Ultra-processed food, and how to tell what is

Ultra-processed is not a synonym for unhealthy, and it is not a synonym for junk food. It is a description of how something was manufactured. A supermarket sandwich and a home-made one can contain the same ingredients and sit in different groups, because the question is not what is in it but what was done to it and by whom.

The term comes from NOVA, a classification developed at the University of São Paulo and now used in public health research worldwide. It sorts everything you can eat into four groups by degree of industrial processing. It is not a scoring system and there is no total.

The four groups

The whole system rests on one distinction: whether a food was made from other foods, or assembled from substances extracted from them. Group 1 is food with things removed and nothing added. Group 2 is what you cook with, extracted from group 1 by pressing, refining or grinding. Group 3 is group 1 with group 2 added, which is most of what a kitchen produces. Group 4 is the one that is different in kind rather than in degree.

The line people expect to matter is between fresh and packaged, and it is not. Tinned tomatoes are group 3. So is a block of cheddar, a jar of passata and a slab of bacon. The line NOVA draws is between food made from food and food formulated from ingredients you would not find in a domestic kitchen.

Where the term came from

NOVA was proposed in 2009 by a research group at the University of Sao Paulo led by Carlos Monteiro. It was built to answer a question existing systems could not: national diets were getting worse in ways that nutrient-based measures were failing to capture, because the products people were switching to were not obviously worse per gram of fat or sugar.

The proposition was that the relevant variable is not the nutrients in a food but what industrial processing does to it, and to how it is eaten. That is why the classification looks at manufacturing rather than at a nutrition label, and it is also why it produces answers that look strange against a traffic-light label on the front of a packet.

The criticisms, which are real

NOVA is contested, and a page claiming otherwise would be selling something. The most substantial objections are worth knowing, because they change how much weight the number on your receipt deserves.

• Group 4 is very broad. A fizzy drink and a fortified wholemeal loaf sit in the same group, and few people think those are equivalent in any sense that matters.
• Classification is not perfectly reproducible. Given the same product, different researchers sometimes assign different groups, particularly around the group 3 and 4 boundary.
• The mechanism is disputed. Critics argue that where research finds associations with health outcomes, the cause may be nutrient content or how the food is eaten rather than processing itself.
• It says nothing about portion size, frequency, or what the rest of a diet looks like.

Defenders point out that associations found in large studies persist after adjusting for nutrient content, which is the argument for processing being a variable in its own right. That debate is live and this site does not have a view on it.

The four confusions worth clearing up

Processed and ultra-processed are not the same word

Almost everything in a kitchen is processed. Bread, cheese, bacon, tinned tomatoes and olive oil are all processed and none of them is ultra-processed. Conflating the two makes the term useless, and it is the single commonest error in writing about this.

Fresh is not the test, and packaged is not either

Frozen peas are group 1. A tin of chickpeas is group 3. A fresh chilled soup can be group 4. Refrigeration, freezing and packaging add nothing to a food, so none of them moves it.

Price does not track processing

Premium ranges are not systematically less processed. Ready meals are the clearest case: the additives solve a manufacturing problem that a higher price does not remove. Dried pulses, rice, oats, frozen vegetables and whole chickens are among the cheapest things in a supermarket and are all group 1.

The version marketed as better is often the group 4 one

This is the pattern that surprises people most, and it recurs across a shop. Removing something from a food usually means adding something to do its job. Reduced-sugar baked beans gain a sweetener, low-fat yoghurt gains a thickener, and reduced-fat peanut butter gains a filler. Each addition is a group 4 marker.

Where it usually is in a real shop

People expect the ultra-processed share of their shopping to be the obvious things, and on a real receipt it usually is not. Crisps are cheap and few in number. The larger share tends to sit in weekly staples: bread, breakfast cereal, sausages and other packaged meats, oat milk and anything from the chilled ready meal aisle.

Some categories land the other way. Peanut butter is usually group 3. Most cheese is. Plain chocolate can be. Hummus generally is. That gap between where people expect the number to come from and where it actually comes from is the whole reason an itemised breakdown is worth having rather than a guess.

What share of a UK shop is ultra-processed

Published estimates for the UK vary considerably, and the variation is not noise: they measure different things. Some report a share of energy intake, some a share of food weight, some a share of items bought. A figure by energy will run higher than one by weight, because group 4 products tend to be more energy dense.

This site does not quote a headline figure it has not produced, and the reason is that a number without its denominator is not information. Bagged measures share of SPEND on food and drink, which is a fourth thing again, chosen because it is the one your receipt can actually evidence.

How to tell, in a shop

The practical test is the ingredients list, and specifically whether it contains things whose job is to make the product work rather than to feed you. Emulsifiers, stabilisers, humectants, bulking agents, modified starches, protein isolates, invert sugars, flavourings and colourings are all markers of group 4. One of them is usually enough, and each is defined plainly in the glossary.

The awkward part is that a receipt does not carry an ingredients list. It carries an abbreviated product name, roughly twenty characters of it, and every retailer abbreviates differently. Working backwards from that name to a classification is exactly what this site is about, and how a receipt is structured turns out to be most of the problem.

The markers, in the order you will meet them

• Emulsifiers. E471, mono- and diglycerides, soya lecithin. The commonest single reason an ordinary product is group 4, and the reason most sliced bread is.
• Modified starch. Controls texture through freezing and reheating, which is why it is in almost every ready meal.
• Non-sugar sweeteners. Sucralose, aspartame, acesulfame K, stevia. Usually the marker in a reduced-sugar version of something.
• Protein isolates and concentrates. Milk protein, soya protein, whey. The usual reason a high-protein product is group 4.
• Flavourings and flavour enhancers. Distinct from an ingredient: flavouring does not mean the food it tastes of is present.
• Humectants, stabilisers, bulking agents, firming agents, anti-caking agents used as additives.
• Added vitamins and minerals, which counts as fortification and is a marker even where the rest of the list is short.

Each of those is defined plainly in the glossary, including what job it actually does, because most of them are described far more alarmingly elsewhere than the facts support.

A shortcut that works most of the time

If every item on the list is something you could buy on its own as an ingredient, and could plausibly use in a kitchen, the product is group 1, 2 or 3. If any item is something that exists only as an input to manufacturing, it is group 4. That is not the formal definition and it agrees with it on the large majority of a supermarket.

Why a receipt is harder than a packet

Everything above assumes you are holding the product. A receipt is a different problem, and a harder one.

A till prints roughly twenty characters of product name, abbreviated by conventions each retailer invents for itself, with no ingredients list anywhere on it. HOVIS SOFT WHITE MED and a bakery sourdough arrive as the same kind of string. Two products that differ only in whether they contain an emulsifier are indistinguishable on the line.

That is why classification from a receipt is a judgement rather than a lookup, and why it is stated as one: the caveat on every result says the classification comes from product names, which do not list ingredients, and that some items are a judgement call. The alternative would be a confident number built on a guess.

It also differs by shop, and mostly because of loyalty pricing rather than the names. How a receipt is structured covers the codes, the promotions applied below the item lines, and why the lines rarely add up to the total.

What NOVA does not tell you

This matters more than it sounds. A group 4 item can be fortified, low in salt and cheap. A group 1 item can be none of those things. The classification is silent on nutrition by construction, and any page that quietly converts it into a verdict has stopped describing the system it is claiming to use.

It is also silent on cost, which is the part most people actually feel. That question is a different one and it is answered by what your own shopping costs over time rather than by any classification.

Why the share, and not the count

Counting items tells you almost nothing, because a shop is not a list of equal things. Twelve ultra-processed items out of forty sounds like a lot until you notice they were the cheapest twelve. Share of spend weights each line by what it cost you, which is the measure that changes when your shopping changes.

It is also the measure a bank statement cannot produce. A card transaction is one number for the whole shop, and no amount of open banking recovers the lines inside it.