Credit Cards Or Transaction Data
A recent Wall Street Journal (10/27/11) carried a story about using credit card data to predict whether or not you will take your medicines as prescribed.
The article, “Next Frontier in Credit Scores: Predicting Personal Data” describes the Fair Isaac “Medicine Adherence Score” that “is based partly on how long a person has lived at the same address and whether he owns a car.” The Journal notes that “The proliferation of ’scores‘ highlights the widening trade in personal information, which is already fueling public concern about diminishing personal privacy.”
Our view: forget credit card scoring and avoid privacy concerns. There’s a better way.
Scores based on actual transactions produce more reliable, less “iffy” results. For more than two years Loyalty Builders has been calculating a ‘Compliance Score’ that measures whether or not diabetics are taking their insulin on schedule. The big difference between Loyalty Builders’ method and the credit card scorers is that the data we use comes from actual patient transactions. So instead of age, gender, and auto ownership, Loyalty Builders’ scores are based on prescriptions refilled and quantity and interval at which supplies like diabetes testing strips are purchased. Many of the diabetics we score are elderly, don’t own a car, and may not have a mortgage on which to track payments – credit card scoring wouldn’t work.
Transaction scoring has a more direct connection between what’s measured and the desired behavior. It also avoids privacy concerns. Scoring based on transactions is going to outperform scoring by inference every time — it just makes sense.