Suncorp CDI-MDM
Implementation In The
Wake Of A Merger
MDM Asia-Pacific Summit Sydney April 2009 www.mdmsummit.com.auGlenn Mead IT Architect Monica Smith CRM Architect
Australia’s 6th largest Bank; 3rd largest General Insurer; significant Wealth
Management business
Merger & Acquisition activity
• Promina 2007 • GIO 2001
• Suncorp / Metway 1997
•M&A activities are a constant challenge for integration platforms such as CDI (or Data Warehouse, GL etc)
•Suncorp & GIO had a long history with in-house CDI systems, but all were end of life / unable to expand for new customer base / requirements
Existing CDI / MDM Capability
Suncorp
• CDI capability (One brand Suncorp across Banking, Wealth and Insurance) • Standardise and match customer records
• Provide customer services to support Sales and Service process • Provide basis for CRM capability
GIO (M&A)
• CDI Capability (Multiple brands Insurance only) • Standardise and match customer records
• Provide customer services to support Sales and Service process
Promina (M&A)
• No CDI capability (Multiple brands across Insurance and Wealth)
Customer Analytics (Group)
Initial Drivers from Suncorp / Promina Integration
The CDI Hub, providing a view across the entire group = data to enable: Analytics & Marketing
• Understand group customer base, Brand cross-over • Limit cannibalisation
• Single marketing platform
Customer Service:
• Support for existing business processes after system rationalisation
•Short-term outcomes - aligned to medium/long-term Customer & IT roadmaps. •Having these roadmaps in place & socialised was crucial to funding when (M&A / Integration) opportunity arrived.
360 Degree view of customer
• Group-wide view for analytics / marketing
• Allow expansion to operational CRM / customer service Multiple brands
• Provide operational views by brand
• Understand customer holdings across brand House-holding - Properly understand share of wallet Data Quality Reporting
• Improvement on Customer Service / Understanding • Pin-point problem systems / monitor DQ over time Customer data partitioning / protection, e.g:
• Portfolio Protection (products sold through brokers / channels) • Protection of ‘Joint Venture’ Customer Data
External Prospect Lists Customer relationships
A comprehensive uplift from ‘current’ capability is being put in place.
SOURCE
SYSTEMS CUSTOMER (CDI) HUB CRM (CFDM)ANALYTICAL OPERATIONAL CRM
Data silos Quality Controller
The Brains The Playmaker
Siloed storage of customer, product and
transactional data. No single view of the
customer exists
Verifies and manages data quality from all source
systems at an enterprise level
Verifies and manages data quality from all source
systems at an enterprise level
Helps to understand and anticipate the needs of
current and potential customers for tactical and strategic decision
making
Automated support to front line business processes
including customer contact (sales, service
and marketing).
CDI Context
The joys of having Marketing as your partner / customer => no complex system diagrams allowed!
Product Selection
Source: Gartner 2008
IBM
• Full featured CDI Hub with extensive data model / large number of pre-built services
Oracle Siebel UCM
• Extensive financial services data model • Best suited to Siebel CRM sites
Siperian
• No Australian presence at time of selection
Initiate
• ‘Best of Breed’ vendor offering fast & advanced matching, quick start-up
• Australian presence expanding
• Background is Health Services – Single View of patient (hence strong in matching / client
Suncorp chose Initiate as a partner with the right engagement model and people for a mutually successful project.
Key Focus Areas for Selection:
• CDI specifically (not wider MDM) • Fast implementation
• Flexible licensing model to suit the Suncorp implementation model
Learnings:
• Implementation required more ‘lines of code’ than expected – would have structured project differently if this was understood.
• Initiate partner products (Intech, Clover) caused more challenges than expected.
• 2.2m records standardised, matched & loaded < 3 hours
• Data model / Web services extended
• ‘CRM-ish’ front-end connected in < 2 weeks
• ETL extract developed suiting Marketing requirement • Helped clarify roadmap & eased many concerns:
• Supporting various project timeframes & requirements • (Relatively) simple transition from existing systems
POC Highlights (Initiate)
Learnings from our POC:
• Initiate model had many advantages over existing CDI systems (brand protection, source data retention, flexible matching algorithm).
• Many puzzles remained including: what to keep in the Initiate model versus ‘alongside’?; will CRM searching requirements be satisfied?
Initial Phase
•24 sources
•20 million customer records
•2 existing CDI Hubs to ‘synchronise’ with
•Feed to marketing data mart •External data (prospect lists)
Some learnings on the period between POC and project start:
• Needed a longer planning / decision making period before initial project phase.
• Some project technology choices & scope decisions rushed / impacts not fully realised.
Technology Set
• Initiate Matching algorithm, Data Model, Framework • Intech Standardisation (name, address, email, phone) • Clover ETL (data transformation)
• Java (business logic, called by Clover)
• Red Hat Linux (App Server – note CPU intensive activity) • Oracle – DBMS (note, Initiate makes little use of many
Relational DBMS features) • Tableaux – Auto-Deployment
Notes:
• First Suncorp usage of Linux for major CPU intensive data manipulation
• Auto Deployment software critical to management of multiple environments with complex software sets
• Difficult to get business ownership for customer data / required (informed) involvement – BUT Crucial to success
- Project would have failed due to IT politics without Business Support (and funding)
- Customer Roadmap socialisation prior to opportunity made funding / direction decisions far easier
- Business representatives on the project will help drive further adoption
• Tangible business objectives, in terms of data quality metrics, have been crucial to getting business agreement on CDI Hub success
Learnings
• Have plans in place, to enact as opportunities arise
• Agree target data quality metrics upfront and focus ‘testing’ on measuring against these metrics.
• Need to have right resources (e.g. data analysis)
• Need appropriate testing methodology (data management versus functional testing)
• Difficult but crucial to get business agreement on ‘requirements’: - Data Sourcing / Target Data Model
- Data Quality Metrics
- Functional requirements
What is working for us – Split functional development from data quality improvements (matching / standardisation) – separate projects with differing resource
requirements / timeframes etc).
Value - Data Quality Reporting by Source
Value
• Provide summary reporting and details of DQ levels / errors for each source.
• Understand where the real issues are, allow focused resolution • Get early warning of DQ degradation
Enablers
• Initiate data model (retention of source data) • ‘Data Lineage’ information
One of the highest value differences with our Initiate implementation is provision of regular data quality monitoring by source.
Key DQ metrics are reported by source as a whole and by time period (understand DQ trends), and compared with the whole base.
So - Where Do We Want To Be?
SOURCE
SYSTEMS CUSTOMER (CDI) HUB CRM (CFDM)ANALYTICAL OPERATIONAL CRM
Data silos Quality Controller
The Brains The Playmaker
Siloed storage of customer, product and
transactional data. No single view of the
customer exists
Verifies and manages data quality from all source systems at an
enterprise level
Verifies and manages data quality from all source systems at an
enterprise level
Helps to understand and anticipate the needs of
current and potential customers for tactical and strategic decision
making
Provides automated support to front line business
processes including customer contact (sales,
What does this mean?
• Integration into Customer Analytics Roadmap /Marketing Roadmap • Create Services layer to support Operational Systems
• Integrate to CRM Front-Ends
• Real-Time feeds to / from some Source Systems • Decommissioning:
• Legacy Hubs
• Data Warehouse feeds • Source system feeds