Why Homemade Meals Are Hardest to Count

Your dinner changes weight while you cook it#

Homemade food is the hardest thing in your log to count, and the reason is not that you are careless with the oil. It is that a home-cooked meal is the only category where you perform the manufacturing step — and the manufacturing step changes the food's mass. Meat sheds water in the oven. Rice and pasta drink it up in the pot. Anything fried gains weight it did not have on the counter. So "100 grams of chicken" is not one quantity; it is two, and they are not interchangeable.

That single fact drives most homemade counting error, and it is far more consequential than it sounds because of a second thing almost nobody knows: the "cooked" entry your app resolves your dinner to may never have been cooked at all. It can be a raw laboratory measurement pushed through a conversion factor. You are estimating against an estimate — and unlike the stacked error layers elsewhere in tracking, this one you can genuinely fix, because you control the weighing.

What a cooking yield is, and why every database owes it#

The USDA has a name for how much of the food survives the pan. Its Nutrient Data Laboratory defines the term plainly: "Cooking yields describe changes in food weight due to moisture loss (e.g., evaporation or moisture drip), water absorption (e.g., boiling) or fat gains/losses during food preparation and cooking"¹.

Read that definition twice, because it contains all three homemade traps at once. Moisture loss — the roast. Water absorption — the rice. Fat gains — the frying pan. Three different directions, in one sentence, from the same mechanism.

The size is not trivial. When USDA researchers cooked four pairs of retail beef cuts under controlled conditions, ribeye roasts came out at 77 and 76 percent cooking yield, retaining 63 and 64 percent of their moisture, while the equivalent steaks yielded 85 and 83 percent and retained 72 percent³.

Cut	Cooking yield	Moisture retention
Ribeye bone-in lip-on, roasted	77%	63%
Ribeye boneless lip-on, roasted	76%	64%
Ribeye bone-in lip-on, grilled steak	85%	72%
Ribeye boneless lip-on, grilled steak	83%	72%

Data: Nguyen et al., USDA/ARS Nutrient Data Laboratory. This is a conference poster from USDA's own laboratory, supported by the Beef Checkoff — worth stating, though a weight measurement is about as hard to spin as food data gets.

Nearly a quarter of a roast's weight leaves in the oven. And notice that the same cut yields differently depending only on whether it was roasted or grilled — an eight-point spread produced by nothing but your choice of appliance. No database entry can know which one you did.

The cooked entry may itself be a calculation#

Here is the part that reframes the whole problem. Why does the USDA maintain yield tables at all? Because measured data for cooked food mostly does not exist.

The agency is explicit: "Composition data are needed for the nutrient value for both the uncooked and cooked forms of foods, but nutrient data for cooked foods are generally not available. Therefore, nutrient composition of a cooked food may be calculated from the uncooked food by applying cooking yield factors to these data" — and "most public and private sector databases apply cooking yields to nutrient values as part of the nutrient calculation process where analytical data for cooked foods are unavailable"¹.

The "roast chicken" in your app may never have been roasted. It can be a raw chicken, measured in a laboratory, multiplied by a factor for what roasting typically does — using a cooking method, time, and temperature that were not yours.

A companion table does the same job for micronutrients, with retention factors covering "16 vitamins, 8 minerals, and alcohol for approximately 290 foods," and the USDA notes that "most public and private sector databases use these retention factors to calculate nutrient values when analytical data for cooked foods are unavailable"².

Read that list closely and you find the distinction that matters for calorie counting: retention factors are for vitamins, minerals, and alcohol. Energy, protein, fat, and carbohydrate are not on it. For calories, the lever is the weight change, not nutrient destruction — cooking does not burn off the fat, it drives off the water and concentrates everything left. That is why a cooked-weight gram is a more calorie-dense gram than a raw one, and why swapping the two bases is not a rounding error.

Recipe aggregation adds its own layer on top. A review of food-composition databases lists recipe calculation principles alongside food description, energy calculation, and analytical methods as a standing source of divergence between databases⁴ — one reason two apps hand you different numbers for the same home-cooked plate.

Oil arrives from outside the recipe#

Every other ingredient in a home-cooked meal is something you chose and could, in principle, weigh. Cooking fat is different in two ways that compound.

First, it is the densest thing in the kitchen: 9 calories per gram against 4 for protein and carbohydrate⁵. Second — and this is the part specific to home cooking — it is added after the recipe is written, in an amount nobody records, by a hand tilting a bottle over a hot pan. The USDA's own definition already concedes the mechanism when it lists "fat gains" as a driver of cooking yield: food can leave the pan carrying fat that was never in the ingredient list.

I went looking for a solid figure on how much oil a given fried food actually absorbs and could not find one I was willing to print — the numbers in circulation trace to blogs and patents rather than to any primary source I could open. So take the honest version instead: the direction is certain and the magnitude is not, which is itself the argument for weighing the bottle rather than estimating the pan. Oil is the one ingredient where a 10-gram misjudgment costs 90 calories and leaves no visual trace at all — see how to track cooking oils and fats for the mechanics.

Then you divide the pot#

The final error is the one unique to cooking for more than one person, and it has no equivalent anywhere else in tracking. You made four servings. You ate "one quarter." You did not eat one quarter.

A pot of stew is the purest amorphous food there is — no unit structure, no fixed geometry, served by a ladle wielded by someone who is hungry. And the error is a multiplier rather than an addition: if your share was really 30 percent of the pot rather than 25, every calorie in the recipe is off by a fifth for you, including all the ones you carefully weighed at the start. Precision on the ingredients does not survive an imprecise division at the end.

Which points at the fix, and it is a change to your cooking rather than to your logging. Weigh the whole finished dish once, then weigh your plate; your share is the ratio, and it takes ten seconds. Better still, cook to a portion — bake in individual dishes, or divide the tray into visible quarters before anyone serves themselves. Turning one pot into four countable units is the same move that beats eyeballing everywhere else, and it works here for the same reason (estimating calories without a scale makes the general case).

Do those two things — weigh raw and match the basis, and divide by weight rather than by eye — and homemade food goes from the worst-counted category in your log to one of the better ones. It has an advantage nothing else has: you were standing there when it was made. A nutrition label is a stranger's declaration and a restaurant plate is a stranger's improvisation, but you personally know every gram that went into this one. What you lose is only the bookkeeping — which is worth doing roughly and often rather than perfectly and never (how to count calories makes that case), and worth reading as a band rather than a verdict (why calorie counts are ranges).

FAQ#

Should I weigh food raw or cooked?#

Either, as long as the entry matches — that is the entire trick. Weigh raw and log a raw entry, or weigh cooked and log a cooked entry. Mixing them imports the whole cooking-yield gap: roasted ribeye came out at 76 to 77 percent of its raw weight³, and boiled grains move the other way by more. Raw is usually easier, because it happens before the food is in a pan and because it is what recipes specify.

Does cooking change the calories in food?#

It changes the calories per gram, mostly by changing the grams. Cooking yields track "changes in food weight due to moisture loss..., water absorption..., or fat gains/losses"¹. Water carries no energy, so driving it off concentrates what remains, and a gram of cooked meat is denser than a gram of raw. The energy did not appear — the water left.

Do I have to log oil that stays in the pan?#

Log what left the pan with the food, which is the awkward part — some of the oil goes onto the plate and some stays behind, and no household method separates them. Given fat is 9 calories per gram⁵, the safer error is to log what you poured. For anything roasted, sautéed, or dressed, nearly all of it arrives on the plate anyway.

Sources#

This article was researched and drafted with AI assistance and reviewed for accuracy by the BurnWeek team. It is general information, not medical advice. How we research and correct our articles →