Our Hair Products Score Lower Than Our Skincare. The Reason Is Mostly Our Own Method.

Eight of the fifty-four products we sell are for hair. They average 8.99 on our screening score. The other forty-six average 9.36.

A 0.37-point gap sounds like rounding error until you look at the bottom of the list. The two lowest-scoring products in the whole catalogue are both hair products: a JHP argan oil hair essence at 8.0, and a GDS hair-dye shampoo at 8.6. Nothing else we sell scores below 8.7.

We predicted part of this back when we published the eight ingredients that earned our worst grade, and said at the time that four of the eight sat in hair products. We recounted this week. It is six.

We carry more than one score per product and they disagree, so before any of this counts for anything: every figure in this piece is screening_score, recomputed from the fifty-four current products this week, not read off an older note. We published the audit explaining why that distinction matters and will not relitigate it here. One number from that audit is worth carrying over, though. Across all fifty-four products, our screening_score and our curator_score agree exactly twice. One of those two products is the JHP hair essence, at 8.0 and 8.

Six of eight

Across 1,846 ingredient entries we grade exactly eight as high-risk. Six of them are in the eight hair products.

The other two are a menadione in a toner and a Yellow No.4 in a foot cream.

Mid-risk tells the same story with more data behind it. Of the 304 ingredient entries in our hair products, 48 grade mid — 15.8%. Of the 1,542 entries everywhere else, 125 grade mid — 8.1%. Almost exactly double the density.

Take the JHP essence apart and the pattern is naked. Thirteen ingredients. The last five are Fragrance, Hydroxycitronellal, Limonene, Linalool and Citral. One high, four mid, five entries out of thirteen, and that alone drags it to 8.0 — the lowest score we have ever given anything. Its free_of chips say Paraben-free, Sulfate-free, Colorant-free. It has never claimed to be fragrance-free, and it isn't, and the label is honest about it. What it is, is a perfumed hair oil, sold as one.

Then the counter-example

Our best-scoring hair product ties the best score in the catalogue at 9.8.

It is the ANEAN argan oil curling lotion. Fifteen ingredients, zero high-risk, zero mid-risk, every entry low. Water, two humectants, argan oil, a diol, an olive ester, a thickener, a glycol, an emulsifier, tocopherol, glycerin, a preservative, EDTA, sea buckthorn oil, sea lettuce extract. It is a styling product, which means it stays on the hair all day, so the rinse-off argument does not rescue it — it just doesn't need rescuing.

Which kills the clean version of the story. There is nothing about a hair product that forces a low score. Six of our eight hair products score 8.9 or above. The category average is dragged down by two, and one of those two is a permanent hair dye, a product class that genuinely cannot exist without oxidative dye chemistry.

The number our score doesn't have

Here is the part where our method loses to the EU's.

The EU's Scientific Committee on Consumer Safety publishes the exposure assumptions every European safety assessment runs on. In the 12th revision of the Notes of Guidance, each product type gets a daily applied amount and a retention factor — how much of it stays on you. Shampoo: 10.46 g/day applied, retention factor 0.01, so 0.11 g/day of actual exposure. Hair conditioner: 3.92 g/day, same 0.01 factor, 0.04 g/day. Face cream: 1.54 g/day applied, retention factor 1.00, so 1.54 g/day.

At the same ingredient concentration, a face cream delivers about fourteen times the daily dose of a shampoo and thirty-eight times that of a conditioner. Our screening score contains no term for this. It reads an ingredient name, assigns a grade, and does not ask how long the product touches you.

Regulators build the same asymmetry into fragrance rules explicitly. The 26 declarable fragrance allergens have to be listed above 0.01% in rinse-off products and above 0.001% in leave-on — a tenfold difference decided by product format, not by the molecule. The industry's own use limits for hydroxyisohexyl 3-cyclohexene carboxaldehyde put hair styling aids and rinse-off conditioners at 0.2%, against 0.02% for lip products and deodorants. Ten times more of the same allergen, permitted, because it is going on hair.

The SCCS put the general point more bluntly than we would have dared to, in a 2026 workshop report: "A shift toward purely hazard-based decision-making will clearly undermine the scientific rigor and trust in the risk assessment process." Our screen is hazard-based. That sentence is about us.

What the patch-test data actually shows

So we went to the European patch-test data to find out which of our flags survive contact with reality.

PPD survives. In the 2021–2022 European baseline series results, pooled across 59 departments in 14 countries and 18,832 patients, p-phenylenediamine came back positive in 3.85% of the 13,563 people tested for it. For scale, in the same population nickel ran at 18.85%, Fragrance mix I at 6.39%, and paraben mix at 0.81%. PPD is roughly five times the paraben rate. Our high grade on it is defensible.

Resorcinol does not survive. A 26-year German-network analysis covering 2,678 female hairdressers and 6,244 female consumers describes it as "a hitherto commonly used coupling agent for oxidative hair dyes, but a very rare allergen" — and the authors dropped it from their analysis on those grounds. We grade it high, in the same product, at position 11.

Fragrance in hair products is genuinely unclear. That same analysis reports that "no clear association with hairdressing work and contact dermatitis related to hair cosmetic products has been found" for fragrances, while hair dyes were the top sensitiser in both groups. Fragrance mix I still runs at 6.39% in the general patch-tested population, so the allergens are real in general. Whether they are doing much from a rinse-off shampoo, we could not find out.

One more thing cut against our assumption. We had half-expected the scalp to be a delicate site that justified stricter grading. Regional absorption data put the scalp at about 3.5 times the forearm for hydrocortisone, so it is more permeable — but in a study where ten bald men took five successive washes with 20% sodium lauryl sulfate, back skin reacted more than the scalp did. The scalp is not the fragile site we were implicitly grading it as.

What our screening found

Standing in front of the shelf

A lower ingredient score on a hair product is not the same warning it would be on a face cream, and if you have been reading our 9.36-against-8.99 gap as though it were, that is our fault for publishing it without the exposure caveat attached.

Two flags on a shampoo are worth reacting to. A permanent oxidative dye containing PPD is the one with a 3.85% positive rate behind it and a patch test is a reasonable thing to ask for. Everything else on a rinse-off product gets roughly a hundredth of the contact a leave-on gets, by the regulator's own arithmetic, and our score does not know that yet. On a leave-on — a styling cream, a hair oil, a scalp tonic — read the flags the way you would read them on skincare, because the exposure is comparable.

We publish what we flag and how often and our screening records so that this kind of correction can happen in the open. Adding an exposure term is on the list. It is not built.