How to track calories when eating out — without guessing

Restaurant food is where most calorie logs quietly fall apart. There's no label, no scale, and the "Chicken Caesar Salad" entry someone typed into a crowdsourced database describes a different restaurant's salad from a different year. But you don't have to choose between skipping the log and inventing a number. You need a method.

Decomposition: breaking a dish into its component ingredients and estimating each one separately, instead of guessing one number for the whole plate.

Decomposition beats whole-dish guessing for a simple reason: single-ingredient errors partially cancel out, and each component can be anchored to a real USDA record instead of a vibe.

Step 1 — name what's actually on the plate

Look at the dish and list what you can see or reasonably infer: the protein, the carb, the fat sources (oil, butter, cheese, dressing), the vegetables, the sauce. A "chicken burrito" becomes: tortilla, chicken, rice, beans, cheese, sour cream, salsa. Seven cheap estimates are more accurate than one expensive guess.

Step 2 — estimate grams against references you always carry

Step 3 — price each line with USDA numbers

With grams attached, the calories are arithmetic against USDA FoodData Central values:

That last row is the one restaurants hide from you.

The invisible calories: fat is the error bar

Restaurant kitchens cook with far more fat than home cooks — it's the cheapest way to make food taste good. Oil is 884 kcal per 100 g (USDA), so a tablespoon (~13.5 g) is about 120 kcal and leaves no visual trace on the plate. Practical rule: for any sautéed, pan-fried or grilled restaurant dish, add 1–2 tablespoons of oil (about 120–240 kcal) beyond what you can see. Creamy dressings and aioli are in the same family: mayonnaise runs about 680–700 kcal per 100 g, so the "small" ramekin holding 40 g is ~275 kcal.

Accept the error bar — then shrink it with volume

A decomposed restaurant estimate typically lands within 15–20% of the truth. On a 900 kcal meal that's ±150 kcal. That sounds bad until you remember what the log is for: the weekly average. If your estimates are honest (not systematically low), errors scatter in both directions and the weekly picture stays useful. What ruins a log isn't imprecision — it's skipping the meal entirely, which records it as zero.

How Calorie.one does this for you

This decompose-estimate-look-up loop is literally what Calorie.one automates. Type "chicken burrito with rice, beans and sour cream" or photograph the plate, and the AI produces the ingredient list with gram estimates — then each line is priced against USDA FoodData Central, not guessed by the model. You see every line before saving, can correct a portion the AI got wrong, and each line is labeled usda, mixed or ai so you know exactly how firm each number is. The 30-second version of a skill that takes months to build by hand.

FAQ

Should I even bother logging a meal I can't measure?

Yes. An honest ±20% estimate keeps your weekly average meaningful; an unlogged 900 kcal meal enters your log as 0 kcal — a 100% error in the direction that always flatters you.

How many calories should I assume for cooking oil at a restaurant?

Add 1–2 tablespoons (about 120–240 kcal) to any dish that was sautéed, pan-fried, or grilled with a sheen on it. Deep-fried items absorb far more; USDA cooked values for fried foods already include typical absorbed fat.

Are chain-restaurant published calories accurate?

They're a reasonable anchor — chains in the US are required to publish calories, and those figures beat blind guessing. Portion variation between outlets still applies, so treat them as a midpoint, not a certainty.

Is it better to overestimate or underestimate restaurant meals?

Aim for honest, not conservative. Systematic overestimating causes its own problem: it can push you to eat less than planned or erode trust in the log. If you must round, round the fat portions up — fat is where restaurant meals hide their surprises.