Customer loyalty: the mistakes that cost you margin and the right method for your profile

Best option for a room under 15 tables: your own guest list with a product reward

If you run fewer than 15 tables and bill under USD 40,000 a month, your best loyalty option is an owned guest list with a costed product reward, not a points app: it runs between USD 0 and 40 a month against the USD 79-249 charged by commercial platforms, and the annual saving of USD 468 to 2,508 equals the contribution margin of roughly 26 to 138 average checks of USD 28 at 65%. Bain & Company sets the frame: lifting retention by 5% moves profit between 25% and 95%, and that wide range depends almost entirely on what the reward costs you. With your own product you fix that cost at USD 0.90 to 3 and you control it; with a percentage discount the guest decides it. Stripo measured in 2025 that 55% of diners are swayed by a well-built promotional email, so the cheap channel already works.

Where you book the redemption: the P&L line that flips the verdict?

Charge the cost of the giveaway INSIDE food cost for the period, in its own subaccount, and the real effect shows up in month one.

A program that gives away food and books it as marketing spend produces a P&L that lies with decimal precision: the kitchen reads 29% when its true cost is 33%, and you price your menu off a false number for quarters. Four points of food cost on USD 40,000 of monthly sales are USD 1,600 nobody sees and nobody explains. For years I recommended generic points programs because that was what the market offered, and I got the costing wrong; once we reclassified redemptions where they belong, three out of four programs that looked profitable stopped being profitable. The accounting reclassification is the starting point, not the choice of tool. Give away a defined product rather than a percentage: a dessert costing USD 2.80, a starter at USD 3 or a specialty coffee at USD 0.90 turn an open variable into a figure you control.

Why a specific product beats a percentage off the check?

At 15% off a USD 28 check you hand over USD 4.20 of sales, and since that sale carried a 65% margin, the hit to contribution is USD 2.73;

if the table orders wine and the check climbs to USD 90, that same 15% costs you USD 8.78 in contribution without you deciding anything. The USD 2.80 dessert costs USD 2.80 on a Tuesday and on a Saturday. That is the difference between a fixed cost per redemption and a bleed proportional to the check, and it explains why the same program performs differently in two rooms with identical sales. A guest who returns four times a year instead of twice does not double your sales, he doubles your contribution, because the cost of bringing him in the first time is already paid and never repeats. Run it with your own figures: you pay USD 12 of acquisition to bring in someone who spends USD 28 at a 65% margin, meaning USD 18.20 of contribution, leaving you USD 6.20 before payroll and rent.

The number that decides it: what the second, third and fourth visit are worth

If he comes back three more times, accumulated contribution rises to USD 72.80 against the same USD 12 of acquisition. Harvard Business Review and the National Restaurant Association agree that winning a new guest costs five to seven times more than keeping an existing one, and yet independent operators still push their budget into paid acquisition, with restaurant Google Ads CPC already at USD 2.05 according to PPC Chief. Three scenarios exist where the popular option wins and paying USD 79-249 a month makes sense. First, if you run more than 40 tables or three locations: manually handling a list of 4,000 contacts eats six to eight hours of admin a week, and at USD 12 an hour that is USD 288 to 384 a month, dearer than the platform. Second, if your own delivery carries more than 35% of sales, because you need channel attribution a spreadsheet cannot give you, and the market is growing 8.6% a year in Latin America according to Grand View Research.

When NOT to choose the owned list and to pay for the platform instead?

Third, if your average check tops USD 60 with wine and your guest expects a named program with a card; there the platform cost dissolves into fewer than four monthly checks.

Outside those three cases, paying the subscription buys a feature your volume does not justify. When you compare vendors, distrust four concrete signals from the trade. The first: if the contract does not let you EXPORT your guest database as CSV, you are not building an asset, you are renting one, and the day you leave you lose everything. The second: a per-transaction commission stacked on top of the monthly fee — 2% on USD 40,000 is USD 800 added to the USD 149 subscription. The third: the vendor talks about engagement and active users but never shows you visit frequency or member versus non-member average check, the only two metrics that move your P&L. And the fourth, the priciest: programs that give away a percentage with no cap per check, because your biggest sales night of the year is also your biggest bleed.

Four red flags when comparing loyalty programs

Diego F. Parra and the Masterestaurant team reject any program that fails the first one. If you have a neighborhood coffee shop with a low check, between USD 8 and 14, a stamped physical card with a reward on the tenth visit suits you: cost per redemption of USD 0.90 against USD 90 to 126 of accumulated sales, under 1% of the revenue it generates. If you run a tablecloth restaurant with checks of USD 28 to 45, the email list with priority booking and a complimentary starter on the third visit works better, because Lightspeed measured in 2025 that guests ordering or booking through a digital channel visit 67% more often. And if you handle a high-volume operation above 200 checks a day, then the platform with POS integration does pay for itself. The question that settles the decision is not which program is better, but how much margin you are willing to hand over for each additional visit you buy.

What separates a program that pays from one that bleeds?

The difference starts on the P&L line where you record the redemption. A program that gives away food and books it outside food cost produces a statement that lies with decimal precision:

the kitchen shows 29% when its real cost is 33%, and you price your menu off a false number. Charge the giveaway inside the period's food cost, under its own subaccount, and the true effect surfaces in month one. Second difference, the reward type. Giving away a percentage of the check transfers margin without control, because the guest decides the amount the discount applies to and you decide nothing. Giving away a specific product —a 2,80 USD-cost dessert, a 3 USD appetizer, a 0,90 USD specialty coffee— turns an open variable into a constant. On a 28 USD average check, option one costs 5,60 USD and option two costs 3 USD, while the guest perceives equal or greater value because the menu price of the gift exceeds its cost.

What separates a program that pays from one that bleeds — in practice?

Third difference, the most expensive of all: who owns the data.

A restaurant billing 60% through aggregators without a single email address from those guests is renting out its own customer base at 18-30% commission, month after month, with no expiry date. According to Charlie Jeffers, co-founder of Owner.com, most independent operators find out too late that the customer base they assumed was theirs actually lives inside the aggregator's platform. Reclaiming that data is the highest-return hospitality growth lever available today. Fourth difference: where the business stands. A restaurant that just opened has no customer loyalty problem, it has a product-trial problem and an unknown baseline frequency. Standing up a points program in week 3 spends cash retaining people who have not decided whether they will return, and contaminates measurement for a full year. Measure ninety days of clean frequency first, decide after. And the fifth, a matter of judgment rather than tooling: customer loyalty never compensates for a broken operation.

What separates a program that pays from one that bleeds — key points?

If your food cost sits at 36% and prime cost above 68%, every extra visit the program generates multiplies a loss. Correct order: menu engineering, then cost structure, then sales funnel and loyalty.

Reversing it is the most elegant way to accelerate toward the cliff.