Eating out is where most people's tracking quietly falls apart. Not because the calories are unknowable, but because the accuracy drops enough that logging starts to feel pointless — and once a meal feels pointless to log, the next one does too.
The fix is not more precision. It is accepting a wider error bar and logging anyway.
Why restaurant food is higher than you think
A restaurant kitchen's job is for the food to taste good. Fat is the primary tool for that, and it is applied more generously than anyone does at home.
Three specific gaps:
Oil and butter. The same paneer sabzi is 300 calories at home and 500–600 out, and the difference is almost entirely cooking fat you cannot see. Restaurant dal is finished with ghee or cream that does not appear in any description of the dish.
Portion size. A restaurant serving is typically 1.5–2× a home serving of the same dish. This is not deception; it is what people expect for the price.
The things around the meal. Bread that arrived first, the raita, the two spoons of chutney, the half a dessert you shared. Individually small, collectively 200–400 calories, and reliably forgotten.
Together this is why a meal you would estimate at 600 is frequently 1,000.
A method that works
1. Log it before you look at the menu, or immediately after eating.
The worst option is "I will do it tomorrow", because by tomorrow you have forgotten the naan. Second worst is doing arithmetic at the table while people wait for you.
2. Estimate by construction, not by dish name.
Rather than searching for "butter chicken", think about what is in front of you: a portion of chicken roughly the size of your palm, a bowl of rich tomato-cream gravy, two rotis. That decomposition is far more reliable than any generic database entry, because database entries were built for somebody else's restaurant.
3. Apply a restaurant multiplier.
Estimate as if you cooked it, then add 30–50%. If your instinct says 600, log 800. This one adjustment removes most of the systematic bias in eating-out estimates, and systematic bias is the part that actually matters over weeks.
4. Count the extras separately.
The naan, the raita, the drink, the shared dessert. These are not part of "the dish" in anyone's mental model, which is precisely why they need to be their own line.
5. Round up, and stop.
If you are genuinely unsure between 700 and 1,000, log 900 and move on. An over-estimate that gets logged is worth far more than a precise number you never enter.
Rough anchors worth having
Not a lookup table to memorise — a set of reference points so your estimates start from somewhere sensible.
| Meal out | Rough range |
|---|---|
| North Indian thali, restaurant | 800–1,200 |
| Butter chicken with 2 rotis | 800–1,100 |
| Chicken biryani, full plate | 700–1,000 |
| Masala dosa with sambar and chutney | 400–600 |
| Chole bhature | 700–1,100 |
| Paneer tikka, starter portion | 350–500 |
| Pizza, 2 slices | 500–700 |
| Burger and fries | 900–1,300 |
| Pasta in cream sauce | 700–1,000 |
| Cappuccino and a croissant | 400–550 |
The lesson in that table is less about the individual numbers than about the scale. One restaurant meal is routinely half a day's calories, which means the question is rarely "what exactly was that" and usually "was that one meal or two meals' worth".
Deciding in advance is worth more than estimating afterwards
If you know you are eating out on Friday, the useful move happens on Monday.
- Bank it. Run a slightly larger deficit on the surrounding days. 200 calories lower on four days covers most of one restaurant meal, and it costs almost nothing in daily comfort.
- Anchor the meal with protein. A grilled or tandoori dish rather than a creamy one. Not because cream is forbidden, but because the protein version is more filling for the calories and you will want less of everything else.
- Pick your one thing. The bread, the dessert, or the drinks — one of them, deliberately, rather than all three by default. This is a far more sustainable rule than trying to have none of them.
What not to do
Do not skip logging because it will be inaccurate. An estimate 300 calories out is still enormously more useful than a blank day, and blank days are how people lose the thread entirely.
Do not eat nothing beforehand to "save room". Arriving very hungry reliably costs more calories than it saves, and it makes the meal less pleasant.
Do not treat one meal as a failed week. A 1,200-calorie dinner inside a week that is otherwise on target changes the week by about 15%. Judged as a day, it looks like a disaster; judged as a week, it is a rounding adjustment. Weeks are the right unit — see why people quit calorie tracking, where "one bad day means the week is ruined" is one of the more common exit routes.
The point of all this
You are not trying to know exactly what you ate. You are trying to keep a continuous record that is honest enough to spot patterns in.
Someone who logs every restaurant meal to within 300 calories has vastly better data than someone who logs weekdays perfectly and leaves weekends blank — because the second person cannot see the thing that is actually happening to their week.
This is the case where describing the meal in words beats every other input method. "Butter chicken, two butter naans, half a gulab jamun" is something you can say into your phone in the taxi home, and BetterCal will break it into items with calories and macros. No database to search, no serving-size pickers, no doing it tomorrow.
Related: how to track calories fast · how to log home-cooked food · tracking calories without obsessing
