Restaurant promotion ideas: the numbers that decide which ones live

Which restaurant promotion protects margin, and which one eats it?

Added-value promotions built on dishes with low food cost, somewhere between 18% and 24%, protect margin; any percentage discount applied to dishes already sitting near 30% destroys it.

The arithmetic is stubborn: a 100 dish carrying 30% food cost leaves you 70 of gross margin, and a 20% discount does not cost you 20 points of profit but 27 points of your real margin, because the cut comes entirely out of what is left over and never out of the ingredient, which still costs exactly the same. A 2-for-1 on that same dish puts your ticket underwater from the first plate served. Give away a side whose food cost is 1.80 on a 100 check, though, and you buy the same feeling of generosity for under two points of margin. Your guest cannot tell the difference; your P&L can. Price moves traffic like no other lever: according to Circana 2025, 50% of people who had stopped dining out would return if prices came down, precisely half the market you lost to menu inflation.

Price still drives traffic, which is exactly why you must protect it

That figure cuts both ways, because it justifies running a promotion and at the same time tempts you to run it badly. The 50% confirms that price sensitivity is real and that dormant demand is waiting for a reason to come back; what it never says is that the reason has to be a flat cut on your signature plate. In the 26-table restaurant in Medellín I mentioned earlier, 2-for-1 Thursdays filled the room for three months and October's cash closed NINE margin points below July's. Expensive traffic bought, cheap margin sold. Operators in the 90th percentile pull more than 37% of their transactions from loyalty members, per Paytronix's Loyalty Trends Report 2024, and that number is what separates a promotion from a markdown. Turn it around: if over a third of your checks come from identified guests, you no longer need to discount blindly to fill a slow Tuesday, because you can send a surgical incentive only to the 400 diners who have not shown up in 45 days.

Loyalty is the promotion that pays for itself

The cost of that nudge spreads across future visits instead of across tables that were coming anyway. Here sits the trap almost nobody sees: an open promotion subsidizes your most loyal customer, the one who would have paid full price. Diego F. Parra hammers this with every owner who sits down with the Masterestaurant method: identify first, incentivize second. Any promotion that fails to capture a contact detail is money thrown away with style, and the digital evidence backs it: 72% of people use social media to research restaurants (Restroworks, 2025) and complete Google Business profiles earn 7 times more clicks (WebFX, 2026). You are already paying for that attention with time, design work and free food. When a guest walks in on the offer and walks out without you knowing a name, a phone number or a birthday, you bought a visit at customer prices. Lifetime value gets built on the second, the third and the ninth return, never on the first.

With no data captured, you bought a visit and not a customer

A three-field sign-up at the table turns an unrecoverable markdown into a reactivation list you will work all quarter. In Medellín, 71% of the 2-for-1 crowd never paid full menu price again, and there was no way to call them. Shifting orders from the aggregator to your own channel is, in practice, the most profitable promotion available, and the consumer already agrees: 70% prefer to order directly from the restaurant rather than through a third party, according to Paytronix (Online Ordering 2024), a figure Statista supports with 67% preferring the venue's own site or app. Concentration explains the urgency: DoorDash closed 2024 with 60.7% of the US delivery market against Uber Eats at 26.1% and Grubhub at 6.3% (Earnest Analytics), while iFood commands 80% of Brazilian delivery (Grand View Research). With commissions biting 20 to 30 points off the ticket, giving away delivery on your own channel is cheaper than handing that margin to the platform, and you keep the guest's data on top, which is worth more than the order itself.

How to read these numbers in YOUR operation?

Translate every benchmark to your own scale before approving anything. Small restaurant, 20 to 40 tables and one strong service:

forget the 37% loyalty figure from the 90th percentile and aim to identify 15% of your checks within six months using a manual sign-up; your real lever is that 1.80 side dish on plates running 20% food cost. Mid-sized operation, two or three venues: now you build 60-day repeat measurement and compare aggregate contribution margin for the campaign month against the previous month, covers against covers being useless. Group of five venues or more: your fight is DoorDash's 60.7% and the commissions attached to it; every point of orders migrated to your own channel beats any dining-room promotion, and marketing budget turns into technology budget. The numbers you just read come from four different kinds of source, and knowing which is which matters before you move a single price.

Where these benchmarks come from and what they do NOT tell you?

Paytronix and Earnest Analytics measure real transactions from US operators and platforms, so their delivery share and loyalty penetration are solid for that market and merely indicative for Latin America or Spain.

Circana and Restroworks work from stated surveys: people say they would return at lower prices, though saying it costs less than doing it. Grand View Research and Statista model aggregate markets, useful for size and trend, useless for the margin on your menu. None of these firms knows your food cost per dish. Benchmarks set direction; the decision comes out of your own contribution margin report, plate by plate. Before signing off on campaign artwork, demand one figure: contribution margin retained per incremental guest. Say your promotion adds 40 covers on a Thursday and 25 of those were coming regardless; the discount applied to those 25 is pure loss, and the remaining 15 must cover that loss plus the cost of the free food before you earn a single peso.

The question that settles this quarter's calendar

If the math does not close on a napkin, it will not close in December. And if none of those 40 left a contact detail, the experiment taught you nothing you can carry into the next campaign. Pick ONE low-food-cost dish from your menu this week, work out what giving it away really costs, and run it as an add-on above a minimum check, with sign-up required at the table. Measure at 60 days. A percentage discount is not a marketing cost. It is a straight subtraction from contribution margin, which is why it hurts three times more than it looks. Twenty percent off a dish at 30% food cost does not cost you 20 points of profit; it costs 27 points of your real margin, because the discount comes out of what was left over and not out of gross sales. That arithmetic changes the tone of the conversation in about thirty seconds.

Where most of the math breaks?

A promotion without data capture is money thrown away with style. If a guest walks in for an offer and walks out without leaving you a way to talk to them again, you paid for a visit rather than a customer;

guest lifetime value is built on the second, third and ninth visit, never on the first. A two-field sign-up — name, mobile, birthday — turns a one-day promotion into a twelve-month asset. Content that sells does not show the price, it shows the process. The Reels that convert best in this sector are close-up preparation clips with real kitchen sound, and the price or the mechanic appears at the end or in the caption, never as a giant headline over the food; the algorithm rewards watch-through, and a number covering the plate kills retention inside two seconds. The time slot decides whether a promotion creates revenue or eats it.

Where most of the math breaks — in practice?

Discounting Friday night, when you already run at 90% capacity, hands margin to people who were going to pay full price; discounting Tuesday at 7 p.m.

at 40% occupancy converts an empty chair — which costs the same either way — into positive contribution. This is the one rule I apply without exception. QR menus and physical menus do not compete. The printed menu governs service rhythm, menu narrative and suggestive selling at the table, while the QR complements it with delivery, accessibility, fast price changes and analytics on what gets looked at but never ordered. Replacing one with the other costs you the dish the server used to recommend while pointing at it. Keep BOTH, each in its role.