The Real Cost of Restaurant Discounts

The number nobody works out first

Discounts get decided quickly. Bookings are soft, a competitor has an offer on, someone suggests 20% off midweek, and it goes out on Thursday. What almost never happens in that conversation is anyone calculating how many extra covers the offer has to produce before it stops costing money.

The answer is uncomfortable. At a 70% gross margin, a 20% discount needs 40% more covers to hold contribution flat. A 25% discount needs 56% more. A half-price offer needs two and a half times the volume — three and a half times the covers you would normally do — before it breaks even.

These are not pessimistic estimates. They are arithmetic, and they hold regardless of how well the promotion is marketed. A discount reduces the contribution on every cover it touches, and volume is the only lever available to make that back.

The reason the numbers surprise people is a habit of thinking about discounts as a share of price. But price is not what you keep. If a $42 check carries $12.60 of food cost, you keep $29.40. Take 20% off the price and the guest pays $33.60 while the food cost stays exactly where it was, so you keep $21.00. Price fell by a fifth; the money that funds the business fell by more than a quarter.

Why the discount eats margin, not price

The mechanism is simple once seen. Your cost of goods is fixed per plate. A discount cannot touch it. So the entire reduction lands on contribution, which means the percentage impact on margin is always larger than the headline percentage.

The lower your margin, the more violent the effect. At a 75% gross margin, a 20% discount leaves you 55 points of contribution — you keep 73% of what you had. At a 55% margin, the same discount leaves 35 points, or 64% of what you had. And at any margin, a discount equal to or larger than your margin means every discounted cover loses money outright, no matter how many of them you serve.

This is why offers copied from another operator so often go wrong. A high-margin bar can run a promotion that would be structurally impossible for a steakhouse. The headline looks the same on the poster and behaves completely differently on the P&L.

It is also why the discount conversation should start with the current margin rather than with the offer. If you do not know your contribution per cover to within a dollar, you are not in a position to price a promotion.

Cannibalisation is where most of the money goes

The break-even uplift assumes discounted covers are additional. In practice a large share of redemptions come from guests who were already coming. Those covers generate no incremental sale at all — they simply pay you less for the same meal, and every one of them raises the number of genuinely new covers the promotion needs.

The worst version is a discount running during peak trade. Cannibalisation approaches total, because everyone in the room would have been there anyway, and there is no capacity to add covers even if demand existed. The service looks identical to a normal Saturday and the day's contribution is materially lower. This is the single most common way restaurant discounts destroy money, and it is invisible unless someone compares against a baseline.

The corollary is that targeting matters more than depth. A 30% offer sent only to guests who have not visited in six months cannibalises very little, because those covers were not in the forecast. The same 30% offered to everyone who walks in is redeemed mostly by regulars. Deeper and narrower beats shallower and broader, almost every time.

Before running anything, decide honestly what share of your normal covers will use the offer. If the answer is most of them, the promotion is a price cut wearing a promotion's clothes.

The costs that do not show up in the calculation

Even a promotion that hits its break-even cover count can still lose. Several effects sit outside the basic arithmetic:

• Lower spend per head. Guests attracted by price tend to order less around the discounted item, so the pre-discount check falls too. Track your average check through the promotion, not just cover counts.
• Labour steps up. Contribution margin treats labour as fixed, but 40% more covers usually needs another body on the floor and another in the kitchen. That pushes the real break-even above the arithmetic one.
• Capacity limits. If the room and kitchen cannot physically deliver the required uplift in the window, the promotion is unsound before it starts.
• Quality under load. A discounted guest having a slow, chaotic first visit is a negative return, not a break-even one.
• Price expectation. Repeat discounting teaches guests what the meal is really worth to them. Once a segment has learned to wait for the offer, full-price demand on the other days falls.

The last effect is the most expensive and the slowest to appear. It rarely shows in the promotion's own numbers, because it damages the weeks around it instead.

What to do instead of cutting price

Almost every objective a discount is meant to serve has a cheaper route.

• Add value rather than removing price. A giveaway item costing $4 against a $42 check costs 9.5% of the check. A 20% discount costs 20%. Guests frequently rate the giveaway as the better deal.
• Raise the check instead. Moving average spend up a few dollars through better upselling and menu structure produces the same contribution as a large volume increase, with none of the capacity strain. The practical steps are in our guide to increasing average check.
• Fix the menu mix. Steering guests toward high-contribution dishes changes profitability without touching price at all. Menu engineering does this systematically, and the menu engineering calculator shows which dishes are worth promoting.
• Shift demand rather than create it. An early-sitting price that moves guests out of your busiest hour can raise total covers without needing more capacity.
• Discount capacity, not the menu. Offering a lower-priced set menu at genuinely quiet times leaves your full-price proposition intact and puts contribution into hours that were producing none.

That last distinction is the important one. Selling a different, cheaper thing at a quiet time is a pricing decision. Selling your normal thing for less to everyone is a margin decision, and margin decisions are much harder to reverse.

When a discount genuinely makes sense

Discounting is not indefensible. It is defensible when it does a job that the margin cost is worth paying for, and there are a few clear cases.

Idle capacity is the strongest. A Tuesday lunch running at a quarter full has near-zero cannibalisation and fixed costs already committed, so any positive contribution is an improvement on an empty room. Here the break-even uplift is close to irrelevant, because there is almost no baseline contribution to protect.

Guest acquisition is the second, but only when it is measured properly. A first-visit discount pays for itself if the second visit happens at full price. That makes return rate the metric, not redemption volume — and it means you need a way to identify the guest again, which most in-house offers do not provide.

Clearing perishable stock is the third, and it is not really a discount at all. A special built around an ingredient that would otherwise be wasted is recovering cost, and it should be judged against the alternative of binning it rather than against your standard menu margin.

Outside those cases, the honest test is whether you can name the number of incremental covers the offer needs and believe the room can deliver it. If the answer is no, the discount is a way of feeling busy while making less money, and the busyness is the part everyone remembers.

Frequently asked questions

What is a safe maximum discount?

There is no universal figure, because it depends entirely on your margin. A useful rule is to stop well before the discount reaches half your gross margin: at 70% margin, a 35% offer already needs the volume to double. Beyond that point the required uplift climbs so steeply that most rooms cannot physically achieve it.

How do I tell whether a promotion actually worked?

Compare covers and total contribution against a genuine baseline — the same day-part in a comparable recent period — not against zero. Redemption count and a full room prove attendance, not profitability. If contribution during the promotion was below the baseline, the offer cost you money regardless of how it felt on the night.

Is happy hour a discount?

It is, but usually a well-structured one. It targets a specific quiet window, so cannibalisation is limited to guests who would have arrived at that exact hour, and beverage margins are high enough to absorb a meaningful reduction. Happy hour goes wrong when it drifts into hours that were already busy.

Do discounts damage how guests see the restaurant?

Frequent, predictable ones do. Guests use price to infer quality, and a site that is always on offer becomes a site whose full price looks arbitrary. Occasional, clearly bounded promotions with a stated end date carry far less of this risk than open-ended ones.

Are loyalty schemes better than straight discounts?

Usually, because the reward is earned after several full-price visits rather than given at the first. That spreads the margin cost across more covers and rewards behaviour you want repeated. The cost is still real, so calculate it as an effective discount across the full earning cycle.

Should I discount to fill quiet shifts?

It is the best case for discounting, with one condition: the offer must be restricted to those shifts and enforced. An offer that leaks into busy periods converts full-price demand into discounted demand, which is exactly the outcome you were trying to avoid.

What about discounts on delivery orders?

Treat platform commission as part of the discount. A 20% offer on a channel taking 25% commission behaves like a 45% reduction against your margin, which very few menus can carry. The economics are covered in more detail in our piece on delivery profitability.

Run the numbers

Use the free Discount Impact Calculator to apply everything above to your own figures.