There are two costs in running a loyalty program. Most people only count the smaller one.
Cost one: the rewards
This is the real expense and it scales with success.
A ten stamp card giving away one free item costs you the margin on that item every time a card completes. Work it out on the member's total spend rather than on the item's retail price, because that is the number that tells you what percentage you are discounting.
A worked example. A cafe with a £4.20 average order and a coffee that costs £1.30 in goods.
A member fills a ten stamp card. They have spent about £37.80 across nine paid visits. You give away £1.30 of cost. That is 3.4 percent of what they spent.
Now the same cafe on a six stamp card. Five paid visits, £21 spent, £1.30 given away. That is 6.2 percent.
And on a five stamp card: four paid visits, £16.80, £1.30 away. 7.7 percent.
Sensible territory for most hospitality and services is between 5 and 12 percent. Below 5 percent customers do not notice the program exists. Above 12 percent you are usually subsidising people who were coming anyway.
Note what moves that number. It is the threshold, not the reward. Changing what you give away is a blunt instrument. Changing how many stamps it takes is precise.
The part people forget
A share of every reward goes to customers who would have visited regardless. That is unavoidable and it is the cost of the mechanic, not a design flaw.
What you can control is how much of your reward budget goes to changing behaviour rather than to thanking existing habits. A bonus stamp on your quietest day sends the discount toward visits you would not otherwise have had. A flat card sends it wherever customers already were.
Cost two: running it
Software. For a single site independent, loyalty platforms generally sit between free and about £50 a month. Free tiers exist and are a reasonable way to test whether customers engage at all. The limits usually appear as a cap on members, no customer messaging, or a single location only.
Multi site, marketing automation and integrations push it higher. StampClub's pricing is public if you want a concrete number to anchor against.
Printing. QR posters, counter cards, window stickers. Under £50 once, then occasional reprints.
Paper cards look cheaper here and are not, once you count reprint runs and the cards customers lose. A lost card is a customer who stops collecting, and that is a cost that never appears on any invoice.
Staff time. Roughly an hour of training at launch, then a few seconds per transaction. The second part is small but it is not zero, and it is the thing that quietly kills badly designed programs. If joining takes ninety seconds at a busy counter, staff stop offering it, and you conclude that customers were not interested.
Your time. An afternoon of decisions up front, then perhaps thirty minutes a month checking numbers and sending a message. Programs that get no attention drift, so budget the thirty minutes rather than pretending it is zero.
Putting it together
A cafe with 400 active members, each spending about £35 a month with them.
Member spend: £14,000 a month. Reward cost at 3.4 percent: about £476. Software: £29. Printing amortised: about £5. Total: roughly £510 a month, of which 93 percent is the reward.
That ratio is typical. Arguments about software pricing usually occupy far more attention than they deserve relative to reward design, which is where nearly all the money is.
Working out whether it pays
The honest calculation has three steps.
One. Take visits per member per month in the two months before joining, and the two months after. The difference is your uplift. If members averaged 3.1 visits before and 3.6 after, that is 0.5 extra visits per member per month.
Two. Multiply by your gross margin per visit. At a £4.20 order and 68 percent margin, that is £2.86 per visit. So 400 members × 0.5 × £2.86 gives about £572 of extra margin a month.
Three. Subtract total program cost. £572 minus £510 leaves £62 a month.
Thin, but positive, and this example deliberately uses a conservative uplift. The point is not the answer. It is that the calculation is doable with numbers you already have, and almost nobody runs it.
If it comes out negative, the fix is usually the threshold. Push the card from five stamps to eight and the reward cost roughly halves while the motivation barely changes.
What is not in the calculation
Two things, both real and both hard to price.
The customer list has value beyond the visit uplift. Being able to contact four hundred people directly is worth something on the day you need to fill a quiet week, and it costs nothing per message.
And you learn things. Which day is genuinely quietest, which customers stopped coming and when, whether your Tuesday promotion did anything. Most independents run on impressions until they have this data.
The cheapest version that still works
If budget is the constraint, this is the order to spend in.
Start on a free tier with a paper backup. Print one QR poster. Set a ten stamp threshold, which keeps reward cost near 3 percent. Skip every feature that is not stamping and joining.
Then upgrade for one reason only: when you have enough members that being able to message them is worth more than the fee. For most cafes that point arrives somewhere between two and three hundred members.
To design the program itself, how to create a loyalty program covers the threshold maths in detail.