Skip to content
Mir AIhome

How to read an ingredient label abroad

5 min readMir Labs

A packet of biscuits in a shop where you do not speak the language is a small puzzle with a real consequence. This is a practical guide to solving it: how ingredient lists are structured, where the useful information hides, what the numbers mean, and how to behave when you cannot read the thing at all. It is written from building Bitewise, which photographs labels and judges them against a dietary profile, but almost everything here works with nothing but your eyes.

Ingredient lists are ordered by weight

In most food-labelling regimes, including the European Union’s, ingredients are listed in descending order of weight at the time of manufacture. The first three or four items are most of the product. If “sugar” is second, the product is largely sugar. If the ingredient you are worried about is last, it is present but minor.

This ordering is the first thing to look at even when you cannot read a word, because it tells you where to focus. The compound ingredients in brackets, for example “chocolate (cocoa mass, sugar, cocoa butter, emulsifier: soya lecithin)”, are broken out too, and the bracketed sub-list is where allergens most often hide.

Allergens are emphasised, but not everywhere the same way

The EU requires fourteen allergens to be emphasised in the ingredient list whenever they are present as ingredients: cereals containing gluten, crustaceans, eggs, fish, peanuts, soybeans, milk, tree nuts, celery, mustard, sesame, sulphites above a threshold, lupin and molluscs. Emphasis is usually bold, sometimes capitals or underlining. That means a quick scan for bold words in a European ingredient list is a fast first filter, even in a language you do not read.

Outside the EU the rules differ. The United States requires a “Contains:” statement or the plain-language allergen name in brackets, and its list of major allergens is shorter and slightly different. Japan mandates labelling for a smaller set of allergens and recommends it for others. Some countries emphasise nothing. The habit to build is: look for the emphasis first, but do not conclude anything from its absence.

“May contain” is a separate sentence, and a separate question

The ingredient list says what was put in. The cross-contamination statement, usually beginning “may contain”, “may contain traces of”, or “produced in a facility that also handles”, says what might have got in. These are two different claims and they matter to two different people.

If you avoid an ingredient by preference, the ingredient list is your document. If you avoid it because a trace could send you to hospital, the “may contain” line is at least as important, and its absence is not reassurance. Precautionary statements are voluntary in most jurisdictions. A packet that prints no warning has not told you the factory is clean; it has told you nothing.

Bitewise handles these as separate fields for exactly this reason. Declared allergens appear as their own row, the label’s own “may contain” line is quoted rather than summarised, and if you have marked an allergy as severe, a trace warning that names it becomes an avoid verdict rather than a footnote.

E-numbers and INS numbers

In Europe, additives are often printed as E-numbers: E120, E322, E471. The same codes appear in much of the world without the E, under the Codex Alimentarius International Numbering System, so “INS 471” on an Asian label is the same substance as “E471” on a European one. The letter suffix matters: E160a and E160b are different colours.

The code tells you the function and identity of the additive but frequently not its origin, and origin is what a vegan, halal or allergen question turns on. E120 is always made from cochineal insects. E471, mono- and diglycerides of fatty acids, may be plant or animal fat, and labels rarely say which. There is a separate article on which E-numbers are not vegan that goes through the list Bitewise itself uses.

Ambiguous words are ambiguous in every language

Some phrases are vague by design and should be read as “unknown” rather than “fine”:

  • “Natural flavours” / “natürliche Aromen” / “arômes naturels”: the source is undisclosed and can be animal or plant.
  • “Vegetable fat” of unstated origin: usually palm, not always.
  • Enzymes: may be microbial, plant or animal derived. Rennet in cheese is the classic case.
  • Glycerin, lecithin, mono- and diglycerides: plant or animal.
  • Shellac and colourings of unstated origin.

Bitewise treats every item on that list as caution with an explanation, never as safe. The reasoning is that an app under pressure to be helpful will round “it depends on the manufacturer” to “probably fine”, and that rounding is precisely the mistake a person with a restriction cannot afford.

The one rule: absence of evidence is not evidence of absence

If you cannot read the list, or can read only part of it, you do not know what is in the product. That sounds obvious, and it is the rule most often broken, usually by an app or a well-meaning friend saying “I don’t see anything wrong”. Not seeing is not the same as seeing nothing.

This is the founding rule of Bitewise. If the ingredient list is blurred, cut off or only partly visible, no check may return safe; it returns “not sure” and says what is missing, with a tip for a better photo. The overall verdict for a product is the worst of the individual checks, computed on the server rather than left to the model. It is deliberately conservative, and it will sometimes tell you “not sure” about something that turns out to be fine. That is the correct direction to be wrong in.

A practical routine in the aisle

  1. Find the ingredient list, not the nutrition table. It is the paragraph, not the grid.
  2. Look for emphasis — bold or capitals — and read those words first.
  3. Find the “may contain” sentence, usually directly after the list, and decide whether it applies to you.
  4. Scan for E or INS numbers if you avoid animal-derived or gluten-bearing additives.
  5. Photograph the whole list in one frame, flat and lit, if you are using an app. A cut-off edge is the commonest reason for a “not sure”.
  6. If in doubt, treat it as unknown. For a severe allergy, check with the manufacturer. No label reader, human or software, replaces that.

Bitewise is informational and is not medical advice; it can be wrong, and it says so on every result. Used the way this article suggests, as a fast, conservative first read that names its reasons, it makes the puzzle in the shop a good deal smaller. It does not make the consequences smaller, and it should not pretend to.

All articles