Some weeks ago I got a chance to review the SAS Warranty Analysis product. I was doing some due-diligence before my speech on “Next Generation Warranty Systems” to the Warranty Chain Management Conference in April. The folks from SAS began with an Aberdeen quote from 2006:
Warranty analytics is the number one differentiator between Best in Class and laggard service organizations.
They identified a set of warranty issues:
- Problem/identification/resolution cycle takes too long
- Increasing government intervention like TREAD act
- Warranty claims and reserves
- Cost recovery
- Invalid and fraudulent claims
The service chain is typically broken up into pieces and integrated only through business processes/workflow but the same data is used everywhere. Pulling all this together as what they call “Service Intelligence” adds value to the service chain. The core of the SAS Warranty product is that it integrates all the relevant data so you can use analytics, scorecards, to understand it. The product helps companies identify problems faster, prioritize based on impact and conduct root cause analysis. It provides automated reporting and alerting as well as collaboration features. At the end of the day it means fewer products ship with problems, costs are lower 10-20%, and customer satisfaction is improved.
SAS Warranty Analysis offers unique analytical routines and various tools for different users. It has a standard data model that integrates everything around warranty including claims, products, sales, surveys, call center etc. Business rules (about the data) and processes are layered on top to get it ready – by handling situations like missing sales data because the company only found out about a sale when they get a claim. These rules are based on SAS technology and are specific to the warranty product (which is probably OK as they are data rules).
The product has various KPIs/dashboards for different roles as well as a report library and automated “emerging issues” monitoring. For instance claims data is being used to see how long things have been in service when a failure occurs and a change in average time in service might be an indicator of a change in reliability. Problems identified using the tool can be easily saved (as reports, for instance) and shared to improve collaboration – lots of this information thus gets fed back into the service process, albeit in a somewhat hit and miss way at the moment. The product has a set of interactive reports for problem definition from trends to multi-variate statistical drivers (it builds a decision tree to isolate failure). In general these reports help with specificity for instance in recalls but the tool does not take action on these items. Warranty know-how is needed but not deep analytic skills unless you want to use the Enterprise Miner extension.
The product has nicely integrated text analysis – users can do things like “find other claims with text like this one” when down in the details. It also offers integration with SAS Enterprise Miner and Enterprise Guide to allow more sophisticated modeling. These tools can access data sets inside Warranty although the rest of the integration is a bit loose and no-one is yet pushing the models back into production except perhaps in fraud detection. They have at least one customer producing fraud scoring models and feeding these back to the audit division to avoid “pay and chase” fraud detection. This company saved $4M of claims fraud in the first 5 months – a great illustration of the power of putting analytics to work in operational systems (the topic of my presentation on next generation warranty systems).
While SAS Warranty Analysis is not a decision management application it is a nice example of how good data analysis tools can position you well for decision management.
The white paper I wrote for the Warranty Chain Management Conference on Next Generation Warranty Systems is available at decisionmanagementsolutions.com/warranty