Editorial analysis · US manufacturer labels checked November 17, 2026
There is a satisfying kind of food-label detective work that starts with a calculator and ends with an accusation. Multiply protein by four, carbohydrate by four and fat by nine. Compare the result with the calories on the wrapper. If the numbers disagree, someone must be hiding something.
The calculation is real. The accusation often gets ahead of it.
A nutrition label is not a spreadsheet that has promised to balance using its displayed, rounded numbers. Treating it that way can turn an attempt at careful tracking into a new error. The better question is which quantities and calculation method the numbers actually represent.
We applied the familiar 4-4-9 calculation to four current US manufacturer panels and kept the discrepancies visible. The products were selected to illustrate different label situations, not to represent a market or rank their accuracy. This is an arithmetic audit of published information. We did not test the foods in a laboratory or inspect manufacturers' unrounded formulation records.
See the four-label comparison · Try the rounding check · Download the records
The same shortcut misses in both directions
Our calculation is 4 × protein grams + 4 × total carbohydrate grams + 9 × total fat grams. “Difference” below means that calculation minus the manufacturer's displayed calories. A positive difference does not mean we found undeclared calories.
| Exact product and serving | Protein / carbohydrate / fat | Label calories | 4-4-9 result | Difference |
|---|---|---|---|---|
| Quest Chocolate Chip Cookie Dough, 1 bar (60g) | 21g / 22g / 9g | 190 | 253 | +63 |
| Jif Creamy peanut butter, 2 tbsp (33g) | 7g / 8g / 16g | 190 | 204 | +14 |
| Nuun Sport Strawberry Lemonade, 1 tablet (5g) | 0g / 4g / 0g | 15 | 16 | +1 |
| Liquid I.V. Hydration Multiplier Lemon Lime, 1 stick (16g) | 0g / 11g / 0g | 50 | 44 | -6 |
The last two rows are the original Nuun Sport tablet and original Liquid I.V. Hydration Multiplier, not caffeinated, sugar-free or other product lines. Water contributes no additional macros to this calculation. The displayed values come from the selected US pages, not a blend of retailer listings and other countries' labels.
The signs matter. The shortcut produces more calories than the panel for three products and fewer for one. That is a poor foundation for a rule that automatically “corrects” every label upward. It is also why we have not converted the differences into an accuracy league table. Agreement with a shortcut is not validation of a food.
Total carbohydrate is doing more work than the shortcut admits
Quest's panel lists 12g of dietary fiber and 6g of sugar alcohols within its 22g total carbohydrate. Its ingredient list includes polydextrose and erythritol. Multiplying every one of those 22 grams by four makes an assumption about energy that the carbohydrate total alone cannot justify.
The US labeling regulation permits several calorie-calculation methods, including specific food factors. One method separates nondigestible carbohydrate and sugar alcohols from the general carbohydrate factor. The listed sugar-alcohol factors differ: erythritol is 0 kcal/g; sorbitol is 2.6 kcal/g. A single subtraction rule cannot describe both.
That does not give us permission to reverse-engineer the Quest formulation. We do not know the unrounded quantities, the breakdown of every relevant ingredient or the precise calculation the manufacturer used. The 63-calorie gap is a measured difference between two published-number calculations, not a finding that the bar contains 63 hidden calories.
The opposite shortcut fails too. Remove all 12g fiber and all 6g sugar alcohol from the displayed carbohydrate, assign the remaining 4g four calories each, and add the protein and fat. The result is 181 kcal. That number is also not the label's 190. Assigning every excluded gram zero energy has introduced a different assumption.
We show both results because they expose the temptation: choose whichever formula makes the answer look tidy, then forget what was assumed. A useful food record should retain the original calorie value and the original nutrient fields, with their source. It should not quietly rewrite one to make an equation balance.
Rounding matters, but it is not a universal explanation
The same US rule says calorie factors apply to quantities before rounding. Displayed calories above 50 use 10-calorie increments; values up to 50 use 5-calorie increments, with a separate allowance for values below 5. The gram fields have their own rounding rules. See the source rules.
Here is a deliberately limited check. Pretend Quest's displayed 21g protein, 22g carbohydrate and 9g fat each stand for a value within half a gram. Under the all-carbohydrate-at-four assumption, the lowest result would be 4 × 20.5 + 4 × 21.5 + 9 × 8.5 = 244.5 kcal. The upper edge would be 261.5 kcal. Neither approaches the neighborhood of 190.
This shows that ordinary rounding of those displayed fields cannot, by itself, reconcile this particular 4-4-9 calculation. It does not show that the label is wrong. The exercise deliberately holds the simplified energy factors fixed, which is exactly the assumption that needs examining. It is not a legal tolerance test or a measure of actual food composition.
For the Nuun row, the shortcut differs by only one calorie. That is compatible with the general idea that rounded displays need not reproduce another rounded display. We still cannot certify the manufacturer's method from that agreement. Small residuals and large residuals both need context.
What to do when the app disagrees with the package
Start with the food's identity. Does the entry match the flavor, country, formulation and serving weight? Is it using total carbohydrate or a separately defined net-carbohydrate value? Has it confused one package with one serving? Our US/UK label comparison shows why matching only the brand name is insufficient.
Then find out whether the app is displaying an imported calorie field or reconstructing energy from macros. Those are different operations. A screenshot containing both a calorie number and gram values does not tell you which operation happened behind it. If the app offers no explanation, preserve the label and ask its support team rather than inventing a reconciliation.
Our editorial recommendation for a matching packaged-food entry is to retain the manufacturer's displayed values together. Flag a discrepancy for investigation rather than replacing calories with an undocumented estimate. This is a recordkeeping principle, not a claim that every package is error-free or a prescription for how much someone should eat.
For a recipe, the more consequential mistake may be the portion denominator. A correct ingredient total spread across the wrong cooked yield creates a different problem from label rounding. Keep that investigation separate; our cooked-weight worksheet walks through it.
Keep the discrepancy inspectable
Download the four-label CSV, methods and source JSON and blank audit worksheet. The JSON also preserves the rounding-only example and the separate 181-kcal subtraction example. Neither is a replacement nutrient declaration.
These records describe the manufacturer pages checked on the stated date. Formulations and online panels can change, and the package in your hand may differ. We did not establish exact metabolizable energy, absorption, manufacturing compliance or clinical suitability. Missing sugar-alcohol declarations in the dataset remain unknown, not zero.
Vora permits reuse of our original calculations with attribution and these limitations; manufacturer artwork is excluded. Corrections can be sent to support@askvora.com.
Publisher disclosure: we make Vora. Our interest is in making food records easier to understand alongside workouts and wearable context. If conversational logging fits your routine, Vora's text logging guide explains the supported setup. The calculations and downloads above are available without an account.