Two years ago I worked for JD Power and Associates on an R&D project. Ultimately the project was canceled when the entire Dallas, Texas department was shutdown. What we were trying to do was to Monetize Customer Satisfaction by linking it to Customer Loyalty. We were trying to use a wide variety of data about a customer to determine what the long term worth of a customer was to a firm and what was the driving factors in determining the customers Loyalty. By understanding these drivers we hoped to increase a customers Loyalty and value to a firm.
Yes, that is the short explanation. The full story was told in several extended white papers that I helped write. It was a very interesting project, and I strongly believe that we were on to a some very cool and valuable ideas.
Well, I see that these Customer Loyalty Models are finally being reported on in the mainstream press. And like any such idea, there is big upside and a dark down side. This analytic tool can tell a firm how to spend its money to increase customer loyalty, but it can also tell you when maintaining a customer's loyalty is not profitable.
I believe that in the coming years, we will see customer complaints about how they are affected by statistical models that reduce their quality of service or out right being denied service.
The article describes what would normally be consider a good customer for Fidelity Investments who was asked to leave becuase the cost to Fidelity to keep his loyalty was too high. Now Fidelity may be correct to bump this bum, and the story provides scant details. But they better have a high degree of faith in their analytic tools that led them to this decision. Statistical tools can tell you a great deal, but blindly relying on them will get you in trouble.
For example, the customer at Fidelity may have had good reasons to tie up three financial advisers. The customer could be in the middle of a very complex time of their financial life. (Retirement, divorce, starting a business, etc). What if, Fidelity's statistical model is not considering the full life cycle of this customer, who could be a profitable client when considered over a 30 years period. I believe that the importance of the lifecycle analysis is a critical and difficult componet. Most companies don't have detailed data sets hat go back 10 years much less 30 years. And even if they do, the generations are different. The customer expectations of today's techie Gen X'er is different from a 1970's era baby boomer. Both technology and culture have changed.
Most firms will probably use these new tools reasonably well. Some will use them with a great deal of incompetence and hurt their bottom line. And a few firms will find ways to use these new statistical models to justify racism, sexism and other ism's.
posted at: 12:20 | path: / | permanent link to this entry
