© 2014 IBM Corporation
How Nationwide Insurance Transformed and Accelerated its Small Commercial Insurance Process with IBM ODM (BAT-4442) Scott Grau, Nationwide Insurance P Vilas Tulachan, Open Source Edge LLC
Would you wait 3 or more days to learn if your insurance company will write your small business insurance?
Presenters
• Scott Grau, Director IT Architecture Office of the CTO - Commercial Lines & Agency
Nationwide Insurance
• P Vilas Tulachan, IBM ODM Enterprise Solution Architect
IBM Worklight Developer Open Source Edge LLC
Key Points • Key ingredients necessary to transform slow, error prone and
manual process to automated underwriting process • How to build a business centric ODM team to maximize
cooperation and results • Key business and technical risks and how to mitigate them • Why Enterprise Rule Governance should be a key ingredient of
the overall solution?
Agenda
• The Small Commercial Market and Nationwide’s Need (define the problem)
• What is success (define the goals) • The approach
• Team structure • Business involvement • Protocols • Proof-Of-Concept
• The technology • Rule governance • The results • Q & A • Final Comments
The Small Commercial Market • Conning defines Small Commercial business as less than
$50,000 premium and generally less than 30 employees • Conning: Small business market premium $58 billion (2014) • Small Commercial insurance market pressures
• “A BOP is a BOP is a BOP!”: Small Commercial is a commodity product
Need : Improved Small Commercial Capability
Key reasons for not winning business
Automation is required to compete
Nationwide’s Challenge • Dependency on manual processing
• Slow and subject to human related inconsistency • Unable to quickly respond to competitive market conditions
• Manual process negatively impacts price, adds difficulty to the quote process, and by definition includes slow response from an underwriter
Nationwide’s Need • Increase decision speed • Eliminate human related inconsistency in decision process • Respond quickly to competitive market conditions • Reduce expense
Project Success Criteria
• Automate the Small Commercial underwriting decision resulting in:
• Improved price by reducing process costs • Improve decision time for the target risk appetite • Improve consistency of underwriting decisions • Quickly and proactively respond to changing market and
competitive conditions • Simplify quoting process
Project Success Criteria
1. Automate 20% of the small commercial underwriting decisions with initial implementation
2. Business team able to author, test, and validate underwriting rules
3. Ability to deploy updated underwriting rules quickly 4. Ability to run simulation and what-if analysis in order to
optimize results
Approach – Business Team
• Need appropriate level of business sponsorship to drive change • Consider change management needs • Mix of business analysts – experience and expertise • Future rule authors and ODM administrator
Approach - IT Team • Project/Solution Architects
• ODM, Integration, UI, Infrastructure • Development team
• Mix of contract ODM and Nationwide developers – Experienced contract ODM developers mentored in-house ODM
developers during the implementation • IBM Professional Service team – Lead ODM developer – Leveraged them during the initial phase of the implementation
Approach - Business Involvement • A business driven solution • The architects worked with the business team to understand the
underwriters natural approach to evaluating a risk • Train the business team on the capabilities of the ODM • Jointly formulate the solution options
Approach - Business Involvement • The data model is key to enabling the business
• An IBM Modeling consultant was involved in getting the first cut of the business object model
• The Architect team worked with the Modeler and the business team to lay the foundation of the Small Commercial solution architecture
• Semantics matter – business user defined the terminology
Approach - Protocols
• Underwriters do not think like a computer with a series of binary true or false checks
• Adopted “Protocols” loosely based on chemistry and represented in terms of atom/molecules/compound mirroring the way an underwriter reviews a given risk
• Compound is Account • Molecules are Business Strength, Insurance History • Atoms are Financial Score, Years in business
Approach - Protocols • Individual compound gradient score
• Score assigned to each individual compound gradient based on score assigned to each molecule gradient
• e.g. Outstanding - 5, Strong - 4, Standard - 3, Weak - 2, Fail - 1 • Molecules and Compounds can be weighted to emphasize
company preferences and market trends • Risk factor outcome can be tweaked at molecule and compound
level giving business analyst control of the overall risk analysis
Approach - Protocols • Weighted molecules determine the final Protocol decision
outcome • ACCEPT : no manual underwriting • REFER: to underwriter for manual review • REJECT: Fits Nationwide target market but does not meet
Nationwide’s risk appetite
Approach – Proof-Of-Concept • POC helped the business understand IBM ODM • Clarified the importance of understanding the business thought
process and semantics • Mitigated many of the project risk • Better understanding of the challenges • Shaped the final solution • Worth the time and money!
The Technology
• IBM Operation Decision Manager: BRMS technology platform that is business centric
• Business users can author rules on Decision Center • Business users can test rules with Decision Validation Service
• Business logic implemented as rules by business users with little support from the IT
• Simulation capabilities • Built in governance controls • Business can manage rule life-cycle & deploy rule
Key ODM Technology Decisions ! Native vs. Dynamic XOM ! Single vs. multiple BOM ! Protocol Effective Dating ! MTDS vs. HTDS ! Simulation, KPI & Decision warehouse ! Performance
The Technology – Key Decisions ! Native vs. Dynamic XOM
• Java XOM for performance reasons • Team members with Java expertise • Easier code & version management using SCCS
The Technology – Key Decisions • Protocol Effective Dating at Ruleset Level
• This capability allows rule authors to author a set of rules (ruleset) that can be deployed and effective at any given date.
(For example: A ruleset can be deployed today with future effective date. Or an older version of the ruleset can be re-run in case of an audit)
• Two part to implementing this solution – Set effective date meta-data in the RuleApp – Implement Interceptor code that uses the effective date meta-data to
select the appropriate ruleset at run-time • Able to keep deploying updates rules while still be able to run
previous ruleset based on policy effective date
The Technology – Key Decisions • HTDS vs Monitored Transparent Decision Service (MTDS) client
• Due to pre and post processing requirements to meet business need, we chose MTDS over HTDS
• Protocol Effective Dating during pre-processing saving detailed Protocol outcome results stored in a database while sending pared down response to the SOA client
• The request and Protocol response results had to be available to be queried by separate application and later consumed by Data Warehousing team
The Technology – Key Decisions • Simulation & Decision Warehouse
• Enable the business analyst to able to run simulation and what-if analysis in their sand-boxed environment
• So business team could continuously refine the rules to further minimize risks
Rule Governance
• Rule Governance should be integral part of the ODM solution since it lays out how, when and who will be able to perform various tasks pertaining to rule-life cycle on ODM
• Expect chaos without proper Rule Governance
Rule Governance - Roles • Roles & Responsibilities
• Rule Analyst • Rule Author • Rule Architect • Rule Developer • Rule Deployer • Rule Administrator • ……
• Roles should be permission assigned so you control and have processes in place
Rule Governance - Release
• Types of Release: – Business Release – IT Release – Emergency Release
• Types of Change: – Compliance change – Decision Improvement – Others
• Category depending on the cost and risk involved: – Major – Significant – Minor
Rule Governance - Business Console • Leveraged the new business rule governance with Business
console to manage rule release life-cycle management, validation and change activity
• Very business user friendly with many facebook like features
Results Delivered! ! Over 20% of small commercial quotes get immediate
underwriting decision ! Business rules team is able to modify and implement rules
quickly based on actual results and market needs ! Simulation is being used to develop the enhanced rules that will
further increase the automated decision percentage and reduce underwriting risks
! Automated underwriting is a key element of Nationwide’s strategy, and it is being expanded quickly
The Results - Keys to Success ! The rules reflect the business thought process ! The solution is business centric with business users in control ! Business users are able to write and test underwriting rules
easily with minimal IT support ! The solution is flexible, easily extendable, and scalable
The Results - Lesson learned
• Involve your business team from the start of the project, educate them, adopt their thinking and semantics, and let them drive the solution
• Integration with other sub-systems can be challenging so focus on integration early
• Having an experienced business and IT team with ODM expertise really helped
Q & A • Scott Grau, [email protected]
Director IT Architecture Office of the CTO - Commercial Lines & Agency
Nationwide Insurance Des Moines, Iowa
• P Vilas Tulachan, [email protected] IBM ODM Enterprise Solution Architect
IBM Worklight Developer Open Source Edge LLC, M: (408)-420-2723 Denver, Colorado
Final Comment • The success of small commercial automated underwriting at
Nationwide can be attributed to how we: • Modeled the small commercial solution • Mirrored the business thinking rather than translate for IT • Experienced project team • Established proper Rule Governance to achieve consistent real
time decisions and enable ability to quickly react to changing conditions consistently