Eight Ingredients in Our Own Catalog Got Our Worst Grade. Here They All Are.

We keep the ingredient screen in a spreadsheet. Fifty-four products, 1,696 ingredient rows, and one column called risk that only ever holds three values.

Filter that column to high and eight rows come back.

Eight out of 1,696 is a ratio we've written about before, in the piece where we first read all 54 labels, and it isn't the interesting part. The interesting part is that all eight rows sit inside products we currently sell. There is no quarantine shelf. Our worst grade is a grade we hand to our own inventory, and last week we promised a full accounting of it. This is that.

The eight rows, unedited

Six products. Four of them are hair products, which is not a coincidence and is a thread we'll pull separately.

Two of them are the product

PPD (Black formula; Brown and Dark Brown use PTD, toluene-2,5-diamine, in its place) and resorcinol (all three shades) are in the GDS dye shampoo because without them it is a shampoo. Oxidative colour needs a primary intermediate and a coupler, they diffuse into the shaft as small molecules, and peroxide polymerises them into pigment too large to wash back out. Take them away and you have a temporary rinse that sits on the cuticle and leaves in four washes.

So the honest framing isn't "these are dangerous." It's that these are the mechanism, and the mechanism carries a sensitisation cost regulators priced years ago. The EU caps PPD at 2% on hair after mixing and forces the box to say Hair colourants can cause severe allergic reactions, along with a line telling under-16s not to use it and a warning about black henna tattoos. Resorcinol is capped at 1.25% in oxidative dye and 0.5% in shampoos and lotions, and the SCCS signed off on that in 2021 while openly noting resorcinol inhibits thyroid peroxidase — they judged a safety margin of 4,000 sufficient.

The population numbers are smaller than the warning text suggests. A five-country European study patch-tested 2,739 people drawn from the general population and found PPD allergy in 0.8%. Among people who'd had a black henna tattoo it was 3.2% versus 0.6% in everyone else, an odds ratio above nine. In dermatology clinics, where the patients are already reacting to something, the figure runs 4–6%.

We wrote a whole piece on what the PPD question does and doesn't settle, so we won't rerun it. The relevant fact here is what our own benchmark turned up: the #1 grey-coverage shampoo on Amazon skips PPD and resorcinol entirely and screens 8.5 against our 8.6. It gets there by being a gradual reducer that builds over weeks and fades when you stop. A tenth of a point apart, doing two different jobs.

Menadione: the grade we can least defend

This one belongs to the eple:me toner, which scores 9.5 — the highest score of any product on this list. Menadione is vitamin K3, it sits twentieth on a thirty-item label, and every other ingredient in that bottle screens low.

We graded menadione high. Then we went to check the basis, and the basis is thin.

No regulator restricts menadione in cosmetics that we could find. It isn't in the EU's prohibited or restricted annexes. It isn't in the US colour or drug listings. The irony is sharper than that: vitamin K1 has been banned outright in EU cosmetics since 2009, and K1 is the one with a published cosmetic contact-dermatitis case series. The unbanned one is the one we flagged.

The human skin evidence for K3 is five patch-test-positive cases across four papers between 1980 and 2015, collected in a 2022 systematic review of contact dermatitis from vitamins. All five were occupational (pharmaceutical and veterinary workers), and all five involved menadione sodium bisulfite, a different, sulfite-bearing salt. That compound is also a known primary irritant, which is why the review recommends patch-testing it at 0.01–0.1% dilutions: at full strength you can't tell an allergy from a burn.

The strongest regulatory sentence anyone has written about it is EFSA's, from a 2025 opinion on menadione as an animal feed additive: "in the absence of adequate data, the additive should be considered a skin sensitiser." Read that again. It's a default applied because the file is empty, not a result.

Our grade is a precautionary one. It should be labelled as such in the database, and right now it isn't — the field says "high," the same word we use for PPD, which has an epidemiology. Those two things should not share a colour.

The foot cream is yellow on purpose

GOUNBAL's urea foot cream contains Yellow No. 4 and Yellow No. 5. We graded the first high and the second mid.

Yellow No. 4 on a Korean label is tartrazine, CI 19140 — which, under American numbering, is FD&C Yellow No. 5. The two systems collide on the same molecule. A Korean consumer avoiding "황색 4호" and an American consumer avoiding "Yellow 5" are avoiding the identical azo dye, and neither of them can tell from the other country's label.

Tartrazine is legal as a cosmetic colorant on both sides. The EU permits CI 19140 in all cosmetic products with no eye-area or lip restriction; the US allows it "for coloring cosmetics generally, including cosmetics intended for use in the area of the eye," subject to batch certification. The only place the FDA demands a tartrazine warning is ingested and absorbed drug products, and even there its own text concedes "the overall incidence of FD&C Yellow No. 5 (tartrazine) sensitivity in the general population is low."

We went looking for the contact-dermatitis literature. We didn't find it. Searching for topically applied tartrazine causing allergic contact dermatitis returns no case series, nothing we'd call a finding. What exists is a substantial oral literature (asthma challenge, chronic urticaria, fixed drug eruption) that has been quietly transplanted onto skin products by people writing ingredient guides. A urea foot cream is close to the lowest-exposure tartrazine there is.

Which leaves the actual objection, which isn't toxicological. It's that the dye does nothing. Urea softens callus. The yellow makes the tube look like a foot cream. That's a defensible reason to grade it down and an indefensible reason to grade it the same as an oxidative dye intermediate.

Three products, one ingredient, two grades

Rows five and six are where our own screen falls apart.

"Fragrance (Parfum)" appears as a high-risk row in the AgainMe shampoo and the Deepeer hair pack. The same string, spelled identically, appears in the JHP argan essence graded mid. The GDS dye shampoo's fragrance is also mid.

Nothing about the products justifies the split. If anything it runs backwards: JHP declares four allergens on its label (hydroxycitronellal, limonene, linalool, citral) while AgainMe declares none, and AgainMe is a rinse-off. The grade tracks whoever entered the row, not the formula.

Hydroxycitronellal is worth naming properly, because it's the one entry in this group with a real file behind it. The SCCS classes it among the established human contact allergens with over 100 reported cases, its animal potency lands it in the moderate band, and it's capped at 1.0% in all non-oral cosmetics under EU rules — a concentration cap, not just a labelling duty. IFRA restricts it separately, allowing up to 4.1% in rinse-off hair products, so on this ingredient the EU limit is the one that binds.

And the labelling floor is moving. Twenty-four fragrance allergens currently require individual naming in the EU; a 2023 amendment added 56 more that the SCCS identified, with non-compliant products blocked from being placed on the market after 31 July 2026 and off shelves by 31 July 2028. A lot of Korean hair products are about to get much longer ingredient lists. Some of what looks like a reformulation will just be disclosure catching up.

What our screening found

The practical takeaway

In front of a shelf, the useful sort isn't high-versus-low. It's whether the flagged ingredient is doing the job you're buying. All eight rows above split cleanly on that test.

PPD in a grey-coverage dye is the job. You accept it or you buy a gradual reducer and wait three weeks. Hydroxycitronellal in a hair pack is the job too, if the job is that the hair pack smells like something — and if it isn't, unscented versions of nearly everything exist. Tartrazine in a foot cream is not the job. It's decoration on a product nobody looks at.

Then there's the fourth case, menadione, where the flag means we don't know. We'd rather say that out loud than launder it into a colour. Our screening awards page lists the products that came through with nothing flagged at all, which is the shortcut if you'd rather not read a label today.

And when you see a number on this site, remember what we just admitted about our own three.