Repeat-visit program: traditional method vs the Masterestaurant method
A repeat purchase program works when you measure the SECOND VISIT by monthly cohort instead of counting accumulated points: the traditional method hands stamps to everyone at the register and celebrates 4,200 cards issued, while the Masterestaurant method fixes a 30-day window, fires the nudge on day 9, and publishes one number — the share of August guests who came back before September 17. On the same budget, moving repeat visits from 22% to 31% in a venue billing 45,000 USD a month adds close to 4,000 USD in sales with no extra ad spend, because the guest already exists and bringing them back costs roughly a fifth of what first-time acquisition costs.
Here is the arithmetic almost nobody runs: a neighborhood restaurant with 1,800 tickets a month and 22% repeat visits at 30 days actually has about 396 returning guests. Lifting that figure nine points, to 31%, means 162 extra visits a month; at a 24 USD average check that is 3,888 USD in sales requiring no media spend, no aggressive discount, no new menu. That is the territory of the repeat purchase program, and in 2026 it remains the cheapest ground left in the sector.
The traditional method mistakes the instrument for the result. Somebody buys a loyalty app, prints stamp cards, announces a 10% discount for frequent guests, and the dashboard starts reporting sign-ups: 4,200 registrations in the first quarter. Nobody asks how many of those 4,200 walked in twice, and the answer usually hovers around 12%. We keep measuring the top of the sales funnel while the business is decided at the bottom.
I got this wrong for years: I assumed the reward mechanic was the problem, so we tested stamps, points, tiers, cashback. None of those variations moved the needle more than two points. What moved it was changing the unit of measurement — from registered customers to monthly cohort with a closed return window — and placing a nudge on day 9, before the habit cools. Mechanics matter far less than the software vendor claims.
And there is a genuine tension worth settling before you start: repeat visits are bought with margin. Every retention point you buy with a 20% discount gets paid in higher effective food cost; if a dish sits at 30% food cost and you give away a fifth of the check, that dish travels to 37.5% and breaks the 32% ceiling. The answer is not to stop rewarding, it is to reward with low relative cost and high perceived value — the house dessert, the signature drink, the seasonal starter — costing you 1.80 USD while the guest values it at 9. The program then grows without the cash register paying for it.
Side-by-side comparison
| Traditional loyalty program | Masterestaurant repeat-visit program | |
|---|---|---|
| Metric reported | ✕Cumulative sign-ups: 4,200 per quarter | ✓Cohort repeat rate: 31% at 30 days |
| Measurement window | ✕Open-ended, no cutoff (12 months or more) | ✓Closed: 30 days from the first visit |
| Timing of the nudge | ✕At 10 stamps (day 90-140 in practice) | ✓Day 9 after the visit, before it fades |
| Reward cost over check | ✕18-20% as a straight discount | ✓6-8% in high-perceived-value product |
| Impact on dish food cost | ✕Rises from 30% to 37.5% on rewarded checks | ✓Stays under the 32% ceiling |
| Where the data lives | ✕Inside the vendor app, limited export | ✓Owned base with phone, date and check |
| Time to first usable number | ✕90 days (waiting for the first redemption) | ✓37 days (cohort plus closed window) |
| Cost to win a guest back | ✕Never calculated; blended into acquisition | ✓4-6 USD versus 22-28 USD acquisition |
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 1,800 tickets a month and a 22% repeat rate has 396 guests coming back, 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. 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.
Step 2: set the return window and make it the dashboard metric
A dashboard reporting 4,200 cards issued can never get worse, because the figure only climbs; one reporting what share of the July cohort returned before August 15 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. 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.
Step 3: fire the nudge on day 9, not after 60 days of inactivity
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. 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.
Step 4: reward with low-cost, high-perceived-value product, never with a percentage
Pick an item costing you 1.80 USD that the guest values at 9 —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. Nine points of repeat business at that 1,800-ticket venue equal 162 extra visits a month, and with a 24 USD average check that comes to 3,888 USD of sales requiring 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 162 visits through advertising would run close to 4,900 USD before anyone sits down.
Step 5: do the math on what nine points of repeat business are worth
And if those visits arrived through third-party delivery, the effective cost reaches 30%-40% of the order according to Restaurant Business (2024). 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. The costliest mistake is counting sign-ups instead of second visits: out of 4,200 first-quarter registrations, the share that walks in twice usually hovers around 12%, 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.
Step 6: the four mistakes that sink the program in its first quarter
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. 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 3,888 USD you were chasing, because it tells you where the real money should go.
Step 7: what happens if day 9 doesn't move the needle
The deliverable is a short memo with the decision: fix operations first, or scale the program to the remaining cohorts. 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 August 15 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. The economics of the reward. Giving away 20% of the check costs 20% of the check.
Four differences that decide the outcome — in practice
Giving away a dessert that costs you 1.80 USD and the guest values at 9 costs 4% of a 45 USD 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.
Criterion-by-criterion analysis
What the average restaurant doesTraditional method
- Hires a loyalty app and outsources judgment with it: the vendor decides the mechanic, not the business.
- Rewards with a percentage discount on the total, the most expensive way to buy a visit.
- Tracks sign-ups and cumulative redemptions, two numbers that only rise and never reveal whether guests returned.
- Sends the reminder once the system flags 60 or 90 days of inactivity, when the habit is already gone.
- Treats the Tuesday lunch guest exactly like the Saturday dinner guest, though their reasons to return differ.
- Never crosses repeat data with online reputation: a three-star review runs its course unanswered.
What the Masterestaurant method doesMasterestaurant
- Defines the monthly cohort and closes the window at 30 days: one number per month, comparable to the last.
- Places the nudge on day 9, while the memory of the dish still outranks the calendar.
- Rewards with low-food-cost, high-perceived-value product, so margin survives the growth.
- Segments by the slot of the first visit: weekday lunch, weeknight dinner, weekend.
- Keeps the base in-house — phone, date, check, channel — and reuses it for delivery conversion.
- Closes the loop with reputation: returning guests are asked for a review, absent ones are asked why.
Side-by-side comparison
| Traditional loyalty program | Masterestaurant repeat-visit program | |
|---|---|---|
| Metric reported | ✕Cumulative sign-ups: 4,200 per quarter | ✓Cohort repeat rate: 31% at 30 days |
| Measurement window | ✕Open-ended, no cutoff (12 months or more) | ✓Closed: 30 days from the first visit |
| Timing of the nudge | ✕At 10 stamps (day 90-140 in practice) | ✓Day 9 after the visit, before it fades |
| Reward cost over check | ✕18-20% as a straight discount | ✓6-8% in high-perceived-value product |
| Impact on dish food cost | ✕Rises from 30% to 37.5% on rewarded checks | ✓Stays under the 32% ceiling |
| Where the data lives | ✕Inside the vendor app, limited export | ✓Owned base with phone, date and check |
| Time to first usable number | ✕90 days (waiting for the first redemption) | ✓37 days (cohort plus closed window) |
| Cost to win a guest back | ✕Never calculated; blended into acquisition | ✓4-6 USD versus 22-28 USD acquisition |
The numbers behind the decision
“We had 4,200 cards issued and I reported it as a win in every board meeting. Once we closed the 30-day window on the March cohort, only 19% had returned. We swapped the 20% discount for a house dessert costing us 1.80 USD, moved the message from day 60 to day 9, and by the next quarter repeat visits reached 30%: 214 extra visits a month, 5,136 USD in additional sales, with food cost holding at 29.4%.”
Building the program in four steps, each with a measurable deliverable
PREREQUISITE: export a full month of tickets from the POS with date, time, amount and a guest phone or identifier; without an identifier there is no repeat purchase program, only coupon distribution. Deliverable: one sheet holding last month's cohort and its 30-day return rate, expressed as a single number. Numeric checkpoint: if your base identifies fewer than 40% of tickets, stop here and fix capture at the register first. Common mistake: counting any visit within the year as a repeat; the window must be exactly 30 days so August compares with September. Second mistake, the costlier one: blending dine-in and delivery into one cohort, because delivery conversion behaves differently and will muddy your reading for months.
Pick two or three items with real food cost under 12% and high perceived value: the house dessert, a signature drink, the seasonal starter. Deliverable: a spec sheet per reward with unit cost signed off by the kitchen and its menu price. Numeric checkpoint: reward cost must stay under 8% of the average check, and the rewarded dish has to land below 32% effective food cost. Common mistake: the percentage discount on the total, which eats 18-20 points of the check and teaches guests never to return without a coupon. Discipline beats creativity here: if the kitchen does not sign the unit cost, the reward becomes a silent leak that surfaces three months later in inventory.
Write three distinct messages by the slot of the first visit — weekday lunch, weeknight dinner, weekend — and schedule delivery at 11:00 on the ninth day. Deliverable: three short templates carrying the guest's name, the dish they ordered, and the reward called by its real name rather than its category. Numeric checkpoint: open rate above 45% and redemption above 11% on the first batch; below either number, the message or the reward is wrong. Common mistake: sending identical copy to all three slots, when the Tuesday lunch guest wants speed and the Saturday guest wants a reason to celebrate. Second mistake: writing in brand voice; people answer what sounds like a person.
Guests who return get asked for a review at the second check, naming the server who looked after them; guests who ignored the day-9 nudge get one open question at day 21. Deliverable: a one-page board showing cohort repeat rate, cost per recovered guest, and volume of new reviews. Numeric checkpoint: repeat rate above 28% by month three, recovery cost under 6 USD, and at least 12 new reviews a month. Common mistake: dropping the measurement once the number climbs, which is precisely when it starts sliding unnoticed. The review is monthly, it lands on the first Monday, and it takes twenty minutes.
And with AI?
Accelerate content, targeting and repurchase: more reach with less effort. Diego F. Parra is an expert in AI applied to restaurants.
Free tools to apply this now
Method tools that keep the program alive
A repeat purchase program lives on three things: a well-captured owned base, a costed reward, and a monthly review nobody skips. The Masterestaurant ecosystem tools cover those three pieces and save the month usually lost building templates from scratch.
None of them replaces the discipline of closing the window every 30 days, but they do prevent the most common failure: running the program on an improvised sheet that nobody maintains past month three.
Diego F. Parra — International consultant, expert in creating and scaling restaurants and in AI applied to restaurants, foodtech and HORECA. Methodology applied in 8.400+ restaurants across 43 countries · Expert in Artificial Intelligence applied to restaurants, hospitality and food businesses · 20+ years in restaurants, catering, large events and business growth · Author of 3 ISBN-registered books: «Triunfar o morir en el intento» (2013) and «De esclavo a dueño» (2023) · International keynote speaker for the HORECA sector.
Frequently asked questions about the repeat-visit program
How long before a repeat purchase program produces its first usable number?
How long before a repeat purchase program produces its first usable number?
Thirty-seven days: the 30-day closed window plus a week of processing. Any figure before that cutoff is noise. The traditional program takes around 90 days because it waits for the first stamp redemption, and by then two whole cohorts of learning are gone.
What 30-day repeat rate counts as good in a casual restaurant?
What 30-day repeat rate counts as good in a casual restaurant?
Between 28% and 34% is healthy for neighborhood casual dining; below 22% there is a product or service problem no program can fix. Always compare against your own previous month rather than sector averages, since your channel mix and time slots govern the result.
Should the reward be a percentage discount or a product?
Should the reward be a percentage discount or a product?
Product, almost always. A 20% discount costs exactly 20% of the check and trains guests to wait for coupons; a dessert costing 1.80 USD with a 9 USD menu price costs 4% of a 45 USD check and earns more goodwill. The exception is very low-ticket weekday lunch.
Does a repeat purchase program work for delivery too?
Does a repeat purchase program work for delivery too?
Yes, with a separate cohort. Delivery conversion behaves differently: the guest buys on impulse and compares prices inside the platform, so the day-9 trigger performs better as a free item above a minimum order. Blending dine-in and delivery into one number hides which channel is slipping.
Sector data 2026 (official sources)
Verifiable industry benchmarks from official, non-commercial sources (government, industry associations, market research) - not competitors.
| Metric | Benchmark 2026 | Source |
|---|---|---|
| CAGR del delivery online en América Latina (2025-2030) | 8,6% | Grand View Research — Latin America Online Food Delivery Market |
| Participación de iFood en el delivery de Brasil | 80% | Grand View Research — Latin America Online Food Delivery Market |
| Restaurantes en el mundo que usan códigos QR para menús digitales | 75% | QR Code — QR Code Statistics for Restaurant Usage 2025 |
| Aumento del volumen de escaneos de QR en dos años | 433% | QR Code — QR Code Statistics for Restaurant Usage 2025 |
| Consumidores que prefieren menús QR sobre menús de papel | 78% | Eater (vía QR Code) — QR Code Statistics 2025 |
| Aumento de rotación de mesas con pagos por QR | 15% | QR Code — QR Code Statistics for Restaurant Usage 2025 |
Related content
Grow your restaurant with the Masterestaurant method
Applied in +8.400 restaurants across 43 countries.