Rafi Mohammed writes for Havard Business Review. Recently, he wrote about his reasons for recommending that clients abolish their loyalty programs in all but three cases. There’s nothing wrong with writing opinion pieces on a blog. I do it all the time. As a loyalty and engagement strategist, though, I am careful to distinguish my gut opinion from my professional recommendations. When recommending a strategy, I try to cite hard evidence. I cringe when a business strategist makes recommendations based on . . . pure opinion.

Mohammed arrived at his professional recommendation through a little research: he asked his mother if she frequents a restaurant (Subway) more often because of its loyalty program. “No,” she told her son, “but it’s nice of them to offer it.” [source]

There are two big flaws in Mohammed’s opinion of loyalty programs: the weak science he relied upon, and the strong science he ignored.

Weak Science

The survey method is better than nothing, but asking someone to predict his or her future behavior is something of a crapshoot. Most people don’t really know how they’ll respond to an offer. For example, UCLA researchers Lieberman and Falk showed that fMRIs are far more accurate than surveys in predicting future behavior.

And Mohammed’s survey lacks rigor because the sample size is just one. Daniel Kahneman points out the problems of small samples in his recent book, Thinking, Fast and Slow, with this exercise:

1. The US counties with the highest incidence of kidney cancer are rural, mostly white, and mostly Republican. Why do thing that is?

2. The US counties with the lowest incidence of kidney cancer are rural, mostly white, and mostly Republican. Now explain that.

The explanation is small sample size. Extreme events are more likely in smaller samples than in larger. By relying on one person’s unreliable prediction of her own motivation, Mohammed has disguised his opinion as a hard fact.

Strong Science

Mohammed’s recommendation to ditch loyalty programs ignores good science.

Mohammed states that a buy 8 get one free offer do not drive behavior. But a famous experiment at the University of Southern California demonstrates otherwise.

In the experiment, customers of a car wash were randomly given one of two loyalty cards. One card offered a free wash after 8 paid washes. The other card offered a free car wash after 10, but the first two slots were already punched. Both groups needed to buy 8 washes to complete the card.

The group with the head start was more likely to complete the card than the other group (35% to 20%). Of those who did complete the card, the group with the head start finished faster, meaning their frequency was higher. [source]

Mrs. Mohammed Proved Her Son Wrong

What bothers me most about Mr. Mohammed’s piece, though, is its implication that short-term profits are all that matter. He complains that his clients squander money by offering loyalty discounts–money they should be keeping for themselves.

A 5% loyalty discount — $5 off a $100 sale — results in a 50% decrease in profits. The costs remain the same, but instead of earning $10 from the sale, profit is reduced to $5. What appears to be a small discount — in this case, 5%, — can significantly impact profits. It’s important to share with your front-line workers how costly these discounts can be to the company’s bottom line.

Yet, his only evidence that loyalty programs don’t work–his mother’s testimony–demonstrates the true value of the program.

"[I]t’s nice of them to offer it."

Even if the loyalty card fails to increase Mrs. Mohammed’s frequency, she’s not going to the competition! The merchant in Mohammed’s example above would probably prefer his $5 profit to no profit at all.

Here’s what it all boils down to. The USC study, and similar studies at the University of Pennsylvania and elsewhere, demonstrates that Mohammed’s mom doesn’t speak for most of us. The UCLA fMRI study says Mrs. Mohammed may not even speak for herself.