Repeat-purchase program: the one lever that cuts CAC without touching average ticket

1. Why does a repeat-purchase program pay off more than any acquisition campaign?

A repeat-purchase program pays off more because it spreads acquisition cost across several visits instead of amortizing it against a single one.

ChowNow (2025) puts average paid CAC in fast food at 27 dollars, organic at around 9, and fine dining close to 180; none of those numbers drops because you build a database, what changes is their weight per ticket. With a guest who returns three times, those 27 dollars weigh 9 per visit, and that is the line the EBITDA actually feels. Add the 2.05-dollar CPC that PPC Chief (2026) reports for restaurants and food, with a 7.6% CTR, and you will see each paid click costs you twice: once when you buy it, and again when you never speak to whoever walked through that door. Below 500 thousand dollars in annual revenue the decision is not to buy a platform and instead build the list by hand, with a simple threshold: 400 contacts captured before you pay a single dollar of licensing.

2. Under 500 thousand dollars a year: the manual database before the software

This restaurant knows its customers by name and has almost never written it down, which is the awkward advantage of its size. A notebook or a spreadsheet with name, phone, last visit date and favorite dish is enough to start, because Stripo (2025) measures that 55% of diners are influenced by well-crafted promotional emails, and that percentage does not distinguish between a 400-dollar monthly CRM and a disciplined sheet. The enemy here is not missing technology: it is that nobody has been assigned the job of asking for the data at the table. In the 500 thousand to 1 million band the correct move is to automate one single trigger —the reminder 21 days after the last visit— and measure it for a quarter before adding anything else. Lightspeed (2025) measured that people who order online visit 67% more frequently, and that gap comes not from the coupon but from a reminder placed on the right day.

3. Between 500 thousand and 1 million: automate the 21-day reminder

Set the threshold at a 12% reactivation rate over dormant contacts; below that, the problem is the message, not the tool. With Tabular (2025) reporting 10.1% to 14.2% clicks on abandoned-checkout SMS in restaurants, we know the channel responds. An owner in this band does not need more campaigns, they need ONE that runs without being pushed every Monday. Past the million-dollar mark, the decision is to segment by visit frequency rather than ticket value, using three buckets: guests from the last 30 days, those between 31 and 90, and those beyond 90. The reasoning is accounting-driven and it annoys plenty of people who segment by spend. Malou (2025) reports that 79% of restaurant searches are non-brand, so every month you bid on Google for the right to exist in front of guests who already dined with you. Your own database breaks that auction, but only if the message going to the 90-day bucket differs from the one the 30-day bucket receives.

4. Above 1 million: segment by frequency, not by ticket

Operating threshold: fewer than 25% of your base in the hot bucket means your program is retaining the already loyal and reactivating nobody. Above 5 million a profile appears that changes the math: the celebrity-chef restaurant or the large-format themed venue, where personal brand pulls new traffic effortlessly and hides terrible repeat business. There the program justifies itself as risk cover before it works as a growth lever. TouchBistro (2025, via Tablein) measures that 25% of diners would avoid a restaurant over social media criticism, and 41% research on those same networks where to eat; an operator whose flow depends on the social algorithm sits one bad quarter away from an empty room. Threshold: if more than 60% of reservations come from discovery rather than your own database, you do not have a profitable restaurant, you have a phenomenon with an expiration date. Diego F. Parra insists at Masterestaurant on measuring that percentage before food cost.

5. Above 10 million, group or chain: cross-location repeat purchase

A group above 10 million dollars must build CROSS-location repeat purchase, with a single guest identity, and that is the only case where platform investment pays for itself in under twelve months. The arithmetic is direct: if you move 200 thousand guests a year and each visits 1.4 times, lifting that to 1.7 hands you 60 thousand extra visits without opening a location or buying one more click at the 2.05 dollars PPC Chief (2026) marks as sector CPC. Grand View Research projects an 8.6% CAGR for delivery in Latin America between 2025 and 2030, and 7.7% in Europe, so volume grows either way; the question is whose guest that will be once it grows. Governance threshold: one identifier per person across every site, or the program measures smoke. The right reactivation offer gives away a low food-cost dish, never a percentage off the check, because a flat discount eats the margin on the high ticket that was already coming.

6. The offer that does not destroy contribution margin

A dessert at 22% food cost on a 45-dollar ticket costs you around 3 dollars of food; a 15% discount on that same check costs 6.75 and changes nobody's decision. I got this wrong for years, treating repeat purchase as a points-accrual matter when it is really decision architecture: what goes out, to whom, how many days later, and at what real cost. With 37% of Americans dining out less frequently according to Morning Consult and NRN (2025), the margin leaves no room for gifts calculated by eye. If you skip the program this year, the likely scenario is not that sales fall, but that they rise with a growing acquisition cost until your break-even point shifts without warning. Follow the chain: BrightLocal (2026) reports that 88% of local mobile searches end in a visit within 24 hours and that 64% of US diners search Google before going.

7. What happens if you skip the program this year

That traffic exists and you are going to buy it. But every guest acquired and not retained forces you to buy another one next month, at a CPC that PPC Chief (2026) sets at 2.05 dollars with no guarantee it holds. Start tomorrow with the one thing that needs no budget: ask for the email on the check, every table, every shift. The first one is accounting, and it is the hardest to swallow: with no repeat-purchase program, customer acquisition cost amortizes against ONE visit. With a program, it spreads across three or four. The US$27 from ChowNow (2025) does not shrink; what shrinks is its weight per ticket, which is what EBITDA actually sees. The second is channel ownership. While 79% of restaurant searches are non-brand, per Malou (2025), you are re-bidding every month for the right to exist in front of a guest who already knows you.

8. Four differences that rewrite the unit economics

An owned base breaks that auction. Third, frequency. Lightspeed (2025) measured online orderers visiting 67% more often. That lift does not come from the coupon, it comes from the well-timed reminder, and industrializing reminders is precisely what this program does. The fourth difference is risk. With 37% of Americans eating out less often (Morning Consult / Nation's Restaurant News, 2025), an operation living on acquisition alone stands exposed to market contraction. One with an owned base absorbs that operational variability through a channel that never charges per impression.