Ask a restaurant owner about their menu and you will get twenty minutes of detail — which dish carries the margin, which one only sells on weekends, what the new supplier did to the hummus. Ask the same owner who their best guest is and you usually get a shrug, or a first name and a table preference. The till knows everything about the food and almost nothing about the people. That gap has a price: the couple who came three Fridays in a row and silently stopped, the birthday group that never heard from you again, the Tuesday dining room that sits half empty while your regulars eat somewhere forgettable.
The fix is not a loyalty app — diners do not download restaurant apps, and the punch card dies in a wallet. The fix is a CRM built on the one channel every guest already opens daily: WhatsApp. Here is the whole setup, honestly sized at about thirty minutes, and the small operating habit that follows it.
Why restaurants never built a CRM — and why WhatsApp changes the math
Classic CRM assumed a salesperson typing notes after a call. Restaurants have ninety-second transactions, a queue at the counter, and zero chance of getting a guest to fill in a form. So the data never got captured, and without data there were no segments, no campaigns, no measurement — just gut feel and a discount aggregator taking a cut to rent you your own customers back.
WhatsApp quietly removes the capture problem. Every guest already has it; nobody installs anything; and the phone number is a natural, durable identity. The moment a guest messages you once, you have a record key that survives staff turnover, POS migrations, and the guest changing email addresses. Everything below builds on that one idea: one phone number, one guest record.
Minutes 0–10: put a QR where guests are already looking
A CRM is only as good as its front door. In a restaurant, the front door is a QR code in the three places a guest's eyes naturally land:
- The table. A small tent or a sticker on the menu corner. Give the scan a reason — early access to specials, a secret-menu item, a birthday treat — not a vague follow us.
- The receipt. This is the highest-intent moment of the visit: the guest has just paid and is holding the proof. A QR that delivers a paperless receipt over WhatsApp does two jobs at once — it captures the guest as a contact and ties tonight's spend to their record. It also means one less thermal slip in the bin, which is exactly the every-scan-saves-a-tree habit we built Remarketly around.
- The counter and the delivery bag. Takeaway and delivery guests are the easiest to lose, because they never see your room. A card in the bag is your only handshake with them.
Make the QR open a WhatsApp chat with a pre-filled first message, so one tap and one send is the entire sign-up. You can generate these in a couple of minutes with our free QR code tool. One non-negotiable: the first reply should ask plainly whether the guest wants offers and news, and record the answer. Opt-in is not paperwork — it is the difference between a welcome message and spam, and it protects the number you will come to depend on.
Minutes 10–20: one guest record per phone number
The moment that first message lands, the CRM should do the boring work automatically: create a contact keyed to the phone number, then let every later interaction accumulate on it. Over a few weeks a record stops being a phone number and becomes a guest:
- Visits — each scanned receipt or table QR stamps a visit with a date and time.
- Orders and spend — what they ordered and what the ticket was worth, pulled from the receipt rather than typed by anyone.
- Notes and tags — window table, no nuts, anniversary in March, always asks for the karak. The things a great maître d' keeps in their head, except they no longer leave when the maître d' does.
None of this requires staff to do data entry during service — that is the test of whether a restaurant CRM will survive contact with a Friday rush. If a step needs a waiter to type, delete the step. The Remarketly CRM treats the phone number as the spine and hangs conversations, receipts, visits, and tags off it without anyone touching a keyboard.
Minutes 20–25: the four segments that matter in F&B
Skip the forty-segment marketing fantasy. Four lists run a restaurant, and they should maintain themselves from visit and spend data — not from a spreadsheet someone forgets to update:
- First-timers — one visit, within the last two weeks. The most fragile relationship you have. The entire game with this list is earning visit number two.
- Regulars — roughly three or more visits in sixty days. Protect them, recognise them, and be careful about training them to wait for discounts they would have come without.
- 30-day lapsed — guests who had a rhythm and then went quiet for a month. They have not rejected you; life drifted. They are winnable while the memory of your food is still warm.
- Big spenders — the top slice of your guests by total spend. These people respond to invitations and first access, not coupons. A ten-dirham voucher to this list is an insult dressed as a gift.
Notice what these four have in common: each one implies a different message. That is the whole point of segmenting — not cleverness, just never sending the win-back text to someone who ate with you on Saturday.
Minutes 25–30: turn on the three campaigns
Three automations cover most of what a restaurant needs from remarketing. Set them once; they run on the segments above.
1. The welcome offer
Fires the day after a first visit: a thank-you and a small, time-boxed reason to come back within two weeks — a complimentary dessert, a starter on the house. The incentive is not the message; the message is we noticed you, and we want a second date. Guests who make a second visit behave very differently from guests who made one, which is why this is the campaign to build first.
2. The win-back
Fires when a guest crosses the 30-day lapsed line. One message, specific and human — mention the dish they ordered if you have it on record — with a modest reason to return. One message, not a drip of escalating nags. If they do not come back after two attempts spaced weeks apart, let the silence be the answer.
3. The slow-night fill
The empty-Tuesday campaign. Pick your quietest service, then send a small, targeted list — regulars and big spenders, never the whole database — a same-week invitation with something exclusive to that night: a tasting of the new menu, a chef's table, a dish that only exists on Tuesdays. Kept scarce and personal, it reads as an invitation. Sent to everyone weekly, it reads as desperation and burns the list.
A note on cost, because owners always ask: campaign messages to opted-in guests are paid, per-message, under Meta's marketing category, while your replies inside a live conversation are free. The amounts are small against the value of a filled table, but they are not zero — the indicative per-message ranges for the UAE and KSA are on our WhatsApp API pricing guide, and you should verify the live figures on Meta's rate card before budgeting. The practical rule: small lists, good reasons, no blasting.
Reservation reminders that end no-shows
No-shows are not a guest morality problem; they are a reminder-design problem. People forget, plans wobble, and a booking made on Monday is an abstraction by Saturday. The fix is a boring, reliable three-touch flow:
- At booking: instant confirmation with the date, time, party size, and a one-tap way to change it.
- The day before: a reminder with two buttons — confirm or cancel. Make cancelling easy on purpose: a cancelled table at noon is inventory you can resell; a silent no-show at nine is a dead loss.
- A couple of hours out: a final nudge with directions or parking. By now the guest has confirmed twice; the table is as safe as a table gets.
Two details make this work on WhatsApp where phone calls and email reminders quietly fail. First, replies are one tap, so guests actually answer. Second, these are transactional, appointment-style messages — the cheaper utility category, not marketing — so the flow costs little to run even at volume. Every confirmed cancellation should release the slot back to your booking flow automatically, which is how the same system that ends no-shows also quietly fills the gaps they used to leave.
Measure one number: repeat-visit rate
Restaurants drown in metrics that do not change decisions. Track one: of the guests whose first visit was last month, what share came back within sixty days? That is your repeat-visit rate, and before a CRM you simply could not see it — walk-ins are anonymous, and the till counts tickets, not people. Once every scan ties a visit to a phone number, the number falls out of the data on its own.
- Set the baseline in your first full month. Do not judge it; just write it down.
- Watch the welcome-offer cohort — guests who got the message versus the stragglers who somehow did not. That gap is the campaign's real report card.
- Check it monthly, not daily. Visits are slow-moving; daily checking just teaches you to distrust noise.
The strategic reason this number matters: a restaurant grows by getting more strangers or more returns. Strangers are expensive — ads, aggregator commissions, discount platforms. Returns cost a message. Every point of repeat-visit rate you gain is revenue you no longer have to buy.
The whole setup, on one screen
- QR codes on tables, receipts, and delivery bags — with a reason to scan and an explicit opt-in.
- One record per phone number, fed automatically by receipts and chats.
- Four self-updating segments: first-timers, regulars, 30-day lapsed, big spenders.
- Three campaigns: welcome offer, win-back, slow-night fill.
- A three-touch reservation reminder flow with one-tap confirm and easy cancel.
- One metric on the wall: repeat-visit rate, checked monthly.
Thirty minutes of setup, then it runs while you run the kitchen. If you want to see how this fits together with the inbox, the AI agent, and loyalty for F&B specifically, the restaurants page walks through the full picture — and when you are ready to put a name to every table, join the waitlist and we will help you print your first QR before the dinner rush.