Repeat-visit program: traditional method vs the Masterestaurant method

Step 1: measure your real 30-day repeat rate before you buy anything

The first number you need on the table is what share of one month's guests came back before day 30, and almost no neighborhood restaurant knows it. Take July's tickets, isolate the repeated identifiers —reservation phone, tokenized card, online-order email— and divide repeat visits by the unique guests in that cohort. A venue with a steady monthly ticket count and a modest repeat rate has only a fraction of its guests coming back, and not one more. The deliverable here is a single sheet with three columns: cohort month, unique guests, returns before day 30. You verify it by adding up three consecutive cohorts: if all three land on the same round percentage, your database is duplicating records and you must clean it before moving on.

Step 2: set the return window and make it the dashboard metric

The 30-day window replaces accumulated sign-ups as the headline metric, and that change of unit is worth more than any loyalty software on the market. A dashboard reporting cards issued can never get worse, because the figure only climbs; one reporting what share of the July cohort returned before the following month can indeed fall, which is exactly why it helps you decide. Diego F. Parra puts it this way to every owner who arrives at Masterestaurant with a points panel: if your headline indicator is incapable of delivering bad news, it isn't an indicator, it's decoration. The deliverable is done when the manager's dashboard shows one large number —30-day repeat rate for the live cohort— and sign-ups drop to a secondary column nobody argues about in the weekly meeting.

Step 3: fire the nudge on day 9, not after 60 days of inactivity

Day 9 after the visit is your contact moment, and that choice isn't cosmetic: the casual-dining return curve collapses between day 7 and day 14, while the memory of the dish still competes with the week's routine. Traditional programs wait for 60 days of inactivity before reacting, meaning they write once the guest has already rearranged their outings and adopted another place. Schedule the automated send for 11:30 on day 9, ahead of the lunch decision, with a two-line message and a single offer. The deliverable is a live automation with a verifiable timestamp; you check it by taking five guests from the cohort and confirming in the log that the message reached them between day 8 and day 10, with no exceptions and no manual rescue sends.

Step 4: reward with low-cost, high-perceived-value product, never with a percentage

This is where the whole program's margin is decided, and the rule is blunt: no percentage discounts on the check. If a dish runs at 30% food cost and you give away 20% of the bill, that dish travels to an effective 37.5% and breaks the 32% ceiling set by the Masterestaurant costing rule. Dropping the reward isn't the answer. Pick an item that costs you little and that the guest values far more —the house dessert, the signature drink, the seasonal starter— and hand it over as a named courtesy, not a coupon. The deliverable for this step is the spec sheet for the reward item with its unit cost signed off by the kitchen, plus a monthly unit cap. You verify it by cross-checking courtesies served against ingredient consumption for the period.

Step 5: do the math on what nine points of repeat business are worth

A few extra points of repeat business at a busy venue add up to many extra visits a month, and each one is sales that requires no ad spend, no new menu, no extra Instagram photo. Compare that with the other road: cost per lead on Google Ads for restaurants and food sits at US$30.27 according to WordStream (Google Ads Benchmarks 2025), so buying those extra visits through advertising would cost several times what it takes to bring them back, and that is before anyone sits down. And if those visits arrived through third-party delivery, the effective cost would eat a large share of the order value. The deliverable is a twelve-month projection with three repeat-rate scenarios —current, +5 points, +9 points— signed by the owner. You verify it against the following quarter's actual close.

Step 6: the four mistakes that sink the program in its first quarter

The costliest mistake is counting sign-ups instead of second visits: of all the registrations in a quarter, only a modest share ever walks in twice, and nobody asks for that figure because the app dashboard doesn't show it. Second comes piling up channels without a shared identifier, so the same guest shows up as a reservation, as an online order and as a card, inflating the cohort. Third is switching mechanics every two months —stamps, points, tiers, cashback— hunting in the reward for an answer that lives in the calendar. I got this wrong for years, testing variations that never moved the needle by more than two points. Fourth is leaving execution to whoever happens to be on shift: with no assigned owner, the day-9 message gets skipped during the first busy week and the program dies unsigned.

Step 7: what happens if day 9 doesn't move the needle

Suppose you run three full cohorts with the 30-day window, the day-9 nudge and the cost-controlled reward, and the repeat rate stays nailed at 22%. Before blaming the program, examine the product: if 41% of diners research a restaurant on social media before going, according to the TouchBistro 2025 Diner Trends Report, and someone who already tasted your food still doesn't come back, the problem isn't marketing, it's the plate, the wait time or the treatment at the table. A repeat program amplifies what already exists; over a mediocre experience it merely speeds up the diagnosis. That uncomfortable finding is worth more than the sign-up target you were chasing, because it tells you where the real money should go. The deliverable is a short memo with the decision: fix operations first, or scale the program to the remaining cohorts.

Closing checklist: how to know everything landed right

Your program is built when you can answer five questions without opening any software: how many unique guests last month's cohort had, what share returned before day 30, how many messages went out on day 9 and how many bounced, what the total courtesies cost in ingredients, and what the sales differential was against the previous cohort. If any of the five requires a call to your vendor, the program still isn't yours. Check three details that tend to fail as well: that the database doesn't duplicate the same guest across two channels, that the reward item carries a monthly cap signed by the kitchen, and that the repeat-rate number appears in the weekly meeting next to food cost. Set the review for the first Friday of every month and put one named person in charge of defending that figure.

Four differences that decide the outcome

The unit of measurement. A program reporting cumulative registrations can never fail, because the number only climbs; one reporting the share of the July cohort that returned before the following month can drop, which is exactly why it is useful. Diego F. Parra presses this point with every owner who arrives at Masterestaurant carrying a loyalty dashboard: if your headline metric cannot get worse, it is not a metric, it is decoration. The calendar of the nudge. The return curve of a casual restaurant falls sharply between day 7 and day 14 after the visit, so firing the reminder on day 9 catches the stretch where sensory memory still competes with routine. Traditional programs wait for 60 days of inactivity, meaning they act once the guest has already rebuilt their nights out around a different venue.

Four differences that decide the outcome — in practice

The economics of the reward. Giving away a slice of the check costs exactly that slice of the check. Giving away a dessert that costs you little and that the guest values far above its cost takes a small bite out of the check and produces more gratitude. The arithmetic is uncomfortable in its simplicity, and yet most loyalty schemes in the sector remain anchored to percentage discounts. Ownership of the data. When the base lives inside a vendor app, you are renting your own guests: you cannot cross repeat rates with delivery channel, segment by time slot, or recover phone numbers the day you switch platforms. An owned base — one sheet with phone, date, check and channel — is worth more than any gamification module.