Expert analysis · reviewed 6 Oct 2026
Loyalty programme: measuring what it changes, not who it attracts.
Key distinctions
Three conditions before calculating.
Joining is not a random treatment
Customers who join a programme already buy more often and spend more. Any gap measured after joining mixes this selection effect with the programme's real effect.
Rewards are a cost, not a gesture
Points, discounts and benefits are paid for in margin, including on purchases that would have happened without them. Unredeemed points reduce this cost but remain a liability as long as they can be redeemed.
Customer data has a value, to be proven
The programme identifies customers and makes other actions possible. This value only counts if a decision actually uses it and its gain is measured.
Method
Measuring the effect of a loyalty programme in four steps
- 01Rebuild the pre-joining period
Measure each member's spending over a comparable period before joining, and that of a group of non-members over the same dates.
- 02Isolate the incremental effect
Calculate a difference in differences or, better, randomly assign the invitation to the programme or a change in rewards.
- 03Convert into margin
Apply the contribution margin to the incremental spending, deducting discounts and free products.
- 04Deduct the full cost
Subtract the rewards redeemed, the points still redeemable, the platform and the running of the programme, then express the result per member per year.
From the observed gap to the programme's net effect
Members spend €180 more per year than non-members. Broken down, this gap mainly reveals who joins, and a negative net effect.
| Component | Amount per member per year | Reading |
|---|---|---|
| Observed spending gap | +€180 | Members versus non-members, after joining |
| Gap already present before joining | +€130 | Selection: the best customers join |
| Incremental effect of the programme | +€50 | Difference in differences |
| Incremental margin at 40% | +€20 | What the programme brings in |
| Cost of rewards and management | −€23 | Points redeemed and operations |
| Net effect | −€3 | The programme subsidises purchases already secured |
Comparing members and non-members mainly measures who joins, not what the programme changes.
A €180 gap hiding a loss-making programme
Illustrative example: before joining, future members spent €360 a year and non-members €230. A year later, members spend €420 and non-members €240. The margin is 40%, and rewards and management cost €23 per member.
Incremental effect = (420 − 360) − (240 − 230) = €50 of spending per member. Incremental margin = 50 × 40% = €20. Net effect = 20 − 23 = −€3 per member per year.
The programme seems to bring in €180 more per member; in reality it creates €50, and loses money. Bringing the cost of rewards below €20 or concentrating them on customers whose behaviour changes would make the effect positive.
Acceptance conditions
What must be true to act.
- 01
Never attribute the raw gap between members and non-members to the programme.
- 02
Measure the effect over a pre-joining period or through random assignment.
- 03
Assess the programme on margin net of rewards and operations, per member per year.
- 04
Record unredeemed points as a liability as long as they remain redeemable.
Limits
What this analysis does not prove.
- A difference in differences assumes that members and non-members would have evolved in the same way without the programme.
- A programme's effect can erode when competitors offer comparable benefits.
- Effects on referrals and on the quality of customer data remain hard to quantify.
- A programme measured over one year does not say what it will produce over its members' lifetime.
References
Works cited.
- Leenheer, van Heerde, Bijmolt & Smidts (2007), Do Loyalty Programs Really Enhance Behavioral Loyalty? An Empirical Analysis Accounting for Self-Selecting Members (opens in a new tab)
- Liu (2007), The Long-Term Impact of Loyalty Programs on Consumer Purchase Behavior and Loyalty (opens in a new tab)
- Dorotic, Bijmolt & Verhoef (2012), Loyalty Programmes: Current Knowledge and Research Directions (opens in a new tab)