This is the topic that decides whether your program makes money or quietly bleeds it, and it is the one most owners skip because it involves a little arithmetic. It is not hard arithmetic. If you can work out food cost, you can work out whether a loyalty program pays, and doing it once before you launch is the difference between a tool and a leak.
🧾 A program is a discount with rules
Strip away the app and the branding and every loyalty reward is the same thing: margin you choose to give up. A free coffee is the cost of that coffee, gone. Ten percent back in credit is ten percent of that sale, gone. This is not a reason to avoid programs — it is the reason to be clear-eyed about them. You are spending real money, and the only question worth asking is whether you are getting more back than you spend.
The trap is that the spending is visible and the return is not. You can see the free coffee you handed over. You cannot see, without looking for it, whether that customer came more often because of the card. So owners feel the cost, never measure the benefit, and either kill a program that was working or keep one that was not.
🔢 The only question that matters
Here it is, and everything else is detail: does the reward cost less than the margin on the extra visits it caused? The word that carries all the weight is extra. A visit that would have happened anyway is not a benefit; rewarding it is pure cost. A visit that happened only because of the program is the entire point.
So a program has two effects, and you have to weigh them against each other:
- The cost side. Every reward you pay out, including to customers who would have come without it.
- The benefit side. The margin on the genuinely extra visits — the ones the program caused that would not have happened otherwise.
If the benefit beats the cost, you have a good program. If it does not, you are running a discount for your regulars and calling it loyalty.
☕ Working one example
Take a café with a "buy nine, get the tenth free" coffee card. A coffee sells for a certain price, and only a small part of that is the cost to make it. Watch how the answer depends entirely on behaviour, not on the card itself:
| Customer | What the card does | Result for you |
| Already came 10 times a month | Nothing changes; they just collect a free coffee they earned anyway | You gave away 1 coffee's cost for zero extra visits. A loss |
| Came 6 times, now comes 10 to finish the card | 4 extra visits, each with real margin, to earn 1 free coffee | The margin on 4 visits versus the cost of 1 coffee. A clear win |
| Came once, never finished the card | Collected a few stamps, never redeemed | Extra visits and no reward paid out at all. Also a win |
The card is identical in all three rows. What changes is whether it moved the behaviour. This is why a program aimed at frequent regulars who would come anyway is the most common way to lose money, and one aimed at occasional customers you want to see more often is the most common way to make it.
🧨 Where the math goes wrong
A few patterns turn a program from a tool into a leak, and all of them are visible in the arithmetic before they show up in the bank:
- Rewarding your base. If the people earning rewards are the ones who already came daily, you are paying for visits you already had.
- A reward that costs more than the visits it buys. A free main course carries real food cost; it can wipe out the margin on the visits that earned it. Cheaper, high-margin rewards almost always do better.
- Discounting the wrong people. Tiers that give your biggest spenders the deepest discounts can cost you margin on exactly the customers who were happy to pay full price.
None of this means do not run a program. It means run the numbers first, on your own prices and your own food cost, for one honest scenario. If the extra visits do not cover the rewards, the answer is not a better app — it is a different program, or none. The next topic assumes you did this and decided yes.