Every nutrition line on this site is one division problem plus a set of decisions about what goes into it. The division is the same everywhere. The decisions are where recipe sites quietly differ from each other, and almost nobody publishes theirs.
This is ours, including the parts that are still weak.
What a serving is here
Total the recipe, divide by the number of servings, print the answer. Every recipe on the site states the batch size, and all 56 of them print six numbers rather than one: calories, fat, carbs, fiber, sugar and protein. Publishing fat and sugar alongside calories is the part that keeps the number honest, because leaving either one off is the oldest way to make a dessert look better than it is.
The servings figure carries almost all the weight in that calculation, which is why it gets its own article. Our black bean brownies come out of an 8x8 pan cut into sixteen, so a serving is a two-inch square at 110 calories. Cut the identical pan into twenty-four and the same ingredients print 73 calories a brownie. Nothing about the recipe changed. We would just be handing you a smaller brownie and a better headline.
Cups are the weakest link, and they are ours
A cup is a volume, and flour does not have a fixed density. Spooned gently into the cup and levelled, a cup of all-purpose flour lands near the 120 gram figure most US baking references use. Scooped straight out of the bag, the same cup can come out at 140 or 150 grams, because scooping packs it. That is a swing of roughly a quarter in the largest dry ingredient in a cake, and it moves the calorie count of the whole pan before anything else has a chance to.
Nut butter is worse in a smaller way. A heaped tablespoon against a level one is 30 or 40 calories, and nobody levels peanut butter.
Weighing removes all of it, and weighing is where we are going. Ten of our 56 recipes already print a gram weight in brackets beside the cup or spoon measure, and one more carries a weight on a packaged ingredient, like the carrot cake loaf at 1 cup (120g) whole wheat flour, or the edible cookie dough at 2 scoops (about 60g) protein powder. The other forty-six do not yet. If you want the tightest possible match to our number today, weigh your flour at 120 grams a cup and your nut butter on the scale rather than the spoon.
What the number covers, and what sits outside it
Anything in a photograph that is not in the ingredient list is not in the count. The yogurt beside the slice, the berries scattered on the board, a dusting of cocoa for the shot. If it is not written down, it was not counted.
Optional ingredients are the harder case and we would rather be straight about it than tidy. The rule we hold to is that the number covers the recipe as written, without the optional add-in. Nothing enforces that rule except us reading the ingredient list, and nothing on the recipe card tells you which side of the line a particular item landed on. Two tablespoons of chocolate chips split across two servings is about 35 calories a bowl, which is exactly the size of thing that can sit either side of a rounding decision without looking wrong. If an optional item matters to your day, add it yourself and take the small risk of counting it twice.
The fix for that is a plain line on every recipe stating what the number includes. It is not written yet. It should be.
How sweeteners are handled
This is where a sugar-free dessert stops behaving like ordinary food on a calculator.
For labelling purposes, sugar carries 4 calories a gram. Allulose is treated at 0.4 and erythritol at 0.2, and both of them still count inside total carbohydrate. So a recipe sweetened with either one prints a carbohydrate figure that looks high next to its calorie figure, and the gap is the sweetener sitting in one line and not the other. The label article covers what that does to a packaged product.
If you rebuild one of these recipes in a tracker and get a bigger number than the page shows, this is the most likely reason. Plenty of database entries treat allulose as ordinary sugar at the full 4 calories a gram. That is a disagreement about convention, not about the recipe.
The check we run on every printed number
Protein and carbohydrate run about 4 calories a gram, fat about 9. Add those up for any recipe and you get a ceiling: the printed calories should land at or slightly under that sum, because the sweeteners are counted in the carbohydrate line while contributing almost nothing to the calorie line. A recipe printing calories above its own macros cannot be explained by sweetener, and is simply an error.
We run all of them through that arithmetic. Every recipe in the corpus lands at or below its own sum right now, and the median sits about 4 percent under it, which is the direction the sweeteners predict.
That was not always true. Cookie dough ice cream for one used to print 250 calories where its own 18g of protein, 20g of carbs and 8g of fat added up to a ceiling of 224, a number the arithmetic alone said was impossible. We have since rebuilt it from the ingredient list, and it now prints 315 calories against a ceiling of 344, comfortably inside the line.
That is exactly what this check is good for, and exactly where it runs out. It can only catch a number that is impossible. It cannot catch a number that is wrong and still adds up cleanly, and we have since found one of those too. Dark chocolate almond bark printed 130 calories and 10 grams of fat against 3g of protein and 9g of carbs, a ceiling of 138, and it passed this exact arithmetic the entire time it was wrong. The real numbers are 195 calories and 15 grams of fat, found only once we recomputed the recipe from its own ingredient weights instead of rechecking the printed line against itself. Passing this check is a floor, not a certificate.
Why your number will not match ours exactly
Brands vary. Peanut butter runs from about 90 to 100 calories a tablespoon depending on the jar, and two protein powders can sit 30 calories apart per scoop while the scoops themselves are different sizes.
Some of the batter stays in the bowl. Ours does too, and the arithmetic quietly assumes it did not.
And the piece you actually cut is the biggest one. Sixteen brownies coming out of the pan as fourteen makes every one of them about 14 percent bigger than the number claims, which is a larger error than anything else on this page.
Where to treat all of this as an estimate
Every number here is an estimate. None of it has been evaluated by the FDA, and we are two people who develop dessert recipes rather than dietitians. If a figure matters to you for a medical reason, the number worth trusting is the one you calculate from the brands in your own kitchen, measured against whatever targets your care team gave you.
Two things are queued for this site next: a gram weight beside every cup measure, and a line on every recipe saying exactly what its number covers.
