A restaurant has too many variables for anyone to promise results — and Match Sous doesn't. What it does is bring into the open what today gets decided in the dark: where each dish sits against what your area charges, what was left when the month closed, and what changes when you adjust something. It's the method of people who ran the industry's largest groups, turned into continuous oversight. The decision is still yours — it just comes with a number behind it. And it starts with what's already in your hands: the menu.
The methodology running here was built by professionals who operated large restaurant groups and food-industry companies — the same analysis that today gets bought as an audit, costs a lot, happens once a year and ends in a document nobody opens again. Match Sous turns that method into a continuous process, inside the app, for any restaurant: from the ten-unit group to the diner, the bakery and the neighborhood café that never had access to this kind of analysis because it always cost more than it was worth.
Menu engineering, recipe costing, food cost, prime cost, vendor negotiation — criteria refined over years of real operations at groups and in the industry. What lived in a handful of people's heads became rules built into the product.
Classic consulting is a snapshot: once a year, expensive, out of date by the following month. Here the diagnosis is alive — it changes when your costs change, when your prices change, when the market around you changes.
The same criteria that run in one restaurant run across an entire chain, unit by unit, comparable side by side — without depending on a consultant's calendar or a manager's spreadsheet.
You start with the menu because it's already in your hands — no system integration, no staff training, no waiting for an implementation. It works for the thirty-dish menu and for the diner's chalkboard with twelve. From there the product follows the operation: the month that closes, what the kitchen actually spent, the bills coming due, and every location under the same yardstick.
Each dish compared with the real price distribution in your area, and cross-checked with the margin it leaves. The comparison leaves the screen as a PDF, a spreadsheet or a message — and a consultant can deliver it under their own brand.
See the comparisonWhat came in, what went to suppliers, what went to payroll — and what was left, down to the last line. From the second month on, every number gets its curve against your target line.
See the month-endThe product stops only counting what already happened: what's coming due, how much the business usually brings in over that stretch and — if you tell it how much cash you have today — how long the money lasts.
See what's aheadMore than one restaurant under the same yardstick: who carries the result, who drains it, and the same dish location by location — each measured against its own market.
See the locationsCrossing margin per dish with weekly volume, every item on your menu lands in one of four quadrants. It's the reading a spreadsheet doesn't give you — and it's where the decision to protect, push, rework or cut shows up by name.
The menu's champion. The screen says: protect the price and leave the recipe alone.
Volume that carries the operation, but with cost pushing back. Go after ingredients before touching the price.
A good dish hiding on the menu. The screen suggests reworking its description or placement — not cutting it.
A natural candidate to cut — unless it's part of the restaurant's identity. The reading is yours; the verdict isn't blind.
The dividing line on each axis is the menu's own median — a classification relative to your restaurant, not an absolute yardstick. A dish sitting on the line is flagged as a tendency, not a verdict.
With sales volumes and cost per dish (entered by you or estimated by AI), the platform consolidates the entire menu into a single verdict: your restaurant closes healthy, needs attention, or is critical. Nothing invented — it only covers dishes that already have data, and says clearly when something is missing.
A kiosk, a container or a food-hall stall doesn't compete with the sit-down restaurant down the street: it competes with whoever's twenty steps away. You declare the food court and the comparison shifts to your physical neighbors — every stall together, which is how the customer actually decides where to eat.
Type the dish and the area. Match Sous returns the statistical price distribution, the restaurants that make up the sample and where your price sits relative to them — if it's already on file.
The question every owner answers from memory and gets wrong. You close the month with the numbers you already pull into your spreadsheet — it takes a minute, and you can upload several months at once. Whatever you don't enter stays open; it never becomes zero. From the second month on, every number gets its own curve.
Every restaurant report tells you what happened. This one also looks ahead: you note what's coming due, and the business crosses it with what it usually brings in over that stretch. If you tell it how much cash you have today, the screen points to the day it gets tight — while there's still time to call the supplier and renegotiate.
With more than one restaurant, all of them go under the same yardstick and show up side by side in the closed month — and the same dish is compared location by location, each against its own market. Different prices across neighborhoods aren't a problem; what's worth watching is the same dish sitting in different places in each market.
Nobody types dish by dish: reading the menu and crossing it with the market is done by a model, and the spreadsheet the restaurant already keeps goes in as it is. These are this sample's numbers.
Extra regional coverage in Tamboré/Alphaville, Pinheiros and Itaim — mall food courts and street restaurants.
A vision model reads the menu image and returns each dish, description, portion weight and price, extracted.
A 300 g picanha against a 500 g rib on the same yardstick — for drinks, R$/100 ml.
You upload a photo of the menu or paste the text. A vision model extracts every dish, description and price — no manual typing.
Your area's price database comes from collecting public menus and grows with what gets uploaded to the platform. In this version it's preloaded, and the How it works page says exactly which part is collected data and which is still a calibrated reference.
Dishes with different names but similar descriptions are recognized as the same item. Each gets a median, a range, a sample size and your price's relative position.
The menu is the entry point, and it's still where you start. From there, the whole business: the month that closed, what the kitchen actually spent, the bills coming due, and the locations side by side. One module isn't ready yet — and it's precisely the one that keeps the rest from going stale.
Each dish against what your area really charges. What to protect, what to rework and what to cut — and the whole menu in a single verdict. With the recipe cost estimated by AI when you don't have it at hand, and descriptions rewritten to sell the dishes that carry the margin.
Once the month closes, you see what came in, what went to suppliers, what went to payroll — and what was really left, down to the last line. From the second month on, the curve shows where the business is heading. And if you say how much cash you have today, how long the money lasts.
The cost your menu promises against what the month actually spent, with the gap in currency and the dishes to start looking at. Your ingredient prices are stored with the date they took effect — which is what keeps recipe costs from quietly going stale.
Several restaurants under the same yardstick: who carries the result, who drains it, and the same dish location by location — each measured against its own market, because different prices across neighborhoods aren't a problem. The consolidated view knows which part of the result is yours.
Half of it already works: the products-sold report your point of sale exports goes in as it comes out — spreadsheet, PDF or a screenshot — and matches your menu's dishes on its own. What's missing is the last piece: the restaurant not needing even that. It's what we're building now, and it's what guarantees the numbers don't die in two weeks: data that depends on someone remembering to type it is data that stops being fed.
No screen saves a restaurant — there are too many variables, and anyone who runs one knows that better than any software. What Match Sous does is transparency and oversight: your operation's numbers in plain sight, week after week, and the industry's disciplines — menu engineering first — working in your favor. The decision is still yours. It just gets easier to make.
Start with my menu