How to log a restaurant meal you didn't cook

Every food log has the same failure point. Six days go fine, then Saturday night happens at a restaurant, the entry becomes a research project, and the day gets skipped. Do that four times and the log no longer describes your month.

The fix is not better data. It is a rule you apply without thinking.

Why restaurant food costs more than the same dish at home

Two reasons, and both work in the same direction.

Fat. Professional kitchens cook for flavour, not for your log. Butter finishes sauces, dishes are tempered generously, and the ghee in a restaurant dal would alarm you if you saw it go in. A home dal at 150 kcal per katori can easily be 280 in a restaurant. This is the single largest reason a restaurant dish beats its home-cooked database entry, and it is the same mechanism described in how much oil is actually in your curry.

Portion size. Restaurant servings have grown steadily for decades. A single main course is frequently two of what you would serve yourself.

So the first rule is: when in doubt, estimate high. A log that is generous about restaurant meals and honest about home meals errs in the direction that keeps your weekly total truthful.

The default table

Stop researching individual dishes. Use defaults, adjust them once you have a few weeks of your own data, and apply them in four seconds.

• Restaurant main course, any cuisine: 800 kcal
• Rich or creamy main (butter chicken, korma, alfredo, biryani): 1,000 kcal
• Shared starter, your share: 300 kcal
• Naan or paratha: 300 kcal each; plain roti 120 kcal
• Rice, restaurant portion: 350 kcal
• Dessert, shared: 350 kcal; own: 600 kcal
• Beer, 330 ml: 150 kcal; glass of wine: 150 kcal; cocktail: 250 kcal
• Soft drink, full sugar, 300 ml: 130 kcal

These are deliberately round. Precision is not available here, and pretending otherwise is what turns a four-second entry into a ten-minute one. More on alcohol specifically in how to track alcohol without ruining the week.

Log it at the table, not at midnight

Memory is not neutral. Reconstructing a meal hours later systematically drops the bread basket, the second drink and the bit of someone else's dessert. The entry made while the plate is in front of you is closer to the truth even if the numbers are rougher.

If logging at the table feels awkward socially, log it in the taxi home. The point is to do it while the meal is still a picture rather than a summary.

When the restaurant publishes real numbers

Chains often do, and those figures are worth using — they come from actual recipes with measured quantities. Treat them the way you treat any label: subject to a tolerance, but far better than a guess.

Independent kitchens have nothing to give you. Asking a waiter produces a number invented on the spot, which feels authoritative and is not. Your own default is more honest.

Buffets, weddings and anything served in rounds

These defeat estimation entirely, because there is no plate to look at and no end to the meal. The honest approach is a flat number rather than an addition.

An Indian wedding dinner, eaten normally, is somewhere around 1,800 to 2,500 kcal. A hotel buffet breakfast is 900 to 1,400. A long lunch with drinks is 1,500 or more. Log the flat number and do not attempt to itemise it, because itemising a meal you grazed through for three hours produces a false precision that is lower than reality every time.

What helps more than accounting is a decision made in advance: eat the things you came for, skip the things that are merely available, and drink water between the others. That is a better lever than any entry you could write afterwards.

The weekly view is the one that matters

One restaurant meal a week at 1,200 kcal instead of your usual 600 adds about 600 kcal to the week. If your deficit is 500 a day, that is 3,500 minus 600 — still a solid week.

This is why the daily number matters much less than people think, and why the weekly weight trend is the instrument to watch. It is also why banking calories by eating almost nothing all day before dinner tends to backfire: arriving at a restaurant genuinely hungry costs you more than the calories you saved.

Ordering choices that move the number more than the estimate does

Since you cannot measure a restaurant meal, the leverage is in what arrives rather than in how you record it.

Grilled, tandoori, roasted, steamed cost far less than fried, creamed or buttered versions of the same protein. A tandoori chicken and a butter chicken start from the same bird and end up several hundred calories apart.

Order the bread you want and stop there. Bread and rice at a restaurant are usually unlimited in spirit if not in price, and they arrive already buttered.

Sauce on the side is not a diet cliché, it is the single most effective request, because the sauce is where the fat is.

Drinks are the invisible third of the meal. Two beers and a soft drink is 430 kcal that leaves no memory of having eaten.

None of this requires ordering sadly. It requires deciding before the menu arrives, which is a much easier decision than deciding after.

Build your own defaults from your own data

After a month or two of logging restaurant meals with generous estimates, you will have something better than any published table: a record of what your weight actually did in weeks with two meals out versus weeks with none.

If the trend matches your log, your defaults are fine. If the trend is worse than the log predicts, raise them by 20 per cent. That feedback loop is the same one described in how to track calories, and it is the only calibration that has ever been about you specifically.