Ask an owner how their loyalty program is doing and you usually get a story rather than a number. The reason is that most programs are unmeasurable by design: a card in a wallet, stamped by whoever's on the till, redeemed when someone remembers to bring it. There's no way to know how many were issued, how many came back, or whether the ones who came back would have anyway.
The last question is the whole game, and almost nobody asks it. A discount handed to somebody who was already walking through the door on Friday isn't loyalty. It's margin you gave away for company.
Start with the number that matters
Independent restaurants tend to obsess over new customers and quietly lose the ones they already have. The arithmetic runs strongly the other way.
Take a modest venue: two hundred covers a week, average bill four hundred rupees. If a quarter of your diners visit twice a year instead of once, that's fifty extra visits a week, twenty thousand rupees a week, roughly ten lakh a year — from people who already know where you are and already like the food. To find the same revenue in new customers you'd be paying for reach, discounts and the usual attrition on the way.
Second visits are the cheapest revenue in the building. They're also the only kind you can affect on a Tuesday afternoon.
So the thing to measure is not enrolments. It's the share of diners who come back within ninety days, before and after you start. If your program can't tell you that, it isn't a program, it's stationery.
Points, visits or tiers
Three models, and the right one depends almost entirely on how wide your bill sizes are.
| Visit-based | Spend-based points | Tiers | |
|---|---|---|---|
| Best for | Cafés, casual, consistent spend | Wide range of bill sizes | High-frequency, higher spend |
| Diner understands it | Immediately | Usually | Needs explaining |
| Rewards big tables fairly | No | Yes | Yes |
| Drives frequency | Strongly | Moderately | Strongly, at the top |
| Drives bill size | No | Yes | Yes |
| Admin load | Low | Low | Real |
For most independents the honest answer is visit-based, because it's the one a diner can hold in their head. "Fifth coffee free" needs no explanation and no mental arithmetic at the counter. Points work when a solo lunch and a table of six shouldn't earn the same thing, which is true in most full-service restaurants and false in most cafés.
Tiers are excellent and almost always premature. They need enough regulars for the top tier to feel like something, and enough discipline to keep the benefits real. A three-tier structure at a venue with forty regulars is a spreadsheet pretending to be a strategy.
What the reward should be, and what it should cost
Two rules do most of the work here.
Price it on food cost, not menu price. A dessert that sells for ₹200 and costs you ₹50 to make is a ₹50 reward that the diner values at ₹200. A ₹50 discount off the bill is a ₹50 reward the diner values at ₹50. Same cost to you, four times the perceived value. Give away food, not money — and give away a specific thing rather than a percentage.
Keep the total give-away around five to eight percent of qualifying spend. Below that it's not worth changing behaviour for. Much above it and you're funding visits that were happening anyway. Do the sum before you launch: a free ₹200 dessert after five visits averaging ₹400 is ₹50 of cost against ₹2,000 of spend, which is two and a half percent — comfortable, arguably too cautious.
Make the first stamp free. A card that starts at one of six instead of zero of five gets meaningfully more completions, for the same number of purchases. People finish things they've started.
Don't make it expire quickly. A reward that lapses in thirty days reads as a trick, and the goodwill you lose from one expired reward outweighs the cost of honouring it. If you must have an expiry, make it generous and remind them before it hits.
Where the ethics line sits
Loyalty and reviews are two systems that should touch each other as little as possible, and this is the single most common way restaurants get themselves in trouble with Google.
Offering loyalty points, a free item, a discount or a prize draw entry in exchange for a review is an incentivised review. Google's policies prohibit it. It doesn't matter that you're offering it for any review rather than only a positive one — the incentive is the problem, because it changes who reviews and what they say. Reviews collected that way can be stripped, and a pattern of it puts the Business Profile itself at risk.
Reward people for coming back. Never reward them for what they write.
Joining your loyalty club and leaving a Google review are both things a diner might do after a good meal. Wiring the second to the first turns your review profile into paid content, and both Google and consumer regulators treat it as exactly that.
The same reasoning rules out the softer versions: a discount for showing a five-star review at the counter, a draw entry for "tagging us", points for a screenshot. If the reward is conditional on public praise, it's bought praise.
Consent, and the messages you send
A loyalty program is a list of phone numbers, which makes it a data responsibility as much as a marketing one.
Collect an explicit opt-in when the diner joins — a clear tick or tap, not a pre-checked box and not "by dining here you agree". Say what you'll send and roughly how often. Keep a record of when consent was given, because that record is the thing that protects you if it's ever questioned. Honour opt-outs immediately and permanently, and make opting out a single tap rather than a reply-with-the-word-STOP maze.
In India this sits alongside telecom rules on commercial messaging, and WhatsApp has its own policies on business messaging that are enforced by account restriction rather than by anybody writing to you first. A number lifted from a booking sheet or a delivery order is not a number you have permission to market to.
On frequency: fortnightly is plenty for most restaurants, and the venues that get unsubscribed from are almost always sending weekly. One good message a month that mentions something real — a new dish, a menu change, a genuinely useful offer on your quietest night — beats four that say "we miss you".
Lapsed, not lost
The highest-return message in a restaurant's entire marketing is the one to somebody who used to come every fortnight and hasn't been in for two months. They already like you. Nothing needs to be sold. Something just interrupted the habit — they moved desk, the usual Friday group broke up, they tried the new place.
Which means the useful thing a customer list gives you is segments rather than a single broadcast.
- Lapsed regulars. Three or more visits, nothing in sixty days. The most valuable list you have.
- One-and-done. One visit, over thirty days ago. A different message: they haven't formed a habit, so give them a reason rather than a reminder.
- Frequent, high spend. Not a discount audience. Invite them to the tasting night, hold them a table on the busy Saturday, use the name.
- Recent birthdays. Only if you collected the date honestly and the message doesn't read as automated.
Send the lapsed-regular message on a Tuesday or Wednesday, mid-afternoon, for a visit later that week. Weekend nights are full anyway; the point of the exercise is filling the quiet ones.
Referrals, which are the underrated half
A referral is a loyalty mechanic wearing different clothes, and it outperforms most discounting because the recommendation carries a person's own credibility with it.
Keep it simple enough to explain in one line: your regular sends a code, the friend gets something on their first visit, the regular gets something when that visit actually happens. Reward on the completed visit rather than the sign-up, or you're paying for names instead of covers.
And keep the friend's reward at least as good as the referrer's. A scheme where the existing customer profits and the newcomer gets a token feels like being sold to by a friend, which costs your regular something socially. That's the fastest way to have people stop using it.
What to actually do this month
- Pick one model — visit-based unless your bill sizes vary widely — and one reward priced on food cost.
- Write down your ninety-day return rate today, so you have a before.
- Collect consent properly at join, with a real opt-in and a working opt-out.
- Build one segment: regulars who've gone quiet for sixty days. Message them once.
- Keep reviews entirely out of it. No points, no draws, no discounts for what somebody writes.
- Look at the return rate again in three months and change one thing.
Almost every failed loyalty program failed at the measurement step rather than the design step. If you know your return rate, an average program you can tune beats a clever one you can't.
RatingEcho runs the loyalty club, referrals, customer segments and campaigns off the same diner records the feedback QR already collects, so joining takes a tap rather than a form. Coupons are validated server-side, so a screenshot can't be redeemed twice. Nothing in it rewards a diner for leaving a review, for the policy reasons above — review invites go to everybody, and what they write is entirely their business.