The most common fat on food labels doesn’t name an oil

Open Food Facts is the largest open database of packaged food in the world: crowd-contributed, free to query, and specific enough to say which ingredients a given barcode declares. We asked it a question with a single-word answer — which fat do food labels name most often?

The answer is not a fat. On August 28, 2026, 299,937 products in the database declared “vegetable oil” and left it there: no plant, no species, nothing a shopper could act on. That is more than sunflower and palm oil combined (218,961), and it is the largest single entry by a wide margin.

Why a word that names nothing is the interesting one

“Vegetable oil” is not a mistake or a gap in the data — it is what the package says. It is a category that can legally contain sunflower, rapeseed, soybean, palm, cottonseed or a blend that changes between production runs, which is precisely why manufacturers use it: a declaration that does not name a plant does not have to be reprinted when the supplier changes.

The consequence lands on everyone who has decided to avoid a specific oil, for whatever reason — allergy, religious rule, a nutrition belief they hold strongly, or simple preference. For 299,937 products, the label cannot answer them. Nor can any app, ours included: a scanner reads what is declared, and if the declaration is a shrug then the honest output is a shrug.

The ranking everyone quotes is wrong, because of a name

Rapeseed and canola are the same plant. Canola is a mid-1970s Canadian cultivar of rapeseed bred for low erucic acid, and the name is what North American labels use where European ones say rapeseed. In the database they are two separate tags, so every ranking built on those tags splits one plant in half.

Merge them and the order changes: 125,392 plus 41,121 is 166,513 — ahead of sunflower, and second only to the unnamed category. The single most-declared seed oil in the world’s food database is the one that goes by two names.

What is barely there

The animal fats that dominate online conversation about cooking oil are almost absent from the shelf. Beef tallow appears on 200 products in the entire database; lard, 1,548. Against 299,937 for the unnamed vegetable oil, that is a ratio of about 1,500 to one.

Two readings are available and the data does not choose between them. One: the tallow revival is a restaurant and home-cooking phenomenon that has barely touched packaged food. Two: packaged food is where the cheapest fat wins, and tallow is not cheap. Either way, a shopper swapping to tallow-based packaged products is choosing from a shelf with about 200 items on it.

Butter, for scale, is on 56,553 products and olive oil on 45,994 — both far ahead of the animal fats, both far behind the word that names nothing.

What to do with this

Treat “vegetable oil” as unknown, not as safe or unsafe. It is the one declaration on a label that carries no information about the thing it names. If a specific oil matters to you, the product that names its oil is the one you can actually evaluate — whichever way you then decide.

Do not read the ranking as a health claim. Nothing here says any of these oils is good or bad for you; that argument is live, contested, and covered separately with both sides. This is a measurement of what labels say.

Watch the two-name problem elsewhere. Rapeseed and canola are the clean example, but the same split hits any ingredient with a regional name. Any count built on ingredient tags — ours included — is a count of words before it is a count of things.

Method

Every figure above is one call to Open Food Facts’ public /api/v2/search endpoint, filtered to a single ingredients_tags value, with the returned count read off it and nothing else kept. Read on August 28, 2026. 19 fats were queried; the script that does it is in the repository behind this site, so the numbers are reproducible rather than something you have to take on trust. OFF answers about ten such requests a minute and 503s anything greedier, which is why the sweep is small and paced.

These are counts of records, not of a market. Open Food Facts is contributed by volunteers and skews European; a fat common in a region with few contributors is undercounted here. A product listing two fats is counted under both. And a count of tags inherits the taxonomy’s decisions — the rapeseed/canola split above is exactly that problem, visible.

One number we could not get: the size of the whole corpus. An unfiltered search is the single query OFF refuses every time, so the figures here are absolute counts rather than shares of the database.