Think Dirty vs Yuka: How the Two Scanners Differ — and the Question Neither One Asks

How each one actually decides

The reason two apps hand you different verdicts on the same bottle is not that one has better data. It is that they are running different rules, and both rules are defensible.

Yuka is a worst-ingredient system. It reads the full INCI list, assigns each ingredient to a risk band, and then lets the single worst one set the ceiling. One red-flagged ingredient and the product cannot score above 25 out of 100, no matter how unobjectionable the other thirty ingredients are. That is deliberately unforgiving, and it is why a product can drop from 90 to 20 on the strength of one preservative.

Think Dirty is a chemical-content system with a philosophical lean. It rates on what is in the formula rather than on brand reputation or marketing claims, which is genuinely to its credit — it will not be talked round by the word “clean” on the front of the box. But its underlying view tends toward natural is safer, so competent synthetic formulations get marked down in ways a cosmetic chemist would often dispute.

The criticism they share, stated fairly

Both apps attract the same objection from cosmetic chemists, and it is worth taking seriously rather than waving away. Hazard is not the same as risk. Hazard is what a substance could do in principle; risk is what it will do at the dose you actually encounter. Neither app models dose.

So an ingredient present at a fraction of a percent, within a legally permitted limit, is scored as though it were the dominant component. A formula can be penalised for a preservative that exists specifically to stop the product growing bacteria — a real, immediate hazard that neither app scores at all, because microbial contamination does not appear on an ingredient list.

This cuts both ways and we are not going to pretend otherwise. It is a genuine methodological weakness, and it applies to any tool that reads a label rather than a lab result — which includes ingredient-level microplastics screening. What a label-reading tool can tell you reliably is what is in the product. What it cannot tell you is how much of it will reach you. Every honest scanner is on the same side of that line.

The question neither app is built to ask

Here is the gap, and it is a gap of scope rather than of quality.

Yuka rates five things: endocrine disruption, allergenic potential, irritancy, carcinogenicity and pollution. That is a list of toxicological hazards — ways a molecule can interfere with a body. It is a sensible list, and for its purpose it works.

Now consider what a cosmetic microplastic actually is. Acrylates copolymer, nylon-12, polyethylene, PET glitter, polyurethane-11. These are not endocrine disruptors. They are not carcinogens or irritants. They are biologically inert — that is the entire point of them. A polymer is chosen precisely because it does not react with skin, does not degrade, and holds a film together indefinitely.

Which means the property that makes a synthetic polymer a microplastics problem — that it persists, does not break down, and accumulates — is invisible to a scanner looking for toxicity. It will pass every check, because the checks are looking for something else. A rinse-off exfoliant built on polyethylene beads can be, in strict toxicological terms, entirely unremarkable.

That is not Yuka being wrong or Think Dirty being careless. It is the wrong instrument for the measurement, in the same way a thermometer is not a poor scale. See the clean beauty loophole for how the same blind spot works in marketing claims, and the worst microplastic ingredients for the INCI names that carry it.

Which one should you actually use

• For food, Yuka, without much argument. Think Dirty does not cover food, and Yuka’s catalogue depth means it will usually return an answer rather than a blank.
• For cosmetics, Yuka finds more products; Think Dirty is stricter. If you want the cautious read and do not mind gaps, Think Dirty. If you want coverage and a slightly more evidence-weighted verdict, Yuka.
• Treat a bad score as a prompt, not a verdict. Look at which ingredient caused it. A product capped at 20/100 by one preservative present at 0.1% is a different proposition from one capped by three separate flagged ingredients.
• Do not read a good score as “no plastic”. This is the practical takeaway. A high Yuka score and a low Dirty Meter rating both mean “nothing on our hazard list”. They do not mean the formula is free of synthetic polymers, and they are not claiming to.
• Use two tools if the polymer question matters to you. Hazard and plastic content are genuinely separate questions, and no single score collapses them honestly. We would rather say that than pretend one number covers everything.

For how we compare directly against Yuka on the microplastics question, see MicroPlastics vs Yuka. For the ingredient names to recognise on a label, see microplastics in cosmetics.