Estée Lauder owns the beauty brand that AI engines cite most often, and ranks eighteenth under its own name. 5W’s Beauty AI Visibility Index, released on 2 July, puts The Ordinary first on citation share at 7.0%. The house that bought it, with decades of advertising history behind the name, earns 1.5%. The distance between those two numbers is not a failure of content strategy. It is a naming decision The Ordinary made in 2016, for reasons that had nothing to do with machines. 5W’s own report also credits part of Estée Lauder’s slide to a pullback in content-infrastructure spend in early 2026, an incremental factor next to a naming decision made a decade earlier.

The Ordinary’s products are named after the query. Niacinamide 10% + Zinc 1%, Hyaluronic Acid 2% + B5, Retinol 1% in Squalane: the label is the specification, and the specification is what a buyer types. 5W ran more than 80 buyer-intent prompts through ChatGPT, Claude, Perplexity, Gemini and Google AI Overviews, among them “best niacinamide product.” A brand whose product name is the answer string is not being retrieved. It is being matched. Brandon Truaxe launched the line in August 2016 under a company he called The Abnormal Beauty Company, and the brand still describes its own work as a refusal of “fleeting trends or exaggerated claims.”

The trade read this as independents beating incumbents, which is wrong on the index’s own facts. 5W groups The Ordinary, Drunk Elephant and Glow Recipe as “ingredient-led independents” holding 31% of citations against 4% for legacy mass brands, and Glossy carried that framing to the industry. Two of those three are conglomerate assets. Estée Lauder completed its purchase of Deciem, The Ordinary’s parent, on 31 May 2024, closing out a stake it had been building since 2017. Shiseido took 100% of Drunk Elephant in 2019. The category 5W calls independent is a shopping list the incumbents already worked through.

L’Oréal’s position makes the pattern harder to read as an accident. CeraVe, La Roche-Posay and SkinCeuticals sit second, fourth and ninth on the index, and together they out-cite The Ordinary by two to one. L’Oréal bought CeraVe from Valeant in January 2017 and filed it beside La Roche-Posay in its Active Cosmetics division, back when the case for it was drugstore shelf space and nobody was modelling citation share. Those brands were built to be described by dermatologists, in the flat declarative register a clinician uses. That register now works as a retrieval asset. It was bought as a distribution strategy.

Citation share is not revenue, and the index does not claim otherwise. What makes the number worth reading is that it moves with the money in the one place we can check: Estée Lauder reported The Ordinary’s net sales up double digits against flat skincare categories, as Glossy noted in the same report. A brand can be cited and never bought, and 5W’s method cannot tell you which. But a brand that is never cited is not in the consideration set at all, and in a channel that returns a handful of names rather than a page of them, absence compounds faster than it used to.

The advantage accumulated before anyone knew it was an advantage — and a conglomerate can buy the brand that has it without moving its own name an inch.


The strongest objection is that the index measured its own prompt set. Ask “best retinol for beginners” and a product called Retinol 1% in Squalane wins by construction; the instrument and the reading share a vocabulary, and 5W sells the generative engine optimization its findings make the case for. For the thesis to fail, narrative brands would need to hold the ground wherever the question is not shaped like a chemistry problem. A second dataset comes close to showing exactly that. eMarketer’s independent read of more than 5,200 ChatGPT responses across nine personal care and beauty categories in the same quarter found La Roche-Posay recommended in 81% of facial skincare queries, and no clear winner at all in makeup. Charlotte Tilbury, the most narrative-driven name in 5W’s top five, is the top brand by citations among primarily makeup labels, Glossy found in the same data set.

So the finding is about skincare, not about beauty. Specification wins where the category admits specification: where the question has a right answer and the answer has a molecule in it. Fragrance is the nearest thing to a control. The highest-placed scent brand in 5W’s top 25 is Le Labo, twenty-third, at 1.0%, in the category that spends the most per dollar on story and offers a model the least to specify. Narrative is not being repriced everywhere. It is being repriced in the aisle where the buyer’s question has a correct answer, and beauty’s mistake would be reading a skincare result as a verdict on the whole shelf.

If beauty reads this index as a content brief, it will spend two years publishing ingredient pages into a corpus that already discounts brand-owned copy. The engines mostly repeat what other people wrote, which was the point in May when beauty started buying book clubs rather than keep renting an introduction. Every brand can put a dermatologist in a byline, and the moment they all do, the trait stops discriminating and the model falls back on whatever else does. The real question for a beauty CMO is whether the product survives being described accurately by a stranger in one line. That is answered in the lab, not the copy deck. The Ordinary answered it in 2016 without meaning to; everyone else has to answer it on purpose, and on purpose costs more.