Customer loyalty in restaurants: the points-app myth versus real LTV

Chapter 1 — Is a points app loyalty, or just a deferred discount?

No: it's a deferred discount, and what it inflates is the vanity metric of "enrolled members," not real loyalty.

By the end of 2025, 80% of restaurants will run a loyalty program (LoyaltyPass, 2026) and 78% of adults have already downloaded a food app (National Restaurant Association), but enrolling isn't returning. In the operations I review, I run into the same scene every time: the tech gets paid for —the app, the stamps, the push— and nobody calculates how much margin is given away per redemption. Real loyalty is measured in LTV per diner: frequency times average check times margin times years of life. A member who redeems a two-for-one at 30% food cost isn't giving you loyalty, they're giving you a visit that pays for half the plate. Before applauding 110 sign-ups per store per month (Paytronix, 2024), calculate that cohort's incremental margin.

Chapter 2 — Why does retaining matter more than acquiring in 2026?

Retaining matters more because acquiring got 222% pricier over the eight years to 2025 (Marqii, 2025), a pace no healthy contribution margin absorbs forever.

And 33% of industry professionals name attracting and retaining customers as their top challenge for 2026 (Toast, 2026). If acquisition keeps getting pricier, every lost visit from an already-won guest is capital leaking out — that's the cruel arithmetic of it. But retention only adds up if you retain with positive margin. A program that hands out 20% off to bring back someone who was returning anyway doesn't retain, it subsidizes. We frame it plainly at Masterestaurant: first you measure the LTV of your cohorts, then you decide how much of that LTV you can reinvest in loyalty without touching prime cost. Retaining cheaply is margin; retaining expensively is a leak with a strategy label on it. Every redemption cuts direct contribution margin, a cost rarely shown in the sign-up report the vendor brags about.

Chapter 3 — How much margin does an app redemption really give away?

For years I looked at that report and celebrated the enrollment count myself, until a client made me work out the real cost of a redemption — that's when I switched metrics.

An US$18 average check at 30% food cost leaves US$12.60 of gross margin over ingredients; if the program gives away an US$8 plate after ten visits, the real cost of that redemption is the US$2.40 of ingredient plus the margin you chose not to charge. Multiply that by the guests who'd have returned without the reward, the free-rider effect, and the deferred discount eats 8-12% of the cohort's margin. Value per customer rises 23% with well-designed programs (Paytronix, 2024), but that 23% is gross: subtract the redemption and the net can land in the red. It doesn't: acquisition cost climbed 222% in eight years (Marqii, 2025), and the app just relabels customers you already had as "members." That blind spot gets expensive in more than one expansion board meeting.

Chapter 4 — Does the app lower acquisition cost? No, and that's the blind spot

The owner opens the dashboard, sees 110 new sign-ups per store per month (Paytronix, 2024) and concludes the program "acquires"; false — most of those sign-ups were already going to pay, now with a coupon on top. LTV management, by contrast, turns each member visit into recurring margin measured against a control group. The difference sits in the accounting: the app reports its ROI in sign-ups, a vanity sum, while the LTV framework reports it in incremental frequency and check against non-members. Without a control group there's no ROI, just an optimistic spreadsheet. Segmented email and SMS reactivate better, precisely where app push tends to fail on low open rates. Personalized email sends lift opens by 26% (Stripo, 2025) and the channel averages a 25.1% open rate (Omnisend, 2024); SMS, for its part, raises food-and-beverage engagement by 25% (Tabular, 2025). A points-app push competes with dozens of other notifications, and a good share of users mute it after the first week.

Chapter 5 — Which channel re-activates with traceable ROI instead of app push?

The operational edge of email and SMS is that you can trace them: tie each send to a reservation or a coded ticket and measure the real incremental.

At Masterestaurant we recommend shifting the app's "features" budget toward cohort segmentation, because an email to 60-day inactives with a margin-calibrated incentive outperforms a mass push that hands a discount to someone who was already coming every week. It's calculated by multiplying frequency times average check times margin times a one-year horizon, not by counting how many enrolled. Sector data confirms that well-run loyalty moves cash: 55% of restaurants report their members' checks grew faster than menu prices (Paytronix, 2024), and 39% of U.S. visits already come from members, double the 2019 share (Restroworks, 2025). Those figures, though, describe the winners, not the average app. A member who visits 12 times a year, spends US$20 and leaves 68% margin after food cost hands you a gross LTV of US$163 a year; subtract the cost of their redemptions and the cost of running the program.

Chapter 6 — What does a loyalty member's real LTV look like?

If the net beats what you'd spend to re-acquire them, ever pricier (Marqii, 2025), the program creates value. If not, you're buying frequency you already had.

Rewarding the discount trains your clientele not to come without a coupon; that damage is hard to reverse. The app hooks people on the mechanics of saving; LTV management anchors loyalty to the experience —the plate, the service, the table that turns well— a far less vulnerable edge in the price war across the street. Channel numbers show where the healthy lever sits: short video accelerates audience growth 2 to 3 times (Restroworks, 2025) and a food reel averages 135,200 views (Restroworks, 2025), while the Google profile gets 7 times more views than the site (Malou, 2025). That demand comes in through desire, not discount. Bluntly: if your traffic depends on the coupon, you don't have a brand, you have a promotion wearing a loyalty costume, and the day you switch it off you'll find out you were never chosen, just the cheapest.

Chapter 7 — How much to reinvest in loyalty without breaking break-even?

Reinvest only up to the point where the cohort's incremental margin still covers its redemption cost and break-even stays intact. Masterestaurant's operating rule is simple:

a plate's food cost shouldn't exceed 32% as a ceiling, and the program's effective discount adds to that variable cost, it never hides inside it. If a redemption pushes the average check's effective food cost above 32%, you're selling the margin you need to cover payroll and rent, fixed costs that belong to break-even, not to the plate. What happens under ingredient-inflation stress? That slack disappears fast: a 10% rise in food cost can turn a profitable redemption into one that destroys EBITDA. The question isn't whether you have an app, but how much incremental LTV it generates, measured against control, and how much of that LTV you can reinvest without touching prime cost.

Chapter 8 — The 6 differences that decide whether your program adds or subtracts EBITDA

The app optimizes sign-ups. The LTV framework, instead, optimizes incremental margin per cohort: vanity math versus cash math. Each app redemption cuts margin because it's a deferred discount; LTV management rewards frequency and check without touching prime cost. Acquisition cost rose 222% in 8 years (Marqii, 2025) and the app doesn't lower it. LTV retention, instead, turns every member visit into recurring margin. App push opens poorly. Segmented email (+26%, Stripo 2025) and SMS (+25%, Tabular 2025) reactivate with ROI you can actually trace. The app reports its ROI in sign-ups. LTV reports it in incremental frequency and check against a control group. Rewarding the redemption trains diners to expect the discount — that's the app's structural risk. Anchoring loyalty to experience, instead, mitigates that vulnerability.