What your loyalty programme actually costs

Loyalty programmes are usually justified with enrolment numbers and redemption rates, neither of which is a cost. The number that decides whether the programme is worth running is the cost per incremental visit: what you paid for the visits that would not otherwise have happened.

The cost calculation

Work it out in this order. The arithmetic is simple; the discipline is refusing to skip the holdout.

1. Gross reward cost: food cost of redeemed rewards plus discount value, for the period.
2. Incremental visits: visits from enrolled members above the holdout group’s rate, over the same period.
3. Contribution from those visits: incremental visits multiplied by contribution margin per visit.
4. Programme cost: gross reward cost plus platform fees plus the operational cost of running it.
5. Verdict: contribution from incremental visits minus programme cost. Positive, and the programme earns its place on frequency alone. Negative, and it must be justified on data value instead, which is a legitimate but different argument.

Reward cost is not reward value

A free medium pizza has a menu price of perhaps $16 and a food cost of perhaps $3.50. The customer receives $16 of perceived value; you incur $3.50 of cost, plus the labour that was going to be paid anyway.

This gap is the single most useful lever in reward design. Rewards with high perceived value and low food cost (sides, breads, desserts, upgrades) deliver more felt generosity per dollar spent than any percentage discount can. A percentage off is the worst possible reward structure on this measure, because it costs you full margin for a benefit the customer barely registers.

Give away things with a big price and a small cost. That is not a trick; it is the only sustainable way to be generous.

Carry the liability honestly

Outstanding points and unredeemed rewards are a liability. Track outstanding balance, estimated redemption rate and estimated cost, and revalue quarterly.

The number moves in ways that surprise operators. A promotion with double points creates a liability spike that lands as a cost two months later. A threshold change re-prices every outstanding balance at once. Neither shows up in the week it was decided.

Breakage is a forecast, not a strategy

Breakage, value earned and never redeemed, reduces real cost, and it is legitimate to forecast it. It is not legitimate to design for it. A programme that relies on customers failing to redeem is a programme that works best when it fails at its own purpose, and customers eventually notice.

Practically: assume a breakage rate from your own history, keep it conservative, and treat any surprise increase as a warning that the programme has become too hard to use rather than as good news.

Questions