Where the numbers come from
Every score in the app is one of the formulas below, run on the device over the menu's prices and portions. No secret weighting, no vibes.
The two honesty rules
✓ Verified — a price a person confirmed, or a chain published, within the last 30 days. The app shows the date and how many people agreed. Nothing else ever gets the green check.
🖩 Estimated — nutrition, margins, satiety, ingredient value and every buffet or per-pound projection. Drawn with a calculator, always.
Ranking a menu
Every item gets a score for the goal you picked, highest first. Ties break by the lower price, then alphabetically, so the order is stable. The bar under each row is that item's score as a share of the winner's. The last row is the sucker item; an item priced below its estimated cost is a loss leader.
Protein per dollar
Calories per dollar
Ingredient value
How much of your dollar is actually food. Ingredient value is the estimated retail cost of
what is on the plate; where we have no costing, we fall back to
price × (1 − margin%).
Stay full longest
Hours of not-being-hungry per dollar. About 200 kcal buys an hour for an average adult, then protein, fibre and sheer volume stretch it. Capped at 8 hours so a 4,200-calorie pizza cannot claim a day.
hours = clamp(0.25, 8, (calories ÷ 200) × (0.7 + 0.6 × satiety))
hours ÷ price = hrs/$
Lowest margin
Where the restaurant profits least — the item they would rather you did not order. Estimated from ingredient costs and published industry reports, and useful for comparing items on one menu, not as anyone's accounts.
Buffet break-even
How many plates of the good stuff it takes to out-eat the door price. Rice, bread and pasta are near-zero value and are excluded from the average — that exclusion is the whole point of the mode.
break-even = buffet price ÷ average plate value of the non-filler dishes
verdict: (your plates × average plate value) − buffet price = what you win or overpay
Pay by weight (hot bars)
Everything costs the same per pound, so the only question is how much protein is in a pound — and how much of that pound is sauce you could have drained.
value = protein per lb ÷ price per lb = g/$
drain saving ≈ (140 g scoop ÷ 453.592) × price per lb × drainable share
drainable share: 30% for sauce-heavy dishes, 5% otherwise
Anything under 3 g/$ — or any pure filler — lands in “Skip”. 6 g/$ and up is “Load up”.
Building an order
The combo builder is an exact bounded-knapsack search at one-cent resolution over the whole menu: the most of your goal that your budget can buy, up to six of any one item and three lines. Drinks and desserts are excluded. It also looks for bundles that cost the same as buying units (five slices for the price of a whole eight-slice pizza) and for upgrades whose extra dollars buy almost nothing.
What “verified” means exactly
- A price is verified for 30 days after the last confirmation.
- The count is the number of different people who reported the same price, within 2%, in that window.
- A report with a different price resets the window and the count — the price changed — and notifies everyone watching that item.
- A price published by a chain counts as verified for 30 days, recorded as a system confirmation rather than a person.
Where the inputs come from
- Prices — people in the app, chain-published menus, and menu-board scans.
- Nutrition — the chain's own figures, else a USDA FoodData Central match scaled to the portion, else a model estimate.
- Places — Google Places, for finding venues near you.
- Menu photos — read with Cloud Vision and structured with Gemini, then stored as reported (never verified) prices.
Spot something wrong? Verify the price in the app. A dated correction from a person beats our best guess, every time.