What loyalty programmes change, and for whom

Loyalty programmes are everywhere, and many of them reward customers for purchases they would have made anyway. The research on what programmes change is more specific than common practice would suggest.

Customers speed up as the reward gets closer

Ran Kivetz, Oleg Urminsky and Yuhuang Zheng studied a real café programme: buy ten coffees, get one free. Customers bought more often the closer they got to the reward. The goal-gradient effect, first described in animal studies, held for coffee too. Those who accelerated more were also more likely to stay in the programme and come back for the next reward.

A head start helps as well. Joseph Nunes and Xavier Drèze gave 300 car-wash customers one of two cards, each requiring eight purchases: an eight-stamp card, or a ten-stamp card with two stamps already filled. Over nine months, 19% completed the first card and 34% the second. Kivetz and colleagues found a similar effect with a twelve-stamp coffee card that came with two bonus stamps. Customers completed the ten required purchases faster.

The impact is not where the points go

Yuping Liu followed members of a convenience-store franchise's programme over time. Heavy buyers were the most likely to claim their rewards, yet the programme left their purchasing unchanged. The customers who did change had started out as light or moderate buyers. Over time they bought more and became more loyal. Judged by who redeems the most, a programme can look like a success while changing very little.

A review by Matilda Dorotić, Tammo Bijmolt and Peter Verhoef calls the evidence equivocal overall. Programmes tend to increase purchases over time, but the effect varies across customer segments and markets, and different research methods have produced conflicting results.

What follows for design

• Make progress visible and let the next reward feel within reach. Acceleration happens near the goal.
• Give a credible head start rather than a bigger discount.
• Judge the programme by change in behaviour among light and moderate customers, not by redemptions among heavy ones.
• Measure by segment. An average effect hides who actually moved.

In the behaviour-based loyalty framework I worked on at FLO, tiers moved on engagement rather than on spend alone. Incentives went after the behaviour we wanted, not the purchase already made. The studies above are why those choices were more than cosmetic.