Packaged food is the easy case. It has a label, a fixed serving, and a barcode.
Home cooking has none of those, which is why most calorie trackers quietly fall apart on the meals you eat most. A curry your mother made has no entry in any database, and the entries that look close were submitted by strangers describing a different kitchen.
Here is a method that works.
Why it is genuinely hard
Three things make home-cooked food resistant to database logging, and it is worth understanding them because the workarounds follow directly.
No standard portion. There is no packet telling you the serving size. One person's katori of dal is another's half-bowl.
Invisible fat. The same sabzi is a different food cooked in a teaspoon of oil versus three tablespoons of ghee. This is usually the single largest source of error, and it is invisible in any photo.
Composite by default. A home dinner is four or five things on one plate. Logging them separately is four or five interactions for one meal, which is exactly the friction that ends most people's tracking.
Log the components, not the dish
The instinct is to search for "mixed vegetable curry" and take whatever number appears. Resist it — you are inheriting a stranger's assumptions about oil and portion.
Instead, describe what is actually in front of you:
"One katori aloo sabzi, two rotis, half a katori dal, small bowl of curd"
Each component is something you can judge. The dish as a whole is not.
Get the cooking fat right, once
This is the highest-value thing on this page.
You do not need to measure every meal. You need to know roughly how your kitchen cooks, once, and then apply it consistently:
- Watch how much oil goes into the pan for a normal family dal or sabzi
- Divide by the number of servings it makes
- That is your per-serving fat, and it barely changes week to week
For most Indian households this lands somewhere between half a teaspoon and two teaspoons of oil per serving — a range of roughly 20 to 80 calories. Knowing which end you are at is worth more than any other single correction.
Ghee on rotis is the other one. Two rotis with ghee is roughly 80–90 calories more than two without, every single day.
Use your household's own units
Real kitchens do not measure in cups and ounces. They measure in katoris, bowls, ladles, spoons and pieces.
Use them. "One katori dal" is a real, stable unit to you in a way that "185 g of cooked lentils" never will be. Any tool worth using should accept it — and if you tell it once that your katori is a large one, it should stay told.
Build a reference for the ten meals you actually eat
Most households rotate through a fairly small set of dinners. Get those ten right once and you have covered most of your year.
Spend one week being deliberate — pay attention to the oil, note your portions — and after that logging is recall, not estimation. This is the point where tracking stops feeling like work.
For shared dishes, estimate your share
A bowl of something in the middle of the table is not one serving. Judge what proportion of it ended up on your plate — a third, a quarter — and log that.
Being slightly wrong here is fine. Being consistently slightly wrong is genuinely fine, because your weight trend over three weeks corrects for a fixed bias automatically. What breaks the method is guessing generously some days and honestly on others.
Rough starting points
Useful until you have your own numbers.
| Component | Typical serving | Calories |
|---|---|---|
| Dal, light tadka | 1 katori | 120–160 |
| Dal, generous ghee | 1 katori | 180–240 |
| Dry sabzi | 1 katori | 90–150 |
| Aloo sabzi | 1 katori | 150–220 |
| Home chicken curry | 1 katori | 220–320 |
| Rice, cooked | 1 cup | 180–210 |
| Roti, no ghee | 1 | 80–110 |
| Roti with ghee | 1 | 120–155 |
| Curd | 1 katori | 80–110 |
| Cooking oil | 1 tsp | ~40 |
A fuller version is in the Indian food calorie chart.
Accept the error, then be consistent
You will not get home-cooked food exactly right, and that is fine. Nobody does — not you, not a food database, and not a photo model that cannot see how much ghee went in.
What matters is that your estimate is stable. A consistent method that runs 10% low still tells you truthfully whether you are eating more or less than last week, and your weight trend will correct the offset within a fortnight.
Inconsistency is the real enemy: counting the oil some days and not others, logging lunch carefully and guessing dinner from memory. That produces noise you cannot correct for because you cannot see it.
How BetterCal handles this
You write the meal the way you would say it — "dal, two rotis and curd" — and it estimates each component against how that food is actually cooked and served where you live, rather than against a Western database average.
Correct anything that looks wrong once, and the correction sticks. Over a few weeks the estimates converge on your kitchen rather than an average of everyone else's.
Related: how many calories in a roti · Indian food calorie chart · Indian food calorie tracker
