The second breakout this afternoon was on improving claims throughput by Rod Dunlap, director at BlueCross BlueShield of North Carolina. Their main driver was to reduce the percentage of claims requiring manual intervention from an initial rate of 17% to a mere 2% by the end of 2009; this didn’t quite turn out to be realistic, but they are headed for 4.5% by the end of 2009. This four-fold performance increase will result in $20M in annual savings, which buys a lot of Pega licenses, but more importantly, this is seen as the implementation of transformational change that will change the corporate culture.
The obstacles are pretty typical of what I’ve seen: middle management in this conservative company was risk averse and threatened by change, with a predisposition to failure from some of the key business management. There was a history of trying to use any new technology as a silver bullet, which typically ends in some sort of disaster. Furthermore, they had a traditional waterfall development methodology governed by an enterprise project office that tended to get in the way of innovation.
To address this, they developed a number of innovations:
- Business people writing business rules: about 180 rules so far. As with any model-driven design, this eliminates translation errors, increases accuracy and increases time to delivery. The problem, of course, is that business people don’t understand programming basics, and Dunlap admitted that Pega isn’t quite as easy to use as the brochures claim. 🙂
- To resolve this, they built their own claims optimization framework. Claims could come in in any format, and rules (written by the business) causes the claim to be routed to the appropriate area for processing or rejected to the old process if it couldn’t yet be handled by the framework application. The framework appears to be a set of common services such and logging and reporting, plus a configurable application layer. They’re using IPD as their workflow engine currently, with Pega being used primarily as a rules engine, but in some cases the claims do not have to hit the workflow engine at all since the rules can determine the disposition of the claim. They plan to replace IPD in the future with Pega’s BPM.
- Implement Agile/SCRUM concepts in an iterative development methodology, with 30-day sprints that result in a code release every 30 days — no schedule slippage allowed, although features may slip into the next version. Features within a sprint are prioritized based on business payback, effort and dependencies. They’ve moved away from use cases and into user stories (scenarios and personas): this is the best practice for modern user design interaction. The combination of prioritizing features by payback and the use of personas allows some edge-case features to be completely dropped off the schedule.
They co-located the business and IT people on the project — key to success in Agile projects — and keep the teams small. They kicked the project off in December 2007, and completed their first sprint in February, with the framework delivered in July 2008.
With what they’ve accomplished so far, 86% of duplicate claims are handled automatically, saving 20 FTE; 48% of corrected claims do not need manual intervention, saving another 18 FTE. They’re still working on the goal of 4.5% manual touches on claims by adding automated claims adjustments and building new types of claim repairs. Then, they’ll expand beyond claims to bring the rules technology to membership, billing and finance, and start implementing BPM.