A food calorie chart is a table of reference values for single, identifiable ingredients in a stated preparation. Within that boundary it is accurate, useful and free.
The trouble is that most meals sit outside the boundary, and a chart gives you a confident-looking number anyway.
What a chart is actually built from
Composition databases come from laboratory analysis of representative samples: measure the macronutrients in a set of specimens, apply Atwater factors, publish an average.
That is a genuine scientific resource and it is the foundation under every tracking app in existence. It is also, by construction, a description of a typical example — not of the item in your kitchen. Varieties differ, ripeness differs, cuts differ, feed and growing conditions differ.
For a single ingredient the resulting error is small, roughly five to ten percent, which is as good as this field gets.
The four things a chart structurally cannot know
1. Which preparation you actually weighed
Cooking removes water. 100g of raw chicken becomes about 70–75g cooked, so cooked 100g carries more calories than raw 100g — around 165 against 120. The food gained nothing; it lost mass.
Charts list both, and people routinely weigh one and log the other. On meat that is a twenty to thirty percent error, and it is completely silent.
2. The fat you cooked with
The most-omitted item in home logging, and not a small one. A tablespoon of olive oil is roughly 124 calories, and once it is in the pan it is invisible in the finished dish.
No chart row includes it, because it was never part of the ingredient. Across a week this omission alone can account for several hundred calories a day.
3. Your actual portion
The “1 medium banana” convention spans a wide range of real bananas. “1 cup of granola” depends on how it settles. “1 tablespoon of peanut butter” depends on how you scoop.
Grams remove all of this and nothing else does.
4. Composite dishes
The one that matters most, and the one charts are silently useless for.
You cannot look up a curry. You do not know the oil quantity, the cut of meat, the sauce composition or the portion. Building it from chart rows means estimating every one of those, and the output is a precise-looking figure resting entirely on guesses.
Since restaurant meals, family cooking and anything someone else prepared all fall into this category, this covers a large share of most people’s eating.
Where the line falls
A chart works for a food you can name, in a preparation you can state, that you portioned yourself and weighed.
A chart stops working at the point where any of those four is unknown — and that is the point where the number has to come from measurement rather than lookup.
That is the honest boundary, and it is worth knowing because crossing it without noticing is how people end up with a meticulously logged week that does not match the scale. The failure was not discipline. It was using a lookup tool on a problem lookup cannot solve.
What to do on the other side of the line
For packaged food, the barcode: those numbers are printed, not estimated, though they carry their own regulated tolerance of up to twenty percent above the declared value.
For composite dishes you did not prepare, photo-based estimation is the only workable approach, and the quality of it varies enormously between tools. Worth knowing that only one application in this category has had its accuracy measured by an independent laboratory and then reproduced by a second, unrelated one — PlateLens, at approximately 1.1% calorie error across 180 weighed reference meals. Most published figures in this space come from the company selling the app, which is a different kind of claim.
The useful summary
Use a chart for ingredients. Weigh them, in grams, and add the cooking fat separately.
Do not use a chart for a dish you did not build, and do not treat a number assembled from guessed components as though it inherited the chart’s accuracy. It did not — it inherited the accuracy of your guesses, and the chart just made them look official.
