In brief
The type is decided by three things you already know: how often the same person comes back, what a visit is worth, and whether your till can talk to the program. Most local businesses land on stamps.
The short answer
Three inputs decide the type: how often a typical customer comes back, how much a visit is worth and how much that varies, and whether your point of sale can hand a transaction amount to the program without anyone retyping it. Answer those three and the type picks itself, usually a stamp card.
The six types and what each one is are already set out on types of loyalty programs. This post does not define them again. It chooses one.
The three inputs
Everything else people ask about at this stage is downstream of these. Get them written on a single line before you look at any product.
How often the same person comes back
Not footfall. The interval between one named customer’s visits.
A program is a device for shortening that interval. If the interval is a week, there is room to shorten it. If it is fourteen months, there is nothing to work with: the customer will not be holding a card, will not remember joining, and will make their next decision from scratch.
Estimate it from what you can see. If your regulars come in most weekdays, call it 20 a month. If they come when they need a haircut, call it every five weeks. A rough number is enough, because the decision boundaries are far apart.
What a visit is worth, and how much it varies
Two numbers, and the second is the one people forget.
Average ticket sets what you can afford to give away. The spread between your smallest and largest ticket decides whether counting visits is fair. A cafe where every transaction is between £2.80 and £9 can count visits and nobody feels cheated. A wine shop where one customer spends £11 and the next spends £180 cannot: a visit-based card hands the same reward to both, and the £11 customer is the one collecting fastest.
Wide spread is the single strongest argument for counting spend instead of visits. It is also the argument that runs straight into the third input.
Whether your till can pass a number to the program
Counting spend means something has to know the amount. Either the till sends it to the program automatically, or a person types it in.
A person typing it in is not a workable long-term plan. It adds seconds to every transaction at exactly the busy moments the program was built for, it is wrong sometimes, and it stops happening within a month because the queue is real and the program is optional. This is the input that eliminates the most options, and it is a yes or no question you can answer today by asking whoever sold you the till.
The decision path
Run these in order and stop when one of them stops you.
- Does a typical customer come back at least monthly? If not, skip the program. Money spent on a well-timed reminder will do more, and how loyalty programs work explains why a threshold nobody reaches pays out nothing to anyone.
- Can the person on the counter recognize a qualifying visit in one second, unaided? If yes, a stamp card is available to you. Keep it on the list.
- Does the till pass the transaction amount to the program with nobody retyping it? If no, points and cashback are out. Not “harder”, out. The manual version does not survive contact with a lunch rush.
- Is your heaviest regular worth more than about three times your median regular? If no, tiers are out. That multiple is a judgment rather than a measurement, but the reasoning holds at any threshold you prefer: tiers exist to treat genuinely different customers differently, and where the difference is small the tiers just tell most of your customers they are at the bottom.
- Would customers pay a fee up front, and is the thing they would be buying capped by something physical? Both, or paid membership is out.
- Whatever is still standing is your type. For most local businesses that is one option, and it is the stamp card.
Step 3 is where the most disappointment happens, because points sound more sophisticated and the demo always looks fine. The demo has one customer in it.
Frequency against ticket size
The same decision as a grid, if you would rather read it that way. Find your row, then your column.
| Typical return interval | Small, consistent ticket | Large or highly variable ticket |
|---|---|---|
| Weekly or more | Stamp card, 8 to 10 stamps | Points if the till passes the amount; otherwise a stamp card with a minimum spend per stamp |
| Roughly monthly | Stamp card, 5 or 6 stamps | Spend-based reward, or a service plan on a schedule |
| A few times a year | No program. Use a reminder at the right interval | No program. A referral offer, or a paid plan if the work is already scheduled |
The bottom row is not a gap in the table. It is the answer for garages, opticians, furniture shops, wedding suppliers and anyone else whose customer has no reason to return this quarter.
The minimum-spend variant in the top right is worth a sentence. Adding “one stamp per visit over £6” to a visit-based card recovers most of what a wide ticket spread breaks, without needing the till to send anything anywhere. It costs you one extra rule for staff to remember, which is the real currency at a counter.
Why most local businesses land on stamps
Three of the six types need infrastructure. Points and cashback need a till that can pass an amount. Tiers need a wide spread of customer value to sort. Coalition arrangements need several businesses and a way to settle up between them.
That leaves stamp cards, referral, and paid membership, and the last two are not really substitutes: referral pays for someone else’s first visit, and a membership asks for money before you have proven anything. Both are things to add later. A stamp card is the only one of the six that a single shop with one counter can run properly on day one.
This is a narrower conclusion than most “types of loyalty program” articles reach, and it is narrower on purpose. A menu of six equals is only useful to a reader who has the infrastructure for all six.
If the choice you are actually weighing is stamps against points, the trade is set out in stamp cards versus points, and the conversion-rate problem points create is in how loyalty points work.
What tiers and coalitions need that you do not have
Tiers need enough customers per level for a level to mean something, and enough spread for the levels to describe a real difference. They also need permanence: a customer who reaches gold and is demoted in March has been punished by a program that was sold as a reward. Every tier is a promise with a maintenance requirement attached.
Coalitions, where several businesses share one card, fail on settlement rather than on software. A customer earns at the bakery and redeems at the barber. Money now has to move from the bakery to the barber, on some agreed basis, every month, forever. Three neighboring shops can start this on goodwill and a spreadsheet. The goodwill lasts until the first month where one shop issues most of the stamps and another honors most of the rewards, which will be the second month.
Both are real mechanics that work at real scale. Neither is a first program.
What makes it a good program once the type is picked
Four tests. A program passing all four beats a cleverer one failing any.
- A customer can state the offer from memory. If the sentence needs a comma and a conversion rate, it will be repeated wrongly or not at all.
- Whoever is on shift can validate a visit and honor a reward alone. Not the owner, not a manager on call. The reward moment is the only part of the program a customer will describe to anyone else, and it happens on a Saturday.
- The reward costs less than your margin on the visits that earn it. This is arithmetic, not judgment, and it is worked through in what a loyalty program returns.
- You can read the ledger. In six months you will want to know how many cards were started, how many completed, and whether those people come in more often. A program you cannot answer that from is a cost with no feedback attached.
Three of those four are settled before any software is chosen.
What to do with the three inputs
Write one line: visits per month, average ticket with its range, and yes or no on the till. Take that line to the grid above.
If it lands on stamps, the next decision is the threshold, and unlike the type it changes the outcome. Set it so the reward arrives within six to ten weeks at the rate the customer already visits, then leave it alone for a quarter. Plan a visit-based loyalty programme walks through that number and the four others it depends on.