In brief
A loyalty program returns money through four mechanisms, and three of them can be written as a sum. The break-even is one division: program cost divided by gross margin per visit is the number of extra visits it has to buy.
The short answer
A loyalty program returns money four ways: visits that would not otherwise have happened, larger baskets on the day a reward is collected, a discount that gets paid in cost of goods instead of revenue, and customers who would have drifted away and did not. Three of those can be calculated. The fourth cannot be observed directly and should be treated as upside rather than as budget.
The break-even is one division. Monthly program cost divided by gross margin on a typical visit is the number of extra visits the program has to produce to pay for itself.
The qualitative version of this case, what a program changes about how a business feels to run, is on the benefits of a loyalty program. This post does the numbers instead.
The business used in every example
One set of assumptions, carried through the whole post so the sums can be checked against each other. Substitute yours.
- A lunch counter selling a sandwich and a drink for £7.50, costing £2.60 in food and packaging. Gross margin £4.90 a visit.
- 60 customers enrolled, each coming in about 3 times a month. That is 180 visits a month.
- A ten-stamp card, one stamp per visit, reward is one £7.50 lunch.
- Software on Starter at £9 a month.
- Counter wage £12 an hour.
180 stamps a month completes 18 cards, so 18 rewards a month at £2.60 of goods is £46.80. Add £9 of software and about nine minutes of counter time at three seconds a stamp, which is £1.80. The program costs £57.60 a month. The full version of that calculation, and the four line items behind it, is what a loyalty program actually costs.
Mechanism 1: visits that would not have happened
The largest return, and the only one that grows without limit.
Suppose the enrolled 60 move from 3.0 visits a month to 3.4. That is 24 extra visits, each carrying £4.90 of gross margin, so £117.60 a month against a £57.60 cost.
Now the important part. I invented the 3.4. You cannot. It is the one number in this post that has to come from your own records, and it is the reason a readable ledger matters more than any feature on a pricing page: the visit interval of enrolled customers, before they joined and after, is the whole argument. A program that cannot tell you that number has removed your ability to check whether it works.
Note how small the change is. Going from 3.0 to 3.4 visits a month is one extra lunch every ten weeks per person. Programs are not usually judged against a bar that low, and they should be.
Mechanism 2: the basket on redemption day
A free lunch is rarely collected alone, and the attachment is worth counting.
Say half the people collecting a reward also buy a coffee at £2.60 that costs you £0.70, so £1.90 of margin. Nine of the 18 monthly redemptions attach one, which is £17.10 a month.
That covers roughly a third of the £46.80 reward line. It softens the cost. It does not pay for it, and a program justified on attachment alone is a program justified on the smallest of the four mechanisms.
This one is easy to verify and almost nobody does. Look at what else was on the receipt on the days rewards were claimed, for one month. The answer replaces my invented half with your real fraction.
Mechanism 3: a discount paid in cost of goods rather than revenue
The mechanism nobody advertises, and the one with the cleanest arithmetic behind it.
Two offers that look identical to a customer:
- 10% off everything, permanently. 180 visits a month at £7.50, ten per cent off, is £135 a month given away.
- A ten-stamp card. One free lunch per ten visits is also ten per cent, and the retail value is also £135. It costs you £46.80, because a sandwich you hand over costs £2.60 rather than £7.50.
Same headline generosity. A shade over a third of the cost. The difference is entirely that a discount is paid out of revenue at menu price and a reward is paid out of the stock room at cost.
There is a second difference underneath it. The standing discount pays on the first visit, to everyone, including the person who was walking in regardless and would never have noticed. The card pays after nine visits have already happened. Same money, opposite risk, and the timing is the argument set out in how loyalty programs work.
Mechanism 4: the customer who would have stopped
The largest effect in most descriptions of loyalty, and the one you cannot measure.
The arithmetic is easy enough. A customer at three visits a month is worth £14.70 of gross margin a month, or £176.40 a year. Keep two of those a year who would otherwise have drifted and that is £352.80, roughly half the program’s £691 annual cost.
The problem is the phrase “would otherwise have drifted”. It describes something that did not happen, and no ledger records a departure that was averted. Anybody quoting you a retention uplift for a stamp card is quoting a counterfactual they did not observe either.
One version of it is countable. A customer whose card is halfway complete, who has not been in for six weeks, who receives one message and comes back, is a return you can see happen, because the gap and the message and the visit are all on record. That is the narrow, observable slice of retention, and it is what winning back customers who stopped coming is about. Count that one. Leave the rest as upside.
The break-even, in one division
Monthly program cost ÷ gross margin per visit = extra visits needed per month.
That is the whole calculation. Three businesses, each with its own assumptions:
| Business | Program cost a month | Gross margin a visit | Extra visits needed |
|---|---|---|---|
| Cafe, 100 customers, ten-stamp card | £47.70 | £2.40 | 20 |
| Lunch counter, the example above | £57.60 | £4.90 | 12 |
| Barber, 80 customers, six-stamp card | £27.70 | £22.50 | 2 |
The cafe row carries the assumptions in what a loyalty program actually costs. The barber row assumes a £24 cut with £1.50 of consumables, customers returning about every five weeks, 12 free cuts a month, and a chair that would otherwise be empty during them.
Twelve extra visits a month across 60 lunch customers is one extra visit per customer every five months. Two extra visits a month is a barber persuading two people to come at five-week intervals instead of six. These are the honest bars, and they are low enough that the interesting question is not whether the program can clear them but whether anyone is watching to see if it did.
The row that hides a much bigger number
The barber row is the one to argue with, because a free haircut only costs £1.50 when the chair was empty anyway.
Twelve free cuts a month is nine hours of chair time. If those redemptions land on Saturday afternoons, the shop is not spending £18 on consumables, it is turning away up to twelve paying customers at £22.50 of margin each, which is £270. The same reward, on the same card, costs fifteen times more depending on the hour it is claimed.
That is a real property of any service business where the constraint is time rather than stock, and it has two fixes. Restrict the reward to the quiet part of the week, which every customer understands and nobody resents. Or make the reward a product rather than a service: a bottle of what you use, a beard trim, something that comes off a shelf. How to reward customer loyalty goes through choosing between them.
A shop selling goods off a shelf does not have this problem. A shop selling an hour of someone’s time always does.
What cannot be counted, said plainly
Three things people list as loyalty program benefits genuinely happen and genuinely cannot be measured by a business your size.
Word of mouth. People tell their friends about the free coffee. You will never know how many, and no attribution scheme at this scale is honest.
The conversation. “Shall I put that on your card?” produces recognition between staff and customers that no service initiative buys deliberately. Real, valuable, unquantified.
Being remembered. A card sitting in a phone is a reminder your competitor does not have. It cannot be separated from every other reason someone thought of you.
No number appears next to any of these because I do not have one and neither does anyone else. Use them as tie-breakers between programs that already break even on the arithmetic above. Never use them to rescue a program that does not.
Do the division for your own shop
Write down three figures: what the program costs you in a month with rewards priced at cost of goods, your gross margin on a typical visit, and how many customers are enrolled. Divide the first by the second.
If the answer is more extra visits than your enrolled customers could plausibly make, the problem is the reward or the threshold, not the software. Raise the threshold by two, run the division again, and check whether the reward still arrives inside ten weeks at the rate people already visit. Whether the program then actually produces those visits is a different question, answered in do loyalty programs actually work.