The Journal of Marketing last year (2007) published an article titled “The Long-Term Impact of Loyalty Programs on Consumer Purchase Behavior and Loyalty” by Yuping Liu. It purports to show the impact of a loyalty program on the buying rates and loyalty of those who join the program. The key finding is that very large changes are observed for the lighter and moderate buyers in the loyalty program while the heaviest buyers exhibited no change.

However, this finding, and the consequently very large sales effects that the program seemed to generate, are actually artifacts of the analysis method. In effect there is a “regression to the mean” type phenomenon where the lighter buying loyalty program members were misclassified. In reality they were heavier buyers, when this gradually shows up in the time series data it is mistakenly seen as real change, when it isn’t.

Here is my full (loyaltyselecteffect.pdf) report.

For more on loyalty program effects see here.