Loyalty card marketing: the four mechanics, the arithmetic behind the reward, and the terms to publish
Stamp cards, points, tiers and paid membership compared, the discount rate your card really carries, what to collect at the till, and the numbers to report monthly.
Published by AI Knows Us · Updated 12 October 2026
Loyalty card marketing is any scheme that records a customer's repeat purchases against an identity and rewards them for it. There are four mechanics in common use: the stamp card, points, tiers, and paid membership. Which one suits you is decided by two numbers you already have, how often your customers buy and what margin each sale carries, and the reward you choose has to be funded by the extra visits it causes rather than by the visits you were getting anyway. The card is also a data collection device, and for most small businesses that is the more valuable half of it.
The four mechanics, and what each one suits
| Mechanic | How it works | Suits | What goes wrong |
|---|---|---|---|
| Stamp card | Buy a fixed number, get one free | High frequency, low value: coffee, lunch, car wash, salon blow dry | Cards are lost, so completion is low and the data never reaches you |
| Points | Spend earns a balance that converts to money off | Variable basket size: grocery, pharmacy, pet, hardware | Earn rate set too low to feel worth tracking, so nobody engages |
| Tiers | Cumulative spend unlocks a status with better terms | Higher value, identity driven: beauty, fashion, hospitality, airlines | Top tier is reachable by too many people, so it stops meaning anything |
| Paid membership | A fee buys standing benefits such as free delivery or a flat discount | Predictable repeat purchase with delivery or service costs attached | Benefits priced below what heavy users consume, so the best customers cost the most |
Most independents should start with the simplest mechanic their frequency supports and add complexity only when the data gives them a reason. A tier scheme on a business with four purchases a year per customer is administration without a payoff.
The arithmetic that decides the reward
Work out the real discount rate before you print anything. A card that gives one free item for every ten bought means one item in eleven leaves without payment, which is a discount of roughly nine per cent of units for every customer who completes a card. Multiply that by your gross margin to see what it costs you in money, then apply two adjustments. Not every card gets completed, so the cost on the whole enrolled base is lower than the cost on completers. And part of the spend would have happened anyway, so the extra visits are the only part the discount is buying.
That gives you a simple test to run after ninety days. Compare the enrolled customers against the unenrolled on three things: visits per month, average basket, and the share of visits that include a second item. If the enrolled group is ahead on visits and holding on basket, the scheme is paying for itself. If the enrolled group is ahead on nothing but redemption, you are discounting loyalty you already had, and the fix is to move the reward away from money off and towards something with a high perceived value and a low unit cost: an upgrade, early access, a service add on, a product your supplier funds.
Card, app or phone number
A printed card costs almost nothing, works with no technology at the till, and is the only option that collects no customer data at all. An app or a card tied to a point of sale account collects the purchase history, which is what lets you see frequency, lapse and basket, and it costs a subscription plus the effort of getting people to install something. A middle route works well for small premises: look the customer up by phone number or email at the till, keep the balance in the point of sale, and print the balance on the receipt. No card to lose, no app to install, and the data lands in one place.
Whichever you pick, the enrolment moment is the whole scheme. One sentence from the person at the till does more than any poster: name the benefit, say how long signing up takes, and ask. Write that sentence, put it on a card by the register, and include it in training for new staff.
The terms, written once and published
Every scheme needs a public page, not a laminated sheet in a drawer, that states the earn rate, what a point or stamp is worth, how to redeem, whether rewards expire, whether a balance transfers between locations or people, and what happens if you change or close the scheme. Publishing it protects you in a dispute, removes a question your staff keep answering, and gives the people who research before they commit something to find. If you sell online as well, the same decisions can be expressed in your product markup: Google's merchant listing documentation, checked on 12 October 2026, defines validForMemberTier for a member price and membershipPointsEarned for points earned on a purchase, described on the merchant listing structured data page. Marking member pricing properly is what lets a shopping surface show the member price rather than the standard one.
The data, handled so it stays an asset
Collect the minimum that makes the scheme work, which is usually a name, one contact method and the purchase history. Keep joining the scheme and agreeing to marketing as two separate decisions with two separate ticks, record when and how each was given, and make leaving a one click job. Never store payment card details to identify a customer. Set a retention period for inactive accounts and apply it. A clean list of two thousand people who chose to hear from you is worth more than a dirty list of twenty thousand, and it is the only kind you can email with confidence.
What to do with the data in the first year
- Find the lapse point. Work out the normal gap between visits, then build one message that goes out when a customer passes it by half again.
- Find the second purchase. For most businesses the biggest single lift available is moving first time buyers to a second purchase, so give that its own offer and measure it on its own.
- Find the top decile by spend and ask them something. A short question to your best customers produces better product decisions than a survey of everybody.
- Find the products that appear in high frequency baskets and put them where new customers see them.
- Report four numbers monthly: enrolment rate at the till, active share of enrolled, redemption rate, and visits per enrolled customer.
Loyalty card questions that come up at the till
How many stamps should a card have?
Set it so a regular customer completes a card inside their normal buying cycle, which for a daily purchase means a week or two and for a monthly purchase means a season. A card that takes longer than the customer's own planning horizon gets abandoned, and an abandoned card teaches them that your scheme is not for them.
Should rewards expire?
A stated expiry keeps the liability on your books finite and gives you a reason to contact people, so set one, make it generous, and say it on the terms page and on the card itself. Then honour a near miss at the counter when somebody has been loyal, because the goodwill from that is worth more than the item.
Is a discount or a free item better?
A free item usually wins. It has a fixed cost you control, it feels like a gift rather than a price cut, and it does not retrain customers to wait for a discount before they buy. Keep percentage discounts for clearing stock and for the paid membership mechanic, where a standing discount is the product.
Can a small shop run this without software?
Yes, and plenty do it well with printed cards and a stamp. Accept that you are buying frequency rather than data, keep the design simple enough to reprint cheaply, and put the business name and one contact detail on the card so it also works as a reminder in a wallet.
How do I move an existing paper scheme onto a system?
Run both for one cycle. Honour every paper card at face value, offer a bonus for moving a balance across, set a closing date for paper, and tell staff what to say at the till for each case. Announce the change on the terms page and keep the old terms visible until the last paper card has expired.
What a shopper hears when they ask whether your scheme is worth joining
People research a loyalty scheme before they hand over an email address, and increasingly that research is one question put to an assistant: is this shop's card any good, what do you get, is there a catch. The answer gets built from whatever is readable, which means your terms page if you have one, and a forum thread, a review or an out of date blog post if you do not. A business whose scheme details sit inside an app and nowhere else has handed that answer to strangers. Publishing the terms fixes the supply side of it, and then it is worth knowing what is actually being said: AI Knows Us reads the answers assistants give about a named business and shows which pages and reviews each answer was built from, which is how you find the three year old thread that is still doing your talking.