A loyalty programme is not a nice gesture. It is a deliberate trade: you give up some margin on an order you were going to get anyway, in exchange for orders you would not otherwise have had. Whether that trade is good depends almost entirely on how you configure it, and most places never revisit the settings after the first week.

Here is how to think about each one.

Why the paper punch card fails

Not because it is paper. Because it rewards a behaviour that was already happening and measures nothing. You cannot see who is close to a reward, who has stopped coming, or whether the tenth coffee cost you more than it earned. It also lives in a wallet, which means it is with the customer only by accident.

The point of doing this digitally is not the convenience. It is that every one of the decisions below becomes measurable.

1. Pick an earn model

There are really only two, and they suit different businesses.

Spend-based. Points per rupee. Better where ticket sizes vary a lot, like a restaurant where one order is a coffee and the next is dinner for six. It rewards value rather than frequency.
Visit-based. Points per visit, regardless of spend. Better for a café where the ticket is small and consistent, and what you want is frequency. The risk is somebody buying the cheapest thing on the menu ten times to earn a reward, which you fix with a minimum spend.

2. Set the earn rate with actual arithmetic

This is the setting people guess at. Do not guess. The question you are answering is: what percentage of revenue am I giving back?

Say a customer earns 1 point per ₹10 spent, and 100 points redeems for ₹50 off. They spend ₹1,000 to earn 100 points, which is worth ₹50. You are giving back 5%. That is your real number, and it should be compared against your gross margin, not against your revenue.

For most food businesses, somewhere between 2% and 5% is defensible. Above that and you need to be confident the programme is genuinely changing behaviour rather than discounting people who were coming anyway.

3. Award points on completion, never on placement

This sounds like a detail and it is not. If points land the moment an order is placed, they survive cancellations and refunds, and you have created a way to mint points without buying anything. Award them when the order is actually delivered or served.

4. Cap redemption per order

Without a cap, a customer can save up and take an entire order for free. That feels generous until you notice the order had food cost attached and brought in no revenue to cover it. A common approach is to allow points to cover a set share of the bill, so every order still contributes something.

A minimum spend to redeem does similar work and is easier to explain to customers.

5. Exclude what you cannot afford to discount

Every menu has items running on thin margin: a loss-leader combo, something priced against a competitor down the road, an item where the ingredient cost moved and the price has not caught up. Those should not earn points and should not be redeemable against.

This is the setting that most often turns a losing programme into a working one, and it takes ten minutes to configure.

6. Decide what expiry is for

Expiry has a bad reputation and a legitimate use. Points that never expire are an open liability on your books and they stop creating urgency. Points that expire in a month feel like a trick.

Somewhere in between, clearly stated up front, with a reminder before it happens, is both fair and effective. The reminder is the part that does the work: "you have 240 points expiring this month" is one of the highest-performing messages any food business can send.

The part most places skip: doing something with it

A loyalty programme that only sits there and accrues points is half a programme. The value is in what it lets you notice.

Customers who have gone quiet. Somebody who ordered weekly for six months and has not been seen in three weeks is the single most winnable customer you have. You cannot spot that without a record.
Customers close to a reward. A nudge to somebody sitting at 240 of 300 points converts far better than a discount blasted to everybody.
Your actual best customers. Usually not who staff assume. Ranked by spend over a year, the list is often surprising, and worth treating differently.

The important discipline is aiming. An offer sent to 38 people who have stopped ordering is a retention campaign. The same offer sent to all 4,100 is a price cut for people who were about to pay full price.

How to tell whether it is working

Three numbers, checked monthly:

Repeat rate. The share of orders coming from someone you have served before. If this is not moving, the programme is not doing its job.
Orders per customer per month. The metric loyalty is supposed to move directly.
Cost of the programme as a share of revenue. Points issued and redeemed, against sales. If this drifts above the number you set in step two, something is misconfigured.

A programme that raises repeat rate while holding its cost near your target is working. One that raises neither is a discount, and you should either fix the settings or stop.

Retova includes loyalty, coupons and targeted offers with all of the settings above under your control, alongside ordering and operations. See restaurants,cafés, or the guide to QR ordering. When you want to see it against your own numbers,get in touch.

How to build a loyalty programme that actually pays