In March I wrote here about a man at the airport who pulled a gleaming Lufthansa card out of his wallet just slowly enough for everyone in the queue to notice. Then the upgrade went to the guy in a tracksuit in front of him. At that moment, his card no longer carried any weight. At the time I was interested in him. What the card meant to him, and why something he hadn’t paid for stung more than any delayed flight.
I haven’t forgotten him. I’ve just turned the question around. What is his card worth to the airline?
The answer is not in the miles. It is in the financial statements. Nobody shows it more openly than Delta Air Lines, Air France-KLM’s partner, an airline most Europeans know from transatlantic flights. In January it published its 2025 results: $8.2 billion received from its partnership with American Express, up 11% on the year before. Delta earns money even when its passengers aren’t flying.
The mechanism is simpler than it sounds. American Express issues a credit card with Delta’s logo on it. The cardholder pays for anything with it: groceries, fuel, dinner. Amex buys miles from Delta and hands them to the cardholder for that spending. Delta gets the money before the cardholder redeems a single mile. It settles up later, with a flight.
The cardholder gets miles. Amex gets a person who takes the card out every day, not just before the holidays, and the merchant fees that come with it. Delta gets the cash, plus more than a million new cardholders a year, for the fourth year running. Last year its revenue from economy tickets fell 5%. Payments from Amex grew 11%. The loyalty business grew while economy ticket revenue fell.
Now look at a typical programme in our part of the world, where the cashier asks “Do you have our card?” before saying hello. Points for purchases: one per cent, sometimes two, on every euro at the checkout. The customer collects, the customer redeems, nobody else pays. Every cent comes out of your margin. Delta has a buyer for those miles. You are the only buyer of your own points.
Points on every purchase are just a discount in instalments. Delayed, less visible, spread over fifteen visits. But a discount. By that I mean one thing: the same percentage, returned automatically on every purchase, whatever the customer did differently this time.
The customer wants the discount today. If they have to wait for it, they must not feel you are holding their money. If they have to earn it, it must be clear it was not theirs until they did something for it. Points for purchases miss on both counts. They promise a discount, then pay it out in instalments, and the customer treats them as their own money in your hands. Nobody is grateful to get their own money back. An earned reward carries weight for one reason: it did not exist until the customer earned it.
Let me put it this way. €50 million of revenue goes through the programme, with 1% back in points. Assume 70% of the points get redeemed; the rest expire or sit unused. That is €350,000 a year. You know this number. It is probably in the budget, under marketing. (If half your points sit unused, that is not a saving. It is an alarm.)
Now the question nobody asks in meetings, because the answer is uncomfortable: how much of that €50 million would have happened anyway, without the points?
If the answer is 90%, the programme generated €5 million of extra revenue and paid €350,000 for it. That is not a 1% programme. That is a 7% reward cost on the incremental revenue, nicely disguised. If the answer is 95%, it is 14%. Whether that is a lot depends on your margin. It is certainly not 1%.
You don’t see these numbers because the programme works. Members’ share of revenue is rising, redemption is high, everyone has the app and the dashboard is green from top to bottom. A dashboard that is always green is not a dashboard. It is wallpaper. Success hides the cost: nobody asks any more what the programme costs per euro of changed behaviour.
Here’s where it gets tricky. Delta sells behaviour to a partner. What if you have no partner?
Then you are your own partner. You reward behaviour that saves or earns you money outside the transaction itself, and you pay the reward out of that money. You don’t need to cross the Atlantic to see it. Triglav, Slovenia’s largest insurer, has been doing it since 2015 with its DRAJV app. The app tracks speed, braking, cornering and whether you are holding your phone while driving. Record 400 kilometres a month with a score of at least 90 out of 100 and you get 2% off your car insurance. The cap is 20% a year. Users have logged two billion kilometres. One in ten drivers insuring a car with Triglav has already used a DRAJV code.
Look at what Triglav rewards. Not buying the policy. It rewards you for not ending up in a ditch. The logic is simple. The discount is paid out of claims that never have to be settled (insurers rarely do anything out of kindness). And the customer experiences it as earned, not refunded. Whether the numbers really add up, only Triglav knows. But it chose the behaviour because it could put a value on it before giving away the first percentage point.
Let’s bring this to an outdoor retailer. I know this kind of customer well: I’m one of them. What changes a customer’s value more than one more purchase of the usual kind?
A first purchase in a new category. A customer who buys boots from you and, six months later, a backpack, probably stays longer than one who only buys boots. Still a purchase, but a different one. The change is what you measure.
Collection in store instead of home delivery. The customer who comes for the parcel is standing in your shop with their wallet open. If they try the boots on the spot, they probably won’t send them back a week later. The saved postage is the smallest item on that bill.
Service. A customer who brings in a jacket for reproofing has a reason to return long before they need a new one.
And the climb itself. A customer who logs a hike in your app, with your backpack, tells you how they use your gear. If they share the photo as well, they show it to their friends. A climb is not a purchase. But it is behaviour with value, and you can put a price on it.
All of these are hypotheses. Good ones, but hypotheses.
Before you start calculating, you need three things. A customer you recognise across every channel, because you can neither value nor reward behaviour you cannot see. Margin per transaction, not revenue, because a programme that measures revenue happily rewards the products you lose money on. And the willingness to run a test in which some customers get no reward. Otherwise you will be paying people who would have done all of it for free.
Compare customers with the same past spending, those who showed the behaviour against those who didn’t. That gives you a hypothesis, not a value. Those customers differ in plenty of ways your database doesn’t record. The proof is an experiment. Split customers at random before you offer anything. Offer the reward to one group, but not the other. Then compare the margin of everyone in both groups, not just those who took the reward. Match the length of the test to the behaviour. A collection reward shows its effect far sooner than a reward for a new-category purchase.
Then comes the part most people miss. You pay the reward to everyone who shows the behaviour. Only those who changed it because of the reward create value. Say a store collection is worth €7 to you (saved delivery, an extra purchase, fewer returns) and you reward it with €2. Out of 100 rewarded collections, if the reward triggered 20, you created €140 of value and paid out €200 in rewards. So reward the switch, not the habit. Take the customers who had every parcel delivered to their home last year. Split them at random into two equal groups and send the offer to one. If 40 in that group collect in store and 5 in the other, you paid €80 in rewards for €245 of estimated incremental value.
My rule: total reward cost no more than a third of the estimated incremental value. Out of the rest you still have to run the programme, make a profit and keep a reserve for your own miscalculation.
If you cannot write the number down, don’t reward the behaviour. If you can, the reward pays for itself out of that number. Points for purchases then become the least interesting part of the programme.
I’m not asking how many points you give. The percentage is the least important decision in your programme.
The man at the airport was not loyal to Lufthansa because of a discount. He was loyal because of something he had earned - and which stopped carrying any weight the moment someone else got the same thing for free. A discount in instalments cannot carry that weight. It just makes the customer wait.
So I’m asking something else: what does your loyalty programme reward besides the purchase - and what is that behaviour worth to you?
igor.pauletic@frodx.com
P.S. I’m not saying every retailer needs its own American Express. I’m saying every programme needs at least one change in behaviour it can prove, and a number for what that change is worth. Without that number, you don’t know whether you have a programme or a discount.