A restaurant and café loyalty programme that actually brings guests back
Author:
Paweł Matusiak
·
Paper stamp cards get lost. An aggregator app awards points to them, not you. Your own loyalty programme — points, stamps, birthday coupons — keeps the guest in your venue and gives a reason to order again.
The most expensive guest is the new one from ads. The cheapest is the one who comes back on Thursday because they are two stamps short of a free coffee. A restaurant loyalty programme works when it lives on the phone, counts itself and does not need a waiter with a pen. Otherwise it is a three-week gimmick. Harvard Business Review and Bain have repeated the same line across industries for years: winning a new customer can cost 5–25 times more than keeping one, and a 5% lift in retention can raise profit by 25–95%. In hospitality that is not a slide. It is the difference between a full Thursday and a thin diary you fill with ads.
This is for a café, pizzeria or restaurant owner who “has regulars” but cannot write to them. Below: why paper and aggregator points do not build your list, which models work at the till, a day in a venue with a phone card, simple maths, SaaS versus your own programme, a staged rollout, mistakes and GDPR.
Why paper and someone else’s app fail
A paper stamp is lost, mixed with other cards, or the guest will “remember next time”. At peak the waiter has no time to stamp it, so the programme dies on the first Friday. Worse: points in Pyszne, Glovo or Uber Eats build loyalty to the platform. The guest returns to an app that will also show a competitor at −30%. You paid commission on the first order and you do not have a phone number for the second.
- The card in the wallet is lost, or the guest will “remember next time” — and does not.
- Points in an aggregator build loyalty to the platform, not the venue.
- You do not know who has not returned in a month — so you cannot send a coupon.
- Staff have no time to stamp cards at peak, so the programme dies.
- You cannot join a stamp to table booking or to online orders.
- Fake stamps and staff “topping up” cards with no control.
- No contact consent: even if you have a number on the receipt, you may not text “come for brunch”.
What works in hospitality
Simple rules you can explain at the till in ten seconds. If a waiter needs a flyer and three exceptions, the guest will not get it and the shift will not run the programme after 6 pm. Models we typically ship — alone or combined:
- Digital stamps: 9 coffees = 10th free. QR on the receipt or at the till. King of cafés and bakeries.
- Percentage points on the bill, redeemed for a menu item or a discount. Better in restaurants with a higher average check.
- Birthday visit: an automatic coupon in a 7-day window — low cost, high sentiment.
- Tiers (regular / VIP) for guests who spend a set amount in a month: earlier event access, a dessert, the window table.
- A “come back within 14 days” coupon for people who went quiet — cheaper than a new ad.
- Office lunch: stamps on pickup and delivery from the same card as the floor.
- Referral: both sides get a coupon when a new guest arrives with a code — legally, with measurement.
A list, not only a discount
The programme is the reason to legally collect a phone number and SMS consent. Then you have a cheaper channel than ads: “weekend breakfast”, “soup of the day”, “a table free at 5 pm”. The same list connects to booking and online orders. A discount without a list is a promotion. A list without a discount will not collect consent. We design both together.
A day in the venue with a card on the phone
8.40 — first coffee. The guest shows a QR from the screen or gives a phone number. The waiter scans, the stamp lands in two seconds. Nobody hunts a cardboard card in a bag. 12.15 — lunch. Points accrue on the bill; at the till you see 40 points short of a dessert. The waiter says that in one sentence, with no upsell training.
16.00 — the system sends a “come back in 14 days” coupon to people who have not visited in a month. A share returns for tomorrow’s brunch. 18.30 — a regular books a table from the same card. The hostess sees “VIP, gluten allergy, usually the banquette”. 21.00 — a report: how many stamps, how many coupons redeemed, which weekday is thin. The manager does not guess whether “loyalty works”. They see it.
- Morning: fast stamps at the till, no queue because of the programme.
- Lunch: points on the bill, a light upsell, the same card on pickup.
- Afternoon: automations (birthday, 14-day win-back), not a manual spreadsheet.
- Evening: the hostess sees guest status at table booking.
- Close: coupon redemptions, reward cost vs extra turnover.
Numbers and payback: what a returning guest is worth
Take a café: an 18 PLN average coffee, a regular twice a week, 48 weeks. That is about 1,700 PLN a year from one person before you add a sandwich. The cost of the tenth coffee free is 18 PLN of product and a moment. The cost of a new guest from an Instagram ad can be tens of zloty for a visit that may not repeat. HBR and Bain are not saying “loyalty is nice”. They are saying retention is cheaper than acquisition — in a 5–25× band, depending on the industry and the study.
In a restaurant with a 150 PLN bill, one extra lunch a week from a regular is 600 PLN a month. Twenty such people is 12,000 PLN of turnover you do not have to buy from an aggregator or from ads. The reward (a dessert, −10%, a coffee) is a fraction. The condition: the programme must count itself and must be able to write to the person who vanished. A logo on cardboard will not do that.
- Reward cost (coffee, dessert, −10%) vs the value of the next visits in the month.
- SMS and running cost vs the cost of an ad click for the same Thursday.
- Share of turnover from cardholders — if it falls, the programme is dead, not “in the background”.
- Win-back: how many 14-day coupons return, and how many burn unused.
- Aggregator commission you do not pay when a regular clicks your own online orders.
Off-the-shelf loyalty SaaS or your own programme?
A packaged stamp app can be cheap and enough for a one-site café. It starts to get in the way when you have several addresses, points on the bill rather than on coffee, a POS / booking / delivery link, your own VIP tiers, or when the SaaS holds the list and on the way out you hear that export “is on the roadmap”. Your own restaurant loyalty programme is then the same guest record as the calendar and the cart — not a third login for the waiter.
- A group of venues, one card, a per-site report for the owner.
- Rules a template does not have: lunch vs dinner, bar separate, events, a staff card.
- Automatic points from online orders and from a QR menu.
- Data and consents with you, under GDPR.
- No per-card or per-SMS fee that grows with success.
An app or a card in the browser?
At the start: a site / PWA the guest saves to the home screen. An Android app once you have hundreds of active cards and want push. We do not force an app download to get the first stamp — that kills conversion. A link from a QR on the receipt, an SMS with the card, an “add to home screen” button. A store app is a reward for a list that already exists, not a ticket in.
A staged rollout
We start with one reward and one way of earning. Shift training is fifteen minutes: scan a code or type a phone number. For a week the manager stands at the till and catches exceptions (two coffees on one receipt, a company card, a guest with no phone). Then birthday and win-back automations. Then tiers. If you launch seven rules and VIP on day one, staff get lost and the guest does not understand what the card is for.
- Phase 1: phone card, one reward, scan at the till, GDPR consent.
- Phase 2: automatic credit from POS / online orders, a redemption report.
- Phase 3: birthdays, a 14-day coupon, SMS on thin hours.
- Phase 4: tiers, several venues, a bridge to booking and QR.
- Phase 5: a PWA or a push app once the list is actually alive.
Mistakes that kill a programme in a month
Rules that are too hard. A reward nobody can reach. A mandatory store app. Staff who do not scan because they “have no time”. Daily SMS until the guest withdraws consent. A discount on everything that eats margin and trains people to wait for a promo. Coupons that are not voided — the same code comes back three times. And the classic: a spreadsheet of numbers “for loyalty” with no consent, which is both useless and risky.
- Starting from a Google Play app instead of a QR on the receipt.
- A stamp only on the floor, zero points for delivery from your own site.
- VIP with no criteria — either everyone is VIP, or nobody takes it seriously.
- No expiry on points and a December avalanche of old rewards.
- A different programme in every site of a group, and a guest who does not get why the coffee “does not count”.
How to tell the programme is alive, not just “on the offer list”
A regulars’ card with no numbers is decoration. Once a week the manager should see: how many active cards, what share of turnover comes from cardholders, how many rewards were redeemed, how many 14-day coupons came back, how many SMS consents were withdrawn. If the card share falls, the problem is at the till (they are not scanning) or in the promise (the reward is too far away). If consents fall, you write too often or with no specifics. A restaurant loyalty programme should be a cheaper channel than ads — HBR and Bain give a 5–25× band in favour of retention, but only if you can actually return to the guest, not merely give them a tenth coffee.
Till checklist: the first 30 days of the programme
Days 1–3 — one reward, one scan, the manager at the till catching exceptions (two coffees, a company card, a guest with no phone). Days 4–10 — every shift scans without thinking; if not, the interface is wrong, not “people resist”. Days 11–20 — a birthday coupon and a 14-day win-back on quiet guests. Days 21–30 — the first report: card share of turnover, reward cost, SMS consents. Without that report a restaurant loyalty programme is a poster, not a channel.
- Before go-live: the reward in one sentence next to the QR, marketing consent separate.
- Week 1: scan under 2 seconds, a plan B (phone number), no store app.
- Week 2: points also from online orders, not only the floor.
- Week 3: automations, not a spreadsheet of “regulars”.
- Week 4: dessert cost vs extra visits — if the reward eats margin, you change the threshold, you do not kill the programme.
- Month 2: tiers only once cashiers scan at peak with no queue.
24 months: stamp SaaS versus your own card
A packaged stamp app at 150–300 PLN a month is 3,600–7,200 PLN over two years, often with an SMS cap and no list export. When you reach a few hundred cards you want POS, booking and delivery joined — and you hear that “is another package”. Your own programme has an implementation cost and then cheap SMS. After 18–24 months in a café with repeat traffic, your own card is usually cheaper, and the phone number stays with you when you cancel the subscription.
Also count what SaaS does not show: a new guest from ads vs a regular with a coupon. At 18 PLN a coffee and two visits a week a regular leaves ~1,700 PLN a year. Twenty such people is 34,000 PLN of turnover you do not buy from ads or Glovo. HBR and Bain: retention can be 5–25× cheaper than acquisition. The programme should realise that gap, not decorate the receipt.
- SaaS: fast, cheap to start, the list often at the vendor, export “on the roadmap”.
- Your own card: higher start, one list with the calendar and the cart, zero percent per visit.
- 24 months: add subscription + SMS + no till integration.
- One café, stamps only — a package can be enough. A group and POS — usually not.
The reward, margin, and when a discount eats profit
A tenth coffee free at 3–4 PLN of product is cheap. −20% off the whole Friday bill is expensive if the room was already full. A good reward sits just beyond a threshold the guest almost hits, and does not train people to wait for a promo. VIP with no spend criterion is a discount for everyone. Point expiry protects December from an avalanche of old coffees. That is not stinginess. It is so the loyalty programme brings the guest back instead of only cutting the average check.
If you have regulars and no way to keep them — describe the venue and whether loyalty should sit on coffee, lunch or delivery. We will propose a model staff can run on a Friday night, not only on paper. If a digital stamp without VIP is enough to start, we will say so.
Frequently asked questions
- Does loyalty have to be a native app?
- No. We usually start with a phone card (link + QR). We add an app when the list and notifications need it. Forcing a download for the first stamp cuts take-up.
- How do staff add points at peak?
- The waiter scans the guest code or types a phone number. Points can also land automatically on an online order and on a QR table order. If a scan takes more than two seconds, the interface is wrong.
- Can several venues share one card?
- Yes — one card across a café or restaurant group, with per-venue reports for the owner. Rules can be shared or separate (for example a dessert only at the issuing site).
- How does it pay back?
- From returning guests and cheaper SMS versus paid ads. One extra lunch a week from a regular usually beats the cost of running the programme. HBR and Bain: new acquisition can be 5–25× more expensive than retention.
- Must the guest leave an email?
- No. A phone number is enough for the card and SMS. Email is optional. Marketing needs a separate consent, whichever channel.
- What about GDPR when we collect a birthday?
- Purpose and legal basis must be clear. A birthday for a coupon is not consent for everything. The guest must be able to withdraw marketing and delete the card. Detail in GDPR in applications.
- Should points also count from an aggregator?
- They can, if you can identify the guest. More often we reward your own channel, to give a reason to leave the commission. That is a business call, not a technical one.
- How long does a rollout take?
- A simple digital stamp is weeks. Points on the bill, POS and several sites — the next phase, once cashiers scan without thinking.
Related service:
Custom software for companies
Describe your project