CaloriMateDifferent data types answer different questions
| Record type | What it can represent | What to inspect |
|---|---|---|
| Analytically derived food data | Measured samples of a defined food | Sample description, preparation and variability |
| Manufacturer label | Declared values for a particular product | Market, serving, formulation and date |
| Recipe calculation | Ingredients allocated across a finished batch | References, edible quantities and yield |
| User or AI estimate | A proposed record based on reported information | Assumptions, edits and validation status |
USDA FoodData Central separates data types such as Foundation Foods, Branded Foods and dietary-survey foods. Their presence in the same search system does not make their origins identical.
Match the food before comparing decimals
Check species or variety, preparation, drained state and edible basis. A raw ingredient, a boiled version and a recipe with added oil can legitimately have different energy densities. A brand name can also refer to different formulations in different countries.
If two entries disagree, first ask whether they describe the same food and serving. A difference between per-serving and per-100-g values is not a contradiction. Neither is the difference between a dry grain and its water-rich cooked preparation.
Unknown does not mean zero
A branded label may provide energy, macros and a limited set of nutrients. A database can faithfully reproduce that label while leaving other nutrients unreported. An empty field should not be silently interpreted as an analytical zero.
This matters when adding daily micronutrients. If some foods have incomplete data, the total may be a partial sum. More decimal places will not fill the missing fields. Prefer a display that identifies data coverage and lets you inspect the underlying source.
Calculated data need a visible chain
For a recipe estimate, record which ingredient entries were used, their quantities, preparation state and the final yield. If a meal total changes, you should be able to identify whether the change came from a portion, reference or ingredient edit.
Consider a hypothetical soup calculated as 1,200 kcal in a 2,000-g cooked batch. A 300-g portion is 180 kcal under that calculation. A different yield of 1,500 g would make the same 300-g portion 240 kcal, even without changing the ingredient energy total. The source needs the yield as well as the ingredient list.
Use provenance as a practical filter
Choose the most appropriate record for the task. A current package label can be the best match for the exact product; a detailed generic reference may be more useful for a plain ingredient. A photo estimate is useful when it is treated as an editable proposal with uncertainty.
Our methodology explains how this site's food references and illustrative calculations are presented. We do not treat user activity or an AI-generated number as laboratory verification. When you log a repeat food, keep its source and serving readable so the record can be improved rather than merely repeated.
Common questions
Does a missing nutrient field mean the food has none?
No. It can mean the source did not report that nutrient. Treat unknown data differently from an explicitly measured or declared zero.
Are all USDA food entries laboratory measurements?
No. FoodData Central contains distinct data types, including manufacturer-label and compiled records. Inspect the type and description.
Sources & methods
- USDA FoodData Central data documentationDifferent food data types have different provenance, preparation detail and completeness.
- FDA: understanding the Nutrition Facts labelServing size, servings per container and interpreting a packaged-food label.
- Cronometer data sourcesOfficial distinctions between laboratory-based and label-based food records.