The number on the back of a packet is an estimate, not a measurement of the food in your hand. It comes from an average of test samples, worked out under rules that change depending on where the product is sold, and it says nothing about how much of that energy your gut will actually absorb. Usually the gap is small enough to ignore. Sometimes it is not.
Where the number comes from
The classic method is brutally simple. You dry a sample of food, seal it in a metal chamber with oxygen, set fire to it, and measure how much the surrounding water heats up. That is bomb calorimetry, and it gives you the total chemical energy in the sample.
Your digestive system is not a sealed metal chamber. Some of what burns in a calorimeter passes straight through you. Fibre is the obvious case. Protein costs you energy to break down, so what you keep is less than what went in. Labels handle all of this with fixed conversion factors rather than by testing what a real body does with the food, because testing real bodies is slow and expensive.
Two packets of the same thing can differ
Labels are built from averages. One batch of oats is not identical to the next, a chicken breast is not a standard object, and the figure printed on the box is a representative value with a legal tolerance around it.
How much tolerance depends on where you are. Countries differ on how far a product may deviate from its stated figure, on whether fibre counts toward the total, and on how the numbers get rounded. [ADD SPECIFIC: the permitted deviation in your main market, e.g. the FDA and EU tolerances, with a source]
Rounding does more damage than people expect on small items. A snack listed at 90 calories can sit anywhere inside a band around that figure, and if you eat four of them, the error is applied four times.
Cooking and processing change the maths
Finely milled flour gives up its energy more easily than whole grains. Cooked food is generally easier to extract energy from than raw food. A nut chewed to paste gives you more than one swallowed in pieces.
None of that shows up on the label. Two products with the same printed count can leave you with meaningfully different amounts of usable energy, which is part of why whole foods behave differently from heavily processed versions of the same ingredients.
Serving sizes cause more trouble than the science
Here is the uncomfortable part. The label is usually closer to right than your idea of a portion is.
A stated serving is often smaller than what anyone actually eats, so a packet that reads 200 calories per serving quietly holds two and a half of them. Breakfast cereal is the classic offender. So is anything that comes in a bag you eat straight out of. Give people a bigger container and they take more from it without noticing, and no label can protect you from that. If you want to fix one thing this week, fix this one. Our guide to estimating portions covers how to judge sizes without weighing everything you own.
What to do about it
Weigh the foods you eat most often, at least for a couple of weeks. Not forever, and not everything. You are calibrating your eye, and the handful of foods that dominate your week are the only ones where precision pays for itself.
Then stop treating your daily total as an exact figure and start treating it as a consistent one. If you log the same way every day, a few percent of error becomes a fixed offset, and the trend on the scale still tells you the truth. Consistency beats accuracy here, because a consistent number is something you can act on.
And if your weight is not moving the way the arithmetic says it should, the label is rarely the main suspect. Check how you are tracking first, then check whether your maintenance calories are still where you think they are.