling into Pre-Approval Near Real Time Alert drills into detail Problem is with Credit Checks
Mid region US Bank Operations Dashboard Into Alerts for Credit Check Problem is communications links to Credit Agencies are down, this could be communication, but we can quickly estimate loss The key element is that we can calculate impact of the alert in near real-time. Shows which Mortgages are impacted
Mid region US Bank Business Impact of Key Performance Metrics Operations must meet internal SLA s; otherwise, there are internal fines and charges are levied. Not only can we see counts and trends of SLA violations. Because Process Monitoring and Performance integrate with each other, we can quickly compute the financial impact in near real-time
Agenda Definitions of BI Drivers of BI Transformation and Innovation 5 evolutionary BI Models Examples of application How to control the evolution
The BI evolution without a revolution requires a platform approach to enable all models
Enabling IT Platforms All serving a different purpose Productivity: Largely owned by MSFT and includes email, word processing etc. Operational: The battleground of SAP and Oracle. Is about automating and standardizing business processes to gain maximum efficiency. Intelligence: Is about uncovering information that can be used to derive intelligence to help an organization innovate, be more effective and further improve efficiency.
SAS Enterprise Intelligence Platform Solve it today. Evolve it tomorrow SAS Enterprise Intelligence Platform Allows you to build all strategic roadmap solutions on top of a (pre) integrated platform which provides industry leading Data Integration, Data Management, Analytics and Business Intelligence Key Requirements: Usability Scalability Manageability Interoperability IT Benefits: 1. Consolidation and Standardization Reduce TCO and Risk Integrate systems and processes 2. IT Governance Secure information assets Enforce information integrity 3. Deliver Strategic Solutions Align IT and organization goals Improve internal user satisfaction
Focusing on Appropriate Initiatives Information Evolution Model Five Levels of Evolution Level 5: Innovate - Expand top line Level 4: Optimize - Optimize bottom line Level 3: Integrate - Enterprise view Level 2: Consolidate Departmental silos Level 1: Operate - Individual focus Four Critical Dimensions 1. Human Capital 2. Knowledge Processes 3. Culture 4. Infrastructure
BICC Functional Areas & Role Examples E.g. Executive Sponsor, Business Unit Managers E.g. Business Unit Managers, IT Manager E.g. Chief Data Steward, Data Quality Specialist E.g. Training Consultant, Instructors E.g. Data Manager, Business Specialist E.g. Data Miner, Business Analyst E.g. Warehouse Consultant, Warehouse Architect Application Developer E.g. IT Manager, Application Developer, Business Analyst,
Summary The role of IT and the CIO is changing Information processes are at the base of further efficiency improvement, competitive differentiation and business model innovation BI Evolution requires an appropriate enabling IT platform Take a structured approach to the initial transformation at all levels Technology Methodology Support organization
Size Optimization Planning / Ordering / Allocation 32
Size Profiles Women s s Department Current Size Distribution Actual Demand 36
Cluster-based Allocations Average Extra Small Medium Med/Large Large versus One scheme fits all 39
Size Profile Clusters Women s Department - Example Before Size Optimization: Small Med Large Xlarge Total # Stores 10% 26% 33% 31% 100% 757 With Size Optimization - Actual Customer Demand: XSmall 13.1% 29.3% 30.6% 27.0% 100% 55 Small 10.8% 28.7% 33.4% 27.1% 100% 101 Medium 10.1% 23.9% 35.7% 30.3% 100% 99 Large 9.8% 24.4% 35.0% 30.8% 100% 117 VLarge 10.3% 23.6% 31.1% 35.0% 100% 113 XLarge 8.3% 24.8% 30.1% 36.8% 100% 41 Average 10.1% 26.4% 32.6% 30.9% 100% 231 38