Guest loyalty in restaurants: the numbers before and after
Guest loyalty pays off when the cost of keeping a guest stays under 20 % of the contribution margin that guest leaves in a year. That is the cutoff. A restaurant spending 12 USD to bring in one new visit and 3 USD to bring a regular back is looking at the same till through two different economies, and the second one covers January payroll. Before: a room full of strangers, 1.4 visits per year, permanent discounting to hold volume. After: 34 % of sales from identified guests, frequency at 3.8, and the discount turned into a targeted benefit that costs half as much. The usual mistake is measuring loyalty in signups instead of margin recovered per guest.
A 180-seat steakhouse in Bogotá closed 2025 with 41,000 checks and 29,400 distinct guests. Frequency: 1.39. It paid almost full acquisition every single time somebody sat down. With food cost at 31 % and payroll at 27 %, contribution margin per check sat near 9.80 USD, and out of that came 12.40 USD of acquisition cost per new guest across delivery, aggregators and paid media. It lost money on every table it celebrated as a win.
Guest loyalty is not a points app, it is a decision about financial structure: you stop buying traffic on the open market and start buying it inside your own base, where cost per visit drops between 60 % and 80 %. Diego F. Parra has spent twenty years walking in through the kitchen and out through the till, and at Masterestaurant a loyalty diagnosis always starts in the same place, the P&L, never the CRM.
What follows are the benchmarks that actually drive a decision. Two tables, each figure with its application context, then how to read them in a single unit, in three, and in a group of twelve.
Side-by-side comparison
| BEFORE · no loyalty system | AFTER · loyalty measured in margin | |
|---|---|---|
| Customer acquisition cost per guest | ✕12.40 USD average across paid channels and aggregators | ✓3.10 USD per repeat visit activated from the owned base |
| Annual visit frequency | ✕1.39 visits per identified guest | ✓3.80 visits per identified guest |
| Guest LTV over 24 months | ✕27.20 USD of accumulated contribution margin | ✓74.50 USD of accumulated contribution margin |
| Average check, returning guest | ✕19.60 USD, identical to the new guest | ✓23.90 USD, 22 % above the new guest |
| Effective discount on sales | ✕11.4 % in open promotions available to everyone | ✓4.6 % in benefits segmented by frequency |
| Food cost of the promotional hook dish | ✕38 % on the open 2-for-1, past the 32 % ceiling | ✓29 % on the benefit aimed at the third visit |
| Share of sales from identified guests | ✕7 % of total sales | ✓34 % of total sales |
| Annual cost of the loyalty system | ✕0 USD, yet 68,000 USD a year in acquisition media | ✓9,600 USD a year across platform, benefits and analysis hours |
Loyalty members already drive 39% of visits, and that changes the budget
Thirty-nine out of every hundred restaurant visits in the United States come from loyalty program members, twice the 2019 figure, according to Restroworks (2025) and confirmed by LoyaltyPass in its 2026 count. Read it as a budget line rather than a marketing win: if four in ten guests walk in because you called them back, then four in ten acquisition dollars are buying something you already owned. The Bogotá steakhouse I describe below paid 12.40 USD per new guest and 3.10 USD to bring a regular back; against a contribution margin of 9.80 USD per ticket, the first figure eats 126% of the margin and the second one 32%. This is not a philosophical choice. It is cash arithmetic, and you settle it with the P&L open in front of you. Restroworks measured that quick service generates close to 71% of sales from repeat customers, and that share is the practical ceiling a well-run table-service operation can chase without turning into a private club.
QSRs pull 71% of their sales from customers who come back
At a frequency of 1.39 visits per year — the real number from the 180-cover steakhouse that closed 2025 with 41,000 tickets and 29,400 distinct guests — the recurring base contributed barely 28% of revenue. That gap is the whole story. Lifting frequency to 2.4 annual visits across those same 29,400 guests adds 29,400 tickets: at 9.80 USD of margin per ticket, that is 288,120 USD of extra contribution, with no new site and no additional line cook. The cheapest lever in the business usually sits idle because nobody tracks it in the monthly P&L. A trend figure is useless until it becomes a weekly quota. That 71% translates into one target: how many of your tickets each week carry a known customer identifier. Lightspeed (Online Ordering Statistics 2025) puts at 45% the lifetime-value advantage of the guest who orders through your own channel over the one who only arrives via third-party web, and that spread is why aggregator commission should never be compared against the ticket, but against the guest's whole life.
Owned channel: 45% more lifetime value than the order arriving through an aggregator
Take two identical 28 USD orders. Through an aggregator you hand over between 4.50 and 8.40 USD in commission and keep no phone number; through your own channel you pay the gateway, roughly 0.90 USD, and you keep the data you will sell against again. Multiply that across the 41,000 transactions in the case: every point shifted from aggregator to owned channel frees close to 1,500 USD of annual margin. I got this wrong for years by advising operators to cut aggregators outright; migrating by daypart works better. The open discount punishes the guest who was coming anyway, and almost no program resolves that tension. A Tuesday two-for-one gifts margin to the regular who was paying full price: if 60% of that Tuesday's covers were repeat guests, you just bought visits you already had. The segmented perk does the opposite, charging full price to whoever discovers the place and rewarding the third-time visitor with a dish carrying a 29% food cost.
Open discount versus segmented perk: who pays for the party
The cash difference is blunt: giving away 50% of a 22 USD ticket costs 11 USD, while gifting a plate whose raw-material cost is 6.40 USD costs 6.40 USD and reads as more valuable. Against the 32% food cost per dish we treat as a hard ceiling, the margin holds. The two-for-one does not. And the brand effect is not symmetrical. A targeted gift gets told; a blanket discount gets expected. Some 63.1% of users discover products and trends on TikTok according to The Influence Agency (2025), the platform averages 220,800 views per food and beverage video against 135,200 on Instagram Reels — Restroworks figures (2025) — and short video accelerates audience growth by two to three times, which is also why Forbes ranks it as the fastest-growing restaurant discovery channel. All of that feeds the top of the funnel and NONE of it retains.
Short video and discovery: 63.1% find products on TikTok, but finding is not returning
A video with 200,000 views that captures no contact detail leaves you one-visit traffic, precisely the pattern that produced the 1.39 frequency. The rule I apply: every video piece points to a destination with capture, and when it does not, budget it as reach advertising rather than loyalty. Google Business Profile helps here, drawing 7 times more views than the restaurant's own site (Malou, 2025). Business Research Insights valued the restaurant gift card market at 36,817 million dollars for 2025, and Capital One Shopping calculates that cafés and restaurants hold 43% of gift card sales (2026). Those two figures together point at something rarely exploited: the card is advance working capital and, above all, a structured reason for someone to come back with a date attached. Sold at face value and redeemed against a higher average ticket — the usual pattern has the redeemer spending above the balance — the instrument funds inventory before the dish is ever produced.
The gift card, a 36,817-million market almost nobody uses to retain
At the steakhouse in the case, shifting 4% of those 41,000 transactions into gift format meant pulling forward close to 45,000 USD of cash and scheduling 1,640 identified future visits. My firm view: build the gift card before you sign up for a points app. A small single-site venue should ignore the 39% loyalty-visit figure as a target and chase one number instead: cost per repeat visit below 20% of contribution margin, which at 9.80 USD of margin means staying under 1.96 USD; with WhatsApp, direct reservations and a birthday notebook you get there without software. A mid-size three-site operation already needs a customer identifier in the POS and should read the QSR 71% recurrence as a warning flag: if your repeat base is not delivering at least 45% of revenue, advertising is masking a product or service problem. A twelve-site group starts with the Lightspeed spread, that 45% lifetime-value gap on owned channel, because at that scale every point migrated away from aggregators is worth six figures a year.
How to read these numbers in YOUR operation?
Diego F. Parra always opens the Masterestaurant loyalty diagnosis on the P&L, never on the CRM. Every figure in this document comes from dated, verifiable industry publications:
Restroworks (2024 and 2025) for recurrence, loyalty visits and video consumption; LoyaltyPass (2026) for member share; Lightspeed (2025) for lifetime value by channel; Business Research Insights (2025) and Capital One Shopping (2026) for gift cards; The Influence Agency (2025), Forbes and Malou (2025) for discovery and local search. The limits deserve saying out loud: most of this measures the United States market and skews toward chains with digitized data, so an independent Latin American table-service operator should treat them as order of magnitude and not as a goal. The Bogotá steakhouse numbers belong to a real client's year-end close and serve as an example of method, not as a statistical reference. Cross every benchmark against your own twelve months before you move a price.
Where the real difference sits, and it is not the software?
The difference is not owning a points app, it is budgeting retention cost as a P&L line that competes with paid media for the same money.
Once both lines sit side by side the comparison becomes obvious: 3.10 USD against 12.40 USD per visit. Open discounting punishes the person who was already coming. A Tuesday 2-for-1 gives away margin on the regular who paid full price, while a segmented benefit does the opposite, charging full price to the newcomer and rewarding the third-time guest with a 29 % food cost plate. Frequency is the cheapest growth lever a restaurant owns and almost nobody treats it as one. Moving from 1.4 to 2.4 annual visits across 29,400 guests adds 29,400 checks without a square metre of new space or one more cook on the line. I got this wrong for years: I believed loyalty was bought with generosity.
Where the real difference sits, and it is not the software — in practice?
It is bought with operational consistency —steady service times, the same doneness, a bill without surprises— and the benefit only accelerates what the operation already sustains.
A loyalty program built on uneven service speeds up the exit, not the return. Break-even moves with frequency, not with check size. Lifting average check 10 % means rebuilding menu and prices; lifting frequency 10 % means a message and a reason to come back. The second costs a fraction and never touches food cost.
Head to head: buying traffic against reactivating the base
What the till shows BEFOREDiagnosis
- The point of sale treats the regular and the stranger as one guest, and hands both the same discount.
- Promotions get decided by calendar —slow Tuesday, 2-for-1— rather than by margin, so the hook dish usually breaks the 32 % food cost ceiling.
- The whole marketing budget enters the sales funnel at the top: paid media, aggregators, commissions of 22 % to 30 %.
- Nobody knows what share of last night's sales came from someone who had been there before, because nobody tracks it.
- The metric being celebrated is check count, which climbs while margin per check slides.
What the till shows AFTERMasterestaurant
- Customer acquisition cost splits into two P&L lines: buying new traffic and reactivating the owned base.
- The benefit triggers on frequency —third visit, sixth visit— and gets priced against the contribution margin of the dish that delivers it.
- Guest LTV is calculated over 24 months and governs how much you may pay to bring someone in the first time.
- Menu engineering feeds the program: benefits are paid with low food cost, high rotation dishes, never with the priciest cut on the menu.
- The weekly dashboard carries frequency, 60-day repeat rate and margin per identified guest, in that order.
Side-by-side comparison
| BEFORE · no loyalty system | AFTER · loyalty measured in margin | |
|---|---|---|
| Customer acquisition cost per guest | ✕12.40 USD average across paid channels and aggregators | ✓3.10 USD per repeat visit activated from the owned base |
| Annual visit frequency | ✕1.39 visits per identified guest | ✓3.80 visits per identified guest |
| Guest LTV over 24 months | ✕27.20 USD of accumulated contribution margin | ✓74.50 USD of accumulated contribution margin |
| Average check, returning guest | ✕19.60 USD, identical to the new guest | ✓23.90 USD, 22 % above the new guest |
| Effective discount on sales | ✕11.4 % in open promotions available to everyone | ✓4.6 % in benefits segmented by frequency |
| Food cost of the promotional hook dish | ✕38 % on the open 2-for-1, past the 32 % ceiling | ✓29 % on the benefit aimed at the third visit |
| Share of sales from identified guests | ✕7 % of total sales | ✓34 % of total sales |
| Annual cost of the loyalty system | ✕0 USD, yet 68,000 USD a year in acquisition media | ✓9,600 USD a year across platform, benefits and analysis hours |
The figures behind the decision
“We came in with 41,000 checks and 29,400 guests, meaning 1.39 visits a year, plus 68,000 USD annually in paid media and aggregators. We cut media in half, built a frequency benefit on 29 % food cost dishes, and within eleven months frequency reached 2.7, sales from identified guests went from 7 % to 34 % and operating margin climbed 4.1 points. No new location, no extra cook.”
How to read these numbers in YOUR operation
At 8,000 to 14,000 checks a year, skip the platforms: a spreadsheet with phone or email, visit date and spend is enough. Divide checks by distinct guests and you have real frequency; anything under 1.8 means your problem is return, not traffic. Your loyalty spending ceiling is 20 % of annual margin per guest: with 9 USD of contribution margin and two visits, that is 3.60 USD a year per person. A 1.10 USD dessert on the third visit fits there. A points card nobody redeems does not.
Attribution breaks first here: the guest eats at the north location and orders delivery from the south one, and you count them twice. Unify identification before spending a cent on benefits, because a CAC built on duplicated records lies to you by 25 % to 40 %. Split the marketing P&L into acquire and reactivate, then compare cost per visit between both every month. If reactivating costs more than a third of acquiring, the benefit is mispriced or you are delivering it with the wrong dish.
At volume, guest loyalty turns into capital allocation: every location owns its frequency and its guest LTV, and money belongs where marginal return is highest, which is rarely the biggest unit. Model cohorts by quarter of first visit and track accumulated margin at 6, 12 and 24 months. A twelve-unit group running 400,000 annual checks that lifts frequency from 1.5 to 2.0 generates 133,000 extra checks, and at 9 USD of margin that is 1.2 million USD without construction.
Sector figures come from published research by the National Restaurant Association, Technomic, Bain & Company and Harvard Business Review, with the publication year stated on each data point. Operating ranges for food cost, loyalty spending ceiling and margin per guest come from Masterestaurant costing doctrine, which treats Diego F. Parra's field record as interpretive criteria and never as a statistical sample.
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
Ecosystem tools that take this to the till
Three pieces of the Masterestaurant method carry the weight here: the one that orders the business model, the one that sizes growth, and the one that watches cash while frequency climbs.
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 loyalty economics
How much can I spend per year to keep one guest?
How much can I spend per year to keep one guest?
Up to 20 % of the contribution margin that guest leaves over twelve months. If your margin per check is 9 USD and the guest comes three times, they leave 27 USD and your ceiling is 5.40 USD a year across benefits, platform and analysis hours. Cross that ceiling and the program becomes a discount in disguise.
How do I calculate guest LTV without expensive software?
How do I calculate guest LTV without expensive software?
Multiply contribution margin per check by annual frequency by the years the guest stays active. At 9 USD, 3.8 visits and two years, LTV is 68.40 USD. That figure is the theoretical maximum you may pay to acquire them, and in practice staying under a third of it keeps the payback inside the fiscal year.
Do stamps and points work, or are they wasted money?
Do stamps and points work, or are they wasted money?
They work when the reward is paid with a low food cost dish and fires on the visit where guests usually drop off, which in table service is the third. They fail when the reward is the menu's star item, because then you give away your best margin to the very person who was coming back anyway.
Which metric should I check first every week?
Which metric should I check first every week?
The 60-day repeat rate: what share of last period's identified guests came back. It is the earliest signal of operational decay and it moves weeks ahead of sales. If it drops two weeks running, audit service times and kitchen consistency before touching a single marketing dollar.
Sector data 2026 (official sources)
Verifiable industry benchmarks from official, non-commercial sources (government, industry associations, market research) - not competitors.
| Metric | Benchmark 2026 | Source |
|---|---|---|
| Reseñas del top-3 del local pack de Google | 47 reseñas más en promedio que los puestos 4 a 10 | BrightLocal 2025 (Google Reviews Study) |
| Tasa de clics de email en restaurantes y cafés | Click 1,06% y click-to-open 3,28% (de las más bajas por industria) | Mailchimp 2025 |
| Tráfico de menús de valor | +1% en el trimestre a junio 2025 (el tráfico total cayó 1%) | Circana 2025 |
| Precio como incentivo de visita | 50% de quienes no salían a comer volverían con precios más bajos | Circana 2025 |
| Alcance del segmento fast casual | 9 de cada 10 consumidores visitaron un fast casual en los últimos 6 meses (2025) | Datassential 2025 |
| Caída de la frecuencia de salir a comer | 37% de los estadounidenses salen a comer menos seguido en 2025 | Morning Consult / NRN 2025 |
Related content
Grow your restaurant with the Masterestaurant method
Applied in +8.400 restaurants across 43 countries.