How much does a discount really lift sales in food & drink?

Operators tend to assume a discount lifts sales roughly in line with the cut — 20% off, maybe 20% more sold — while promotional material often implies far more. The evidence contradicts both. In food & beverage the proportional lift is typically smaller than the discount, and a substantial share of discounts fail to cover their own cost. The sections below set out what the data shows, and the arithmetic behind it.

What the data says a discount actually does

Price sensitivity is measured as price elasticity: the percentage change in units sold for each 1% change in price. The most comprehensive review of food-sector elasticity — Andreyeva, Long & Brownell's synthesis of 160 studies — puts food away from home at an elasticity of about −0.8[1] (−0.81 in the review, though the underlying studies range widely), and that was among the most price-responsive food categories. As a first approximation that is roughly 0.8% more units for every 1% taken off the price:

Discount	Expected volume lift (approximate)
10% off	≈ +8%
20% off	≈ +16%
30% off	≈ +24%

The elasticity scaled to each discount — order-of-magnitude, not precise. Elasticity is a local measure, so treat the deeper rows as indicative: the true lift at 20–30% off is model-dependent and can run somewhat higher, but stays well short of retail's response. Individual items, sites and dayparts vary widely.

So a 20% price cut is associated with roughly a 16% volume increase — not the doubling that "20% off" can suggest.

Why food & beverage is less elastic than retail

The far larger figures sometimes quoted — elasticities of −2.5 and beyond, implying a 1% cut lifts units by 2.5% — are genuine, but they are drawn from grocery and packaged-goods scanner data. The Bijmolt meta-analysis of 1,851 elasticities put the cross-study average at about −2.6.[2] That figure isn't a like-for-like comparison — it's brand-level and amplified by shoppers switching between brands, whereas the food number is category-level — so part of the gap is how each was measured. But two structural differences make hospitality genuinely less responsive:

• No stockpiling. A supermarket shopper buys ten discounted units for later; a guest consumes the meal or drink now and can't bank it. The forward demand a price cut unlocks in retail simply isn't there.
• Brand and occasion outweigh price. Venue, location, occasion and loyalty drive the purchase far more than a few percent off, so price moves volume less.
• Capacity is finite. A full Friday can't sell many more covers at any price; there, a discount mainly hands margin back on demand you already had.

The profitability evidence: most deep discounts don't pay

The largest real-world test of deep discounting is the daily-deal era. In Dholakia's self-reported survey of 324 small and medium businesses, only about 55–60% of daily deals were profitable for the operator — roughly four in ten lost money outright, with restaurants among the weakest performers.[3] Two caveats matter: these were deep deals — typically ~50% off, with the platform then taking around half of what was left, an effective give-back near 75% — and the businesses skewed small, so the finding speaks to deep third-party promotions, not a routine 10–20% offer. The mechanisms still travel, though: deal-led customers are price-driven and seldom return at full price; the promotion displaces full-paying regulars; and the volume strains kitchen and service capacity. High redemption looked like success and frequently was not.

Why the lift rarely covers the cost

The reason is arithmetic. A discount is deducted from contribution — the margin left after an item's own cost — so the break-even volume is governed by margin, not by the headline discount. Take an item priced at £10 with £4 of product cost: contribution is £6. A 20% discount drops the price to £8 and contribution to £4 — so you need 50% more volume (£6 ÷ £4) simply to stand still. Set that against the elasticity evidence, which predicts a 20% cut delivers about 16% more volume. The expected lift recovers only a fraction of the volume the discount requires. That structural gap — not poor execution — is why so many discounts lose money, and it widens fast on thinner-margin lines.

How to read these numbers — and their limits

Three caveats keep this honest:

• They're central estimates, not precise predictions. The lift figures are an elasticity scaled to each discount, so read them as order-of-magnitude. The true response at 20–30% off is model-dependent — it can run somewhat higher — but stays well short of retail's.
• The evidence is broad, not hospitality-bespoke. The food figure is a category-level elasticity from mostly US studies up to the late 2000s; the retail figure is brand-level and measured differently again. No public, item-level promotional dataset for hospitality exists — these are the closest proxies, and a single discounted line could respond more strongly than the category average.
• They show association, not a guarantee. These are demand relationships estimated from real markets, not a promise for your next promotion. The one part that always holds is the arithmetic above: the break-even volume is fixed by your margin, whatever demand does.

What it means for operators

The implication holds whether you run a single site or a multi-site group: model the break-even before committing, then test it against an evidence-based lift rather than an optimistic one. If a realistic response — the food & beverage figures above — clears your break-even with margin to spare, the promotion has a credible case. If it does not, which is the typical outcome for deep discounts, the data says hold off, or reach for a lever that doesn't surrender contribution on every sale.