Benchmark

How much profit do restaurants really lose to leak?

Published July 28, 2026 · OpslyIQ

Restaurants lose an estimated 4–7% of sales to “profit leak” — the combination of void and comp abuse, food-cost variance, and labor drift — and because industry net margins run just 3–5%, that leak is often larger than the profit itself. Total restaurant fraud alone averages around 4% of sales, and total shrinkage from waste, theft, and error can reach roughly 20% of profit. The losses persist not because operators are careless, but because the data that would expose them lives in separate systems that never get read together.

Key findings

Where the leak hides — three legs

1. Voids & discounts — ~0.5–1.5% of sales

The National Restaurant Association puts total restaurant fraud near 4% of sales. Void and comp abuse — cancelled or discounted sales used as cover — is one slice of that. A healthy void rate is generally benchmarked under 1–2% of sales; consistently above that points to error or intent. Counted conservatively, the recoverable void/comp slice lands around 0.5–1.5% of sales — deliberately well under the 4% headline.

2. Inventory & food-cost variance — ~1.0–3.0% of sales

Food and beverage cost runs about 32% of sales, and 10–15% of food spend is lost to waste in professional kitchens — roughly 3–5% of sales before you count theft or over-portioning. The recoverable theoretical-vs-actual slice — the gap between what you sold and what you actually used — is a portion of that, conservatively 1.0–3.0% of sales.

3. Labor drift — ~0.5–1.5% of sales

The American Payroll Association associates time theft with 2.2–5% of payroll — early clock-ins, buddy punching, overtime creep. With labor at roughly 30% of sales, that translates to about 0.7–1.5% of sales, and manual time-clock operations can lose 5–10% of payroll annually. Scaled gently by headcount, the conservative labor slice is about 0.5–1.5% of sales.

Add the conservative legs and you get the 4–7%-of-sales headline — against a 3–5% net margin.

It’s not the same for every venue

Leak exposure scales with the service model, which is why the benchmark shouldn’t be a single number. Bars and lounges carry the most — high cash exposure, high void discretion, and overpour that can reach 10–20% of inventory monthly. Full-service is the baseline. Quick-service runs lowest, with tighter comp discretion and portion-controlled inventory. (These map to OpslyIQ’s estimator modifiers: bar/lounge ×1.35, full-service ×1.0, fast-casual ×0.85, quick-service ×0.7.)

A worked example

A full-service restaurant doing $80,000/month ($960k/year), at the conservative combined rate, is exposed to roughly $1,600–$4,800 per month of leak — on the order of ~$38,000 a year, mid-range. A bar at the same volume carries meaningfully more; a quick-service location, less. None of these is a claim about your restaurant — it’s a benchmark for an operation of that shape. Your real number comes from your own POS.

Why it stays hidden

The losses above aren’t exotic, so why do owners typically discover them at tax time? Because each signal lives in a different system: the POS holds the void, the schedule holds the labor, the bank holds the deposit, the count sheet holds the variance — and none of them was built to be read against the others. The leak that spans two or three systems is invisible to all of them individually. It only becomes visible when the clock-in, the void, the tip, and the deposit sit in one place and can be correlated.

Methodology & honesty note

This benchmark is a synthesis of published industry research, not a measurement of OpslyIQ’s own customer base. Every coefficient is taken at the conservative (low) end of its cited range — we under-claim on purpose — and outputs are expressed as ranges, never false-precise single figures. The three-leg model, the service-type modifiers, and the headcount scaling are OpslyIQ’s original framing, applied consistently in the Profit Leak Estimator. Figures were cross-checked against the sources below; restaurant benchmarks drift, so we date and re-check them. When we can responsibly publish first-party leak data measured across live accounts, we will — and we’ll label it clearly.

Sources


The benchmark gives you a range. Your POS gives you the receipts. Estimate your exposure in under a minute, then connect your POS on a free 14-day trial (no card) and see your actual leak — named to the shift it lines up with.

Estimate your leak · Start free — no card

OpslyIQ shows you where to look, with the evidence — it never accuses a person. You investigate; you decide.

← All guides