Do loyalty cards actually work?
Loyalty cards work when three things are true: customers buy often enough to reach a reward while they still care, the reward is worth less to you than the extra visit is worth, and someone is actually running the programme. Where any of those is missing they don't work, and the card becomes a discount you hand to people who were coming anyway. The mechanism that does the work is not the card. It's knowing who stopped coming and being able to reach them for free.
The honest failure cases
Start here, because most pages on this query won't.
| Situation | Does a loyalty card work? | Why |
|---|---|---|
| Under ~30 regulars | No | You already know them. Software adds nothing. |
| Once-in-a-lifetime purchase | No | No frequency, no loop |
| Reward more than ~10 visits away | Rarely | People stop counting before they finish |
| Nobody owns the programme | No | It decays to a paid-for nothing |
| High frequency, flat ticket, someone running it | Yes | This is the case it was invented for |
| Wide ticket range, points or cashback, someone running it | Yes | Fair rewards, plus a lapse signal |
Do loyalty cards actually work according to research?
The best evidence on stamp cards is a field study run inside a real café's "buy 10, get one free" programme, published in the Journal of Marketing Research by researchers at Columbia and Chicago Booth. Three findings matter to any small business considering one:
| Finding | The number | What it means for you |
|---|---|---|
| Customers speed up as the reward gets closer | Interpurchase time fell by an average of 0.7 days, a 20% acceleration; whole-card completion 24.6 days vs 29.4 | The card genuinely changes behaviour, but mostly near the end |
| The feeling of progress works even when the progress is given | A 12-stamp card with 2 free stamps was completed in 12.7 days vs 15.6 days for a plain 10-stamp card | Same real effort, 20% faster, because it starts part-filled |
| Behaviour drops back after a reward is claimed | Purchase rates reset downward right after the first reward, and customers were most likely to defect at that moment | The moment you hand over the free coffee is your riskiest moment, not your safest |
Source: Kivetz, Urminsky & Zheng (2006), Journal of Marketing Research 43(1).
That third row is the one nobody acts on. Most businesses celebrate the redemption and then do nothing, at precisely the point the research says the customer is most likely to leave. A programme that immediately starts the next card, with the first stamp or two already on it, is doing two evidenced things at once.
The bit that does the actual work
Here is the uncomfortable truth about stamp cards: a meaningful share of the rewards go to people who would have come back regardless. On rewards alone the maths is often marginal.
What tips it is the data. A digital card knows that a client with eight stamps hasn't been in for seven weeks on a four-week cycle. That client is leaving, and right now, without a programme, you would never know until they were gone. One free push notification at that moment is where the return lives, not in the tenth-cup-free.
That's why paper cards are the weakest version. Paper rewards loyalty but cannot detect its absence.
What a working programme looks like
- Reward reachable inside four to six visits.
- Enrolment is a front-desk habit, one QR, one sentence, every transaction.
- A lapse trigger: a message to anyone past their normal return window.
- A monthly look at who's active, who's slipping, who's gone.
- One named person responsible.
Anything less than that is a card, not a programme.
Make the offer explicit
People hand over a phone number and their buying history. Say plainly what they get for it. A card that appears in someone's wallet without a clear promise attached is the reason loyalty schemes get ignored, and it is the reason some get resented.
Put three things in front of the customer at the moment they join:
- What they earn, in plain terms. Not "rewards", but "5% back on everything, spendable here".
- What you collect, in one line. Their number, what they buy, and when.
- What you will send, and how they stop it. Frequency and a one-tap opt-out.
That clarity is not only good manners. It is the same disclosure that consent rules expect, so writing it once solves the marketing problem and most of the compliance problem together.
Pro tip. Set a standing job for the week after a redemption. The research says that is when customers are most likely to drift, and it is the one week almost no programme covers.
How to tell whether yours is working
Two numbers, both of which any digital programme can show you:
- Repeat rate among card-holders vs non-holders. If it isn't higher, the programme is a discount.
- Reactivation rate on lapse messages. This is the number that justifies the subscription on its own.
Both are visible in a digital programme from the first full purchase cycle. If you cannot see either number, you do not have a programme. You have a card.
Pro tip. Keep a route in for people without a smartphone. the CMA's 2024 review of supermarket loyalty pricing criticised supermarkets for excluding under-18s and shoppers without smartphones from loyalty prices, and recommended offline registration and physical cards as alternatives. The same fairness point applies at a much smaller scale.
Frequently asked questions
Do people actually use digital loyalty cards? Usage tracks friction. Wallet cards need no download and no login, which is why they enrol far better than branded apps. The customer scans once and it is done.
Aren't I just discounting customers I already had? Partly, yes, and any honest answer says so. The gain comes from the visits the programme causes and from the lapsing customers you can now see and reach.
How long before I know if it's working? One full purchase cycle plus a month. For a café that's about six weeks; for a salon, about three months.
What's the biggest reason they fail? Nobody runs them. The second biggest is a reward set too far away to feel winnable.
Sources
- Kivetz, R., Urminsky, O. & Zheng, Y. (2006). The Goal-Gradient Hypothesis Resurrected. Journal of Marketing Research 43(1), 39–58. Full paper (columbia.edu) · Summary (Columbia Business School)