Restaurant profit margins: the real numbers

What net margin does each restaurant format actually make?

The bands below are industry rules of thumb, triangulated from operator P&Ls and public commentary, not promises. "Well-run" is the operative phrase in every row: the spread between a well-run and a badly-run outlet of the same format, on the same street, is wider than the spread between any two formats.

Two readings owners get wrong. First, percentage is not the business, fine dining's 5–10% on a big ticket can bank more rupees than a QSR's 15% on a small one; pick the format for the capital and market you have, not the prettiest percentage. Second, the cloud-kitchen row is really two businesses: the same kitchen nets close to nothing when every order pays 25–35% effective commission, and a healthy margin once multiple brands share the fixed cost and repeat customers order direct, the mechanics are in the cloud-kitchen guide.

What does a healthy restaurant P&L look like, line by line?

The anatomy, with the band each line should hold in a healthy operation:

Add the midpoints and roughly 85% of every rupee is spoken for before profit, which is the single most important fact in this business. There is no fat line to cut; there are five thin lines that each drift two points when nobody is measuring, and five drifting lines is the difference between 12% net and a loss. That is also why "we are busy" and "we are profitable" are different sentences: a full room with food cost at 38% and an unwatched commission line loses money in front of witnesses.

Make it concrete. On ₹10 lakh a month of net revenue, with every line holding a respectable point in its band: ₹3 lakh to ingredients (30%), ₹2.2 lakh to the team (22%), ₹1.3 lakh to the landlord (13%), ₹1 lakh to electricity, gas, repairs and everything else (10%), and ₹1 lakh to the aggregators (10% of blended revenue). That is ₹8.5 lakh gone and ₹1.5 lakh kept, 15%, the very top of the well-run band, achieved only because every single line behaved. Let food cost slip to 33% and staff to 24% and the same busy month keeps ₹1 lakh less. Nothing broke, nobody stole, no customer complained, two lines drifted, and a third of the profit left the building.

The bands also interact. A delivery-heavy operation trades rent (lower) for commissions (higher); a bar-led room trades licence-heavy "other" costs for beverage margins that flatter food cost. Judge your P&L against your own mix, but if any line sits outside its band for two consecutive months, that line is the project.

One more honesty rule before the bands mean anything: pay yourself inside the P&L. A large share of "profitable" small restaurants are really paying the owner nothing for sixty-hour weeks, a 12% net that depends on free owner labour as biller, buyer and closing manager is not a 12% business, it is a job with inventory. Put a market salary for your own role into the staff line and see what survives. If the answer is still inside the band, you own a business; if not, you now know what to fix before you scale it.

Where does margin actually die? The four leaks

Restaurants rarely lose money in one dramatic place. They lose it in four quiet ones, and every leak has its own playbook:

1. Food-cost drift. The band is 28–32%; the drift is yields nobody re-measured, portion creep, prices that never moved after onion season, and wastage that never hit a report. Fixing it starts with knowing true per-dish cost, the food-cost guide walks the method, and the food-cost calculator does the arithmetic per recipe.

2. Theft and pilferage. Unchecked, staff-side leakage, cancelled bills pocketed, raw material walking out, vendor collusion, is enough to erase a good month's entire net. It is also entirely catchable with attributed audit trails and weekly variance checks; the named scams and their tells are in the theft-prevention guide.

3. Commission creep. Aggregator take starts at the quoted rate and grows, platform fees, ads, co-funded discounts, to 25–35% effective. On a delivery-heavy mix, that is the largest single line after food cost. The counters (negotiation, ONDC at 3–5%, a 0% direct channel) are in the commission playbook.

4. Mispricing. Menus priced by rounding to the neighbour's menu leave margin on the table twice: bestsellers priced too low, and low-margin dishes selling too well. Quarterly menu engineering with real recipe costs, helped along by the menu-price calculator, is the highest-ROI exercise available to a running restaurant.

Notice what all four have in common: none of them announce themselves. They are found by measurement or not at all.

Is a restaurant a good business in 2026?

Honest verdict: yes, conditionally, and the condition is instrumentation, not passion. The bear case is real: five thin lines, 85% of revenue pre-committed, aggregators taking a growing share of the growing delivery mix, and rent and wages repricing upward at every renewal. Restaurants that run on instinct and a bill book are structurally set up to lose money at full tables, and many do.

The bull case is equally real. Demand keeps broadening with every city tier, delivery has expanded the revenue ceiling past the dining room's four walls, and the tooling to run measured, recipe-level costing, daily reconciliation, per-channel P&L, now costs less per month than a single day of an unnoticed leak. The operators netting the top of the 8–15% band in 2026 are not luckier or larger; they are the ones who know their break-even cold, read their variance weekly, and treat every P&L line as a number with an owner. Two identical-looking restaurants on the same street can sit at opposite ends of that spread for years. The difference is rarely the food.

If you are deciding whether to enter, size the whole journey, capital, licences, timeline, with the opening guide, and pressure-test the plan against the leaner cloud-kitchen route before committing dine-in capital.

How do owners actually get to 15%?

Not with one heroic cost-cut, with a measurement loop that never skips a beat:

Daily: a day-close that ties out. Bills, payments, aggregator orders and cash reconciled every night, so a bad day is caught as a bad day and not as a bad quarter. This single habit separates the top of the band from the bottom more reliably than any menu or marketing decision.

Weekly: variance and staff cost. Ingredient-level stock variance against sales (leak detection, per the four leaks above) and payroll as a percentage of the week's revenue, the staffing guide covers what that line should hold and how scheduling to the sales curve keeps it there.

Monthly: the P&L against the anatomy table. Every line against its band; any line out of band for two months becomes the month's project, with a named owner.

Quarterly: re-engineer the menu. Recost every dish, rerun the matrix, reprice the plowhorses.

The loop is simple; sustaining it manually across a seven-day operating week is what breaks most owners. CountStand is an AI-native restaurant operating system for India, offline-first billing, KDS, inventory, GST & compliance, and an autonomous AI manager, in one platform, from ₹999/mo per outlet. In practice that means the POS produces the tied-out day-close automatically, recipe-linked inventory throws the weekly variance without a stock-taking marathon, and the AI manager flags "food cost drifted out of band" the week it happens rather than the quarter after. The loop still needs an owner who reads it, but reading takes minutes when the numbers assemble themselves. Watch the whole cycle on a demo.

What is your own margin math? Run it

Margin bands are the map; your break-even is the territory. The number every owner should know cold is daily covers (or orders) needed to cover fixed costs at your real contribution margin, after food cost and after commission:

Run it three ways: today's mix, a delivery-heavier mix, and a version with a growing direct channel. If break-even sits above 60% of realistic capacity in every scenario, the problem is structural, rent, pricing or format, and no amount of hustle fixes a structural P&L. If it clears comfortably and the margin still is not showing up, revisit the four leaks; the money is going somewhere, and it is findable.