In brief
A restaurant runs two or three businesses out of one kitchen, and a single undifferentiated stamp card subsidizes whichever one already works. The daypart decides the threshold, the reward and the redemption rule.
The short answer
A restaurant loyalty program should count the daypart you want more of, not every visit that happens to occur. Lunch and dinner are different customers on different margins, and one undifferentiated card pays the same reward to both while only one of them needed persuading. Set the threshold against the visit rate of the daypart you are targeting, choose a reward whose food cost you can name, and decide before launch which nights the reward can be spent on.
That last decision moves the cost of the program by a factor of three. Most of this post is the arithmetic behind it.
The assumptions
Every number below comes from this menu. Substitute your own; the method is the point.
- A lunch main sells for £11.00 and costs £3.30 in food.
- A dinner main sells for £18.00 and costs £5.40 in food.
- A dessert sells for £6.50 and costs £1.60 in food.
- The average dinner party is four people on one bill.
Gross margin, therefore: £7.70 on a lunch main, £12.60 on a dinner main, £4.90 on a dessert.
Lunch and dinner are the same number, twelve months apart
A lunch regular who comes twice a week makes 8.7 visits a month. A dinner customer who comes every six weeks makes 8.7 visits a year. Same figure, one twelfth the rate.
That is the daypart problem in one line, and it is what separates a restaurant from a cafe. A cafe has one visit rate and can set one threshold against it. A restaurant has at least two, and they are an order of magnitude apart. The reference version of target setting for food service is on the cafes and restaurants page; this post is about what happens when the page’s advice has to serve two rates at once.
A ten-stamp card is three and a half weeks for the lunch regular and fourteen months for the dinner customer. Issue the same card to both and you have designed a working program for the people who were already coming and a dead one for the people you wanted.
Reward the visit you want more of
The qualifying event is the decision, not the threshold. If Tuesday dinner is the shift you need to fill, the thing you count is Tuesday dinner.
A card that stamps any visit gives your best lunch customer nine rewards a year and your dinner customer none. That customer was coming anyway, so you have converted a retention tool into a standing discount for existing behavior. The qualifying event is where the program is actually designed, and in a restaurant it should carry the daypart in it.
Scoping the count has a cost worth naming. A card that only counts dinner is invisible to your lunch trade, and lunch is where the volume is. The usual compromise is one card that counts every visit with a reward scoped to the daypart you want: the reward is a dinner main, redeemable Monday to Thursday, and the lunch customer has to come back at night to spend it. That is a design that uses the frequency you have to buy the visit you do not.
A free main costs three times a free dessert
Take a six-stamp lunch card. The customer pays for six lunches and the seventh main is free.
| Reward on the seventh visit | Customer paid | Reward at menu price | Share of their spend | Your food cost | Share of your takings |
|---|---|---|---|---|---|
| A lunch main | £66.00 | £11.00 | 16.7% | £3.30 | 5.0% |
| A dessert | £66.00 | £6.50 | 9.8% | £1.60 | 2.4% |
The dessert costs you less than half as much. It also buys more goodwill per pound spent: £6.50 of menu price for £1.60 of food is a ratio of 4.1, against 3.3 for the main. Divide any candidate reward’s menu price by its food cost and pick the highest number. On most menus that is a dessert, a side or a soft drink, because the main is the item carrying the protein and the worst ratio on the board.
The free main has one real advantage, and it is not a small one: it is the thing people actually came for, so it is the reward most likely to produce the visit. Weigh that against paying three pounds thirty instead of one sixty every time a card completes. Both are defensible. Picking without doing the division is not, and the full budget model, with software and staff time on it, is in what a loyalty program actually costs.
Who gets the stamp when four people share a bill
Only the person holding the phone. That is the constraint, and everything else follows from it.
One stamp per bill is the workable rule, and it has a consequence nobody mentions: your discount rate is divided by your average party size. On the six-stamp card above, the solo lunch customer spends £66 to earn an £11 main, which is 16.7% of their spend. A dinner table of four on the same card spends four times as much per stamp. Six stamps is six bills at £72, so £432 of revenue for one £18 main. That is 4.2%.
Same card, same rule, and the group customer is getting a quarter of the deal the solo customer gets. They will not calculate it, but they will feel that the card is not going anywhere, because in their experience six dinners is most of a year.
Per-head stamping fixes the rate and breaks the service. Four phones at the table while the card machine is out is not a thing a restaurant does on a Friday. Two rules that work instead:
- Set the dinner threshold low enough that party size cannot kill it. Three or four stamps, not eight.
- Put a minimum spend on the stamp rather than counting heads. A bill over £40 earns one stamp; a bill over £80 earns two. It keeps the count on the bill, where your team can see it, and it stops a table of six earning the same credit as a solo diner. A stamp card cannot scale credit with spend the way a points program can, which is the honest limit of the mechanic and is set out on stamp cards versus points.
Quick service and casual dining need different numbers
| Quick service, counter lunch | Casual dining, table service | |
|---|---|---|
| Typical visit rate | 2–3 times a week | Every 4–6 weeks |
| Stamps | 8 | 3–4 |
| Reward lands in | About 3 weeks | About 5 months |
| Reward | A main from the lunch menu | A dessert or a side, per person at the table |
| Stamp rule | One per customer per day | One per bill, minimum spend |
| Redeemable | Any weekday | Monday to Thursday |
Read the third row honestly. Even a four-stamp dinner card is five months from the reward, which is well past the point where people keep track of a card. There is no threshold that fixes this, because the constraint is the customer’s appetite for eating out, not your program design.
So for a dinner-only restaurant, a stamp card is a weak instrument. What works better is a card that counts every daypart, plus a targeted offer to the customers whose gaps are getting longer. That means picking the audience from card activity rather than from a mailing list, which is the only way to send a Tuesday offer to people who have stopped coming on Tuesdays instead of to everyone on the list.
The same plate costs £5.40 or £18.00
A free dinner main costs you the food when the table was going to be empty and the full menu price when it was not.
On a Tuesday with tables free, you cook a main for £5.40 and collect nothing. That is the whole cost: £5.40.
On a full Saturday, you cook a main for £5.40 and collect nothing, and the party you turned away would have paid £18.00 for the same plate. You would have spent the £5.40 on food either way. The difference between the two Saturdays is the £18.00 you did not take.
Same reward, three and a third times the cost, decided entirely by a redemption rule you can write in one sentence. “Rewards can be used Monday to Thursday” is that sentence, and it belongs on the card before launch, not added later when someone notices the Saturday covers.
Against that, the software is a rounding error. NeoLoyal’s Starter plan is £9 a month, and one extra dinner main carries £12.60 of gross margin, so a single extra cover a month pays for it with £3.60 to spare. The rewards line is the one that grows when the program works, and it is the only one worth arguing about.
Do this before you set a threshold
Open last month’s till or booking report and split the covers by daypart. Divide each daypart’s covers by the number of days it ran, then divide that by your number of regulars, and you have two visit rates instead of one.
The larger number tells you the threshold for a card. The smaller number tells you whether a card is the right tool at all: anything below about one visit a month is a problem a stamp card cannot reach, and a visit-based program plan will tell you the same thing in more detail. Then price two candidate rewards by dividing menu price by food cost, and take the bigger ratio.
Figures in this post are a worked model with the assumptions stated at the top, not a survey. NeoLoyal prices are current as of publication. Run the same arithmetic on your own menu and your own covers; the numbers will differ and the method will not.