Guides

How to run a loyalty program that actually works

Written for people who work a counter — cafés, barbershops, salons, workshops. By the end you will be able to decide whether a program suits you, with which mechanic, what goal and what reward, and what to measure to know if it is working.

Reviewed 2026-08-04 · 12 min read

A loyalty program works when the goal is reachable at your customers' real visit rate and the reward costs you less than the extra visits are worth. Almost every program that fails does so for one of three reasons: the goal was unreachable, staff never offered it, or nothing was measured except sign-ups.

First: should you run one at all?

The question that settles it

A loyalty program rewards repetition. If your typical customer comes back once or twice a year, there is no repetition to reward — any card you design takes years to finish and gets abandoned long before.

The quick test: does your average customer visit at least monthly? If yes, a program makes sense. If they come every six months, your problem is being remembered, not retention, and that is solved with content and reminders rather than stamps.

  • Good fitCafés, bakeries, barbershops, nail salons, car washes, dry cleaners, takeaway, gyms and studios.
  • Poor fitDentists, opticians, furniture shops, estate agents, major auto repair — visits spaced months or years apart.
  • DependsHigh-ticket restaurants, tattoo artists, photographers: low frequency, but a high ticket can justify recognition of a different kind, not a stamp card.

If your business is a poor fit, no vendor is going to tell you. Selling loyalty software to a dentist is entirely possible; making that program work is not.

What a program does not fix

It does not fix a mediocre product, a dirty room or a forty-minute wait. Most customer loss in small businesses is not a competitor winning — it is one bad visit.

It also does not bring new customers. A loyalty program operates on people who already walked in. If nobody is walking in, this is not what you need yet.

Choosing the mechanic

Stamps count visits

One visit, one stamp, fixed goal, fixed reward. Nothing to calculate and nothing to explain, which is why it survives on paper in thousands of businesses that never adopted anything else.

Its flaw is that it ignores value: someone buying an espresso and someone buying four pastries earn the same. It only makes sense where ticket sizes cluster.

Points count spend

Points scale with the bill, which fixes the fairness problem. The cost is comprehension: almost no customer can tell you what their balance is worth without looking it up, and a reward that needs arithmetic gets ignored.

They also carry accounting weight. Unredeemed balances are an open liability that grows quietly until someone cashes the lot.

Cashback returns a percentage

The easiest to understand and the weakest at changing behaviour, because it is a discount with a delay. It rewards the purchase the customer was already making.

It has a legitimate use: if your aim is getting them back to spend the credit and you do not mind that it is a discount, it works. Just call it that in your own numbers.

Running two mechanics at once almost always backfires: it is two explanations at the counter, and staff end up explaining neither. Pick one, run it a quarter, change it if the numbers say so.

The goal and the reward

Setting the goal

Match it to real frequency, not to a round number. The working rule: the reward should be reachable in about a month of normal behaviour for frequent-visit businesses, and about three months for fortnightly ones.

Ten stamps at a café someone visits twice a week is a month — fine. Ten stamps at a nail salon, with a fill every three weeks, is seven months: nobody gets there.

  • Café or bakery8 to 10 visits, if customers come several times a week.
  • Barbershop6 visits, on a four-week haircut cycle.
  • Nails or hair colour5 visits, on cycles of two to eight weeks.
  • Food truck or stall6 visits: customers see you fewer times a month than a fixed location.

Pricing the reward

The reward costs you its marginal cost, not its menu price. A free coffee costs you beans and milk; the customer values it at what the menu says. That gap is the entire economics of the program.

Define the reward as a specific product, not free choice. 'A free coffee' and 'anything from the case' have very different costs, and the customer will always pick the expensive one — quite reasonably.

A percentage discount is a bad reward. '10% off on your tenth visit' asks the customer to do arithmetic to find out what they won, and the answer is usually 'not much'. A free product explains itself.

The quick sum

Multiply: extra visits per month × average ticket × gross margin = extra profit. Subtract: rewards earned per month × marginal cost of the reward. If the result is positive, it pays.

The fragile number is 'extra visits'. A realistic lift is between 2% and 8%. If your sum only works assuming 20%, you do not have a profitable program — you have an optimistic spreadsheet.

The counter is where it is won or lost

Enrol in seconds or not at all

Enrolment happens mid-transaction, with people waiting. Every extra field, confirmation email and app install loses a share of customers.

If joining requires downloading an app, most will decline while there is a queue behind them. It is the number one cause of programs sitting at ten members after three months.

Staff decide whether it exists

A program nobody mentions at the counter gets no sign-ups regardless of how good the software is. And staff stop offering it the moment it gets in the way: if the flow is slow, at peak hours they simply will not reach for it.

Give them one sentence, not an explanation. 'Want to collect stamps? Scan here, it's free' works. 'We have a new loyalty programme that lets you…' does not.

  • Measure the offer rateAsk your team how many times they offered it yesterday. If nobody knows, it is not being offered.
  • Remove friction before assigning blameIf staff are not offering it, it is nearly always because it is slow or fails, not because they do not care.
  • Scan directionThe business should scan the customer, not the reverse: a code taped to the register can be photographed and used from anywhere.

Knowing whether it works

Sign-ups are not the number

Total enrolments flatter every program ever built and tell you nothing: they rise as long as new customers exist, even if none of them come back.

The number that matters is visit frequency among enrolled customers, before and after. If your regulars came every ten days and now come every eight, it works. If they come at the same rate, you are giving product away to people who were coming anyway.

  • TrackVisit frequency of enrolled customers · redemption rate · how many abandon mid-card.
  • IgnoreTotal sign-ups · stamps issued · 'engagement'.

When to change something

Give it a quarter before touching anything; less than that and you are reading noise. After that: if many people stall mid-card, the goal is too high. If everyone finishes quickly and frequency did not move, the goal is too low and you are giving away margin.

Change one thing at a time. Move the goal and the reward together and you will not know which one worked.

The five ways to ruin it

In order of frequency

None of these is a software problem, which is why switching platforms rarely fixes a program that is not working.

  • An unreachable goalThe card outlasts the customer's patience. Most common failure and the easiest to avoid.
  • A reward not worth the visitsCustomers do the arithmetic. If ten visits earn less than a normal discount, they stop collecting.
  • Requiring an app installYou lose most people at the counter, and staff stop offering it.
  • Not training the teamA program nobody mentions does not exist, however good the mechanic.
  • Measuring sign-upsIt convinces you it is working while frequency stays flat.

If you are about to switch platforms because your program is not working, check which of these five is your case first. Four of the five will follow you to the new platform.

Frequently asked questions

How many stamps should my card require?

As many as your average customer can collect in about a month if they visit often, or three months if they visit fortnightly. Roughly 8 to 10 for a café, 6 for a barbershop, 5 for nails or hair colour.

Points or stamps?

Stamps if your ticket sizes are similar, points if they vary a lot. It is fit rather than preference: stamps ignore how much each person spent, which is simple and also unfair when the spread is wide.

Do I need an app or a POS?

Not necessarily. QR-based programs work with a phone on each side. POS-integrated ones are more accurate because they see the transaction, at the cost of requiring that POS.

How long before I can tell if it works?

A quarter at minimum. Before that you are reading noise, and the number to watch is visit frequency among enrolled customers, not how many enrolled.

What if my business is low-frequency?

A stamp card probably does not suit you. At one or two visits a year, any reasonable goal takes years. No vendor will tell you this, but it is the honest answer.

Guides age. This one carries the date it was last reviewed so you can judge how current the advice is — if that date is far in the past, weigh it accordingly.

Once the mechanic is settled, setting it up takes two minutes

Stamp cards on Bioflow Pro — no customer app, no POS.