How to match ingredients to food-composition data

Why ingredient matching matters

A recipe-calculation system multiplies nutrient values by ingredient quantities. If the selected food record does not represent the ingredient in the kitchen, the calculation can be consistently wrong even though the arithmetic is correct.

FAO/INFOODS publishes dedicated guidance for food matching because food descriptions, preparation states and data sources are not interchangeable.

Match the preparation state

Check whether the recipe quantity refers to the ingredient before or after cooking. A raw grain, a boiled grain and a ready-to-eat packaged grain can have very different water content and therefore different nutrients per 100 g.

The same principle applies to drained canned products, reconstituted powders, cooked meats, peeled fruit, frozen vegetables and other foods whose physical state affects the composition basis.

• Raw vs cooked
• Fresh vs dried or reconstituted
• Drained vs including packing liquid
• Edible portion vs as-purchased weight
• Plain vs prepared with added ingredients

Generic vs branded data

Generic food-composition records are useful for basic ingredients. Branded records can be more appropriate when a specific commercial product has a formulation that matters nutritionally, such as sauces, plant-based products, breakfast cereals or prepared foods.

USDA FoodData Central separates data into types including Foundation Foods, FNDDS, Branded Foods and SR Legacy. The documentation explains that these types come from different sources and have different update patterns.

Use regional data thoughtfully

Food composition can vary by country, variety, production method and fortification practice. FAO notes the value of country- or region-level food-composition tables and maintains an international directory through INFOODS.

For a business operating across markets, a single ingredient library may need source rules. A product sold in the UAE, for example, may be better represented by its manufacturer information or an appropriate regional source than by an unrelated branded record from another country.

Handle missing data explicitly

A food record may not contain every nutrient. Missing should not automatically be interpreted as zero. USDA Foundation Foods documentation notes that a nutrient may be absent because it has not been analysed, while in other cases a nutrient may genuinely not occur in the food.

A professional workflow should distinguish known zero values from missing values and flag gaps that matter to the intended analysis.

Keep an audit trail

Store the source database, food identifier, food description and any matching note alongside the ingredient profile. If the mapping is later changed, the recipe should be recalculated and the change should be reviewable.

This becomes especially useful when a supplier changes, a formulation is updated, or an authority or nutrition reviewer asks how a value was derived.