How do you count macros from a food label?

Macros are the grams of protein, carbohydrate and fat on the label, and 4, 4 and 9 kcal per gram turn them into calories: for Nutella that gives 533 kcal per 100 g against the 539 on the label, a gap of 1.1%.

Six products per 100 g: label calories against 4 × protein + 4 × carbohydrate + 9 × fat. Gap = (label − estimate) ÷ label. Source: food database, build of 16 November 2026.
ProductProteinCarbsFatLabel4 / 4 / 9Gap
Nutella6.3 g57.5 g30.9 g539 kcal533.3 kcal+1.1%
Pringles Sour Cream & Onion6.0 g56.0 g30.0 g525 kcal518.0 kcal+1.3%
Coca-Cola Original Taste0.0 g10.6 g0.0 g42 kcal42.4 kcal−1.0%
Lindt Excellence 85% Cacao12.5 g22.0 g46.0 g584 kcal552.0 kcal+5.5%
Barilla Pesto alla Genovese4.7 g11.0 g47.0 g492 kcal485.8 kcal+1.3%
Chocapic (Nestlé)8.5 g73.8 g4.4 g384 kcal368.8 kcal+4.0%

A positive gap means the label is higher than the sum of the three macros. The six products were picked by hand: a spread, chips, a cola, dark chocolate, a pesto and a breakfast cereal. All six were compared with the live Open Food Facts records.

What macros are

"Macros" is short for macronutrients: protein, carbohydrate and fat. They are the nutrients a label gives in grams that also supply the energy in its calorie figure. Alcohol supplies energy too (7.0 kcal per gram in the same paper), but it is not one of the three macros.

The FAO's paper on food energy gives the general factors: 4.0 kcal per gram for protein, 9.0 for fat and 4.0 for carbohydrate, with alcohol at 7.0. It adds that, as originally described by Atwater, carbohydrate is determined by difference and so includes fiber.

The energy factors on a label

Regulation (EU) No 1169/2011 says the energy value on a label "shall be calculated using the conversion factors" in its Annex XIV. That list goes beyond three numbers.

Energy conversion factors, Regulation (EU) No 1169/2011, Annex XIV
Componentkcal per gram
Protein4
Carbohydrate (except polyols)4
Fat9
Alcohol (ethanol)7
Organic acid3
Salatrims6
Polyols2.4
Fiber2
Erythritol0

In the United States, 21 CFR 101.9(c)(1) allows calories to be worked out with specific Atwater factors, or with the general factors "of 4, 4, and 9 calories per gram for protein, total carbohydrate, and total fat". It also lists a separate factor for each sugar alcohol, for example 2.4 for xylitol, 2.1 for maltitol and 0 for erythritol.

The method, step by step

Labels give protein, carbohydrate and fat per 100 g, or per serving, or both. To work out what a portion contains, scale each value:

grams of a macro eaten = grams of food eaten ÷ 100 × the per-100 g value

Then multiply each macro by its factor and add them up. Here is Nutella, with 15 g eaten (the serving on the jar):

  • Protein: 15 ÷ 100 × 6.3 = 0.945 g, and 0.945 × 4 = 3.78 kcal
  • Carbohydrate: 15 ÷ 100 × 57.5 = 8.625 g, and 8.625 × 4 = 34.5 kcal
  • Fat: 15 ÷ 100 × 30.9 = 4.635 g, and 4.635 × 9 = 41.715 kcal
  • Total: 3.78 + 34.5 + 41.715 = 80.0 kcal

The label's 539 kcal per 100 g scaled to 15 g is 539 × 15 ÷ 100 = 80.85 kcal, so the two routes differ by 0.85 kcal. The post on per serving and per 100 g covers the scaling step on its own.

The share of energy from each macro is that macro's kcal divided by the total. For the Nutella portion: 3.78 ÷ 80.0 is 4.7% from protein, 34.5 ÷ 80.0 is 43.1% from carbohydrate and 41.715 ÷ 80.0 is 52.1% from fat. The shares do not depend on the portion size, because every term scales the same way.

Share of the 4 / 4 / 9 energy that comes from each macro
ProductProteinCarbsFat
Nutella4.7%43.1%52.1%
Pringles Sour Cream & Onion4.6%43.2%52.1%
Coca-Cola Original Taste0.0%100.0%0.0%
Lindt Excellence 85% Cacao9.1%15.9%75.0%
Barilla Pesto alla Genovese3.9%9.1%87.1%
Chocapic (Nestlé)9.2%80.0%10.7%

How Smidge sets its own protein, carbohydrate and fat split is on the method page. This post only describes how to read the numbers on a pack.

How close the label and the estimate are

Across the database, 2,675,583 product groups list calories above zero and all three macros. A group is one name and brand, and it can cover several barcodes. For each, the label's per-100 g calories can be compared with 4 × protein + 4 × carbohydrate + 9 × fat. The median gap is +0.6%, so the label is usually a little higher than the three-factor sum.

Gap between label calories and the 4 / 4 / 9 estimate, groups with complete macros (n = 2,675,583)
GapGroupsShare
Within 2%1,372,23351.3%
Over 2% to 5%675,87325.3%
Over 5% to 10%311,96311.7%
Over 10% to 20%165,5286.2%
Over 20%149,9865.6%

In 3.1% of all groups the label is more than 20% above the estimate, and in 2.5% it is more than 20% below.

Why the gaps happen

Small numbers. A fixed difference weighs more on a low-calorie food. Among groups with a label under 50 kcal per 100 g (287,671 groups), 23.9% sit more than 20% from the estimate. From 50 to 199 kcal it is 4.7%, from 200 to 399 kcal 3.3%, and at 400 kcal or more 2.0%. Restricted to 50 kcal and above, 3.4% differ by more than 20%.

Fiber. The EU text separates carbohydrate from fiber. Annex I, point 7 defines "carbohydrate" as "any carbohydrate which is metabolised by humans" and says it includes polyols. Point 12 defines "fibre" on its own, as carbohydrate polymers "neither digested nor absorbed in the human small intestine". Annex XV lists fiber as a separate line, not as an "of which" line under carbohydrate, and Annex XIV gives it its own factor, 2 kcal per gram. Wasa fibres, a seventh product in the saved query, shows the effect most clearly: protein 12 g, carbohydrate 47 g and fat 5 g give 281 kcal, against a label of 333, a gap of +15.6%. Adding 26 g of fiber at 2 kcal per gram brings the estimate to 4 × 12 + 4 × 47 + 2 × 26 + 9 × 5 = 333 kcal. Chocapic goes from 368.8 to 384.2 the same way, with 7.7 g of fiber. Lindt's entry lists fiber as 0.0 g, so that step adds nothing and the 5.5% gap stays.

The US text gives more than one route. 21 CFR 101.9(c)(6) has total carbohydrate calculated by subtraction of protein, fat, moisture and ash from the total weight, so it includes fiber, and dietary fiber is listed indented beneath it. For calories, paragraph (c)(1)(i) allows several methods, including (A) specific Atwater factors, (B) the general 4, 4 and 9 factors on total carbohydrate and (C) the same factors on total carbohydrate "less the amount of non-digestible carbohydrates and sugar alcohols", with 2 kcal per gram for soluble non-digestible carbohydrates and the listed factor for each sugar alcohol. Option (B) applied to a US label already counts fiber at 4 kcal per gram. Option (C) has the same shape as the "inside" treatment in the table below, except that its 2 kcal per gram applies only to soluble fiber, which the database does not separate out.

The database can show how the gaps are spread, but not which option a maker used. Among groups with fiber listed, the median gap in groups not tagged for the US is +1.3% on plain 4 / 4 / 9 and 0.0% with fiber added at 2 kcal per gram. In US-tagged groups, plain 4 / 4 / 9 gives −0.4%, and adding fiber on top gives −2.1%.

Median gap, and share of groups more than 20% from the estimate, under three treatments of fiber. "Inside" takes carbohydrate to include fiber (4 × (carbs − fiber) + 2 × fiber); "on top" takes it to exclude fiber (4 × carbs + 2 × fiber).
GroupsnMedian 4/4/9Median insideMedian on topOver 20% 4/4/9Over 20% insideOver 20% on top
Fiber listed1,455,363+0.8%+2.2%−0.2%5.2%7.6%5.9%
Fiber listed, US-tagged572,040−0.4%+1.0%−2.1%5.1%5.5%7.1%
Fiber listed, not US-tagged883,323+1.3%+3.0%0.0%5.3%8.9%5.2%
All groups, adjusted where fiber is listed2,675,583+0.6%+1.2%0.0%5.6%6.9%6.0%

In each of the two subsets, US-tagged and not, one treatment has the median closest to zero: plain 4 / 4 / 9 for US-tagged groups, and fiber on top for the rest. A median across many makers does not show which option any one maker used. The share beyond 20% barely moves (5.3% to 5.2% for non-US groups with fiber on top) and rises under the other treatments, so on this test fiber shifts the middle of the distribution and leaves its tail.

Polyols and alcohol. A sweetener such as xylitol counts 2.4 kcal per gram in the EU and carries its own factor in the US list. Alcohol counts 7. The database has no polyol or alcohol field, so these cases cannot be separated out. They fall into the gap.

Rounding. A label rounds the numbers it prints. In the US, 21 CFR 101.9(c)(1) has calories given "to the nearest 5-calorie increment up to and including 50 calories, and 10-calorie increment above 50 calories", per serving. Grams are rounded too (US total carbohydrate to the nearest gram, 21 CFR 101.9(c)(6)), so a calorie figure worked from printed grams can land a few kcal away. On a 20 kcal portion, 2 kcal is already 10%.

Specific factors. The US text also allows specific Atwater factors, and the FAO describes this system next to the general one. Where a maker used them, the label does not follow a plain 4 / 4 / 9 sum.

Estimated and mistyped values. Among groups whose nutrient source is marked as an estimate (22,400 with complete macros), 16.6% are more than 20% from the sum, against 5.5% for Open Food Facts groups and 5.4% for USDA groups. A mistyped number in any of the four fields would also show as a gap, and the database cannot say which groups hold one.

Where this data falls short

Coverage. The comparison covers 2,675,583 of 2,784,193 product groups (96.1%): those with calories above zero and all three macros. Fiber is listed for 1,455,363 of them (54.4%), so the fiber results describe a little over half. A missing fiber value is left out and is never counted as zero.

Contributor-entered labels. Open Food Facts is edited by volunteers and USDA Branded Foods comes from manufacturers' submissions. Both can hold typos. The six products in the table and Wasa fibres in the text were compared with the live Open Food Facts records on 17 November 2026, and every protein, carbohydrate, fat, fiber and calorie value matched the database.

Category tags. Each group carries at most three Open Food Facts category tags, and about half have none. These numbers do not filter by category, so the limit does not change them.

One snapshot. Everything here is the database built on 16 November 2026, from Open Food Facts data through 15 November 2026 and USDA release 2026-10-29. Makers change recipes and labels, so the live record for a product may differ from the entry used here.

Grouping by name and brand. A group can mix sizes, flavors and countries. Each group is represented by the entry of one barcode, so the comparison counts products once and does not weight by sales. The US tag means a product is tagged for the US in Open Food Facts or comes from USDA. It is not a record of which label rules the maker followed, so the US and non-US medians above are consistent with the two rules but do not prove which one applied to each label.

Food data from USDA FoodData Central (public domain) and Open Food Facts (Open Database License). The database is described on the food database page, and any product can be looked up in the food search.

Sources

Know the number. Log it in a sentence.
Smidge for iPhone and Android.
QR code that opens smidge.fit/get

Scan with your phone camera to open Smidge in the App Store or Google Play.