TL;DR
We ran 12 skincare searches through AI, the kind a shopper actually types, and logged every brand it recommended and every source it pulled from. The recommendations came almost entirely from editorial roundups like Wirecutter, Vogue, Elle, and Allure, plus medical sites, and not once from a brand's own website. Reddit, despite its grip on Google, never showed up. Legacy brands like La Roche-Posay and CeraVe won the generic searches, while DTC challengers only broke through in niche categories they could own, like clean, K-beauty, or founder-led luxury. If you run a skincare brand and you want AI to recommend you, the takeaway is blunt: optimizing your own site barely moves it. Getting into the roundups AI trusts, in a category you can actually win, is the game.
This article is the plain-language companion to our working paper on SSRN, A Cross-Engine Content Analysis of Brand Recommendation in AI Answer Engines (Zam, 2026), which sets out the preregistered, cross-engine method in full. Full citation is at the end of this article.
What we did
We ran 12 consumer skincare searches through Perplexity on August 23, 2026, logged out from a US session, and recorded two things for each: which brands the answer recommended, and which sources it cited. The searches spanned concerns (acne, anti-aging, sensitive skin), products (vitamin C, retinol, sunscreen, moisturizer), and categories (best brands overall, K-beauty, clean, affordable, luxury), so the sample covers how a real shopper moves from a problem to a product to a brand.
Two caveats before the findings. This is one engine, one run per search, on one day, so read it as a snapshot, not a census. We used Perplexity because it shows its sources openly, which most answer engines don't, and other engines pull from an overlapping but different set. What repeated across all 12 searches is the mechanism, and that's the part worth acting on.
Where AI's skincare recommendations come from
Across all 12 searches, AI built its recommendations from editorial roundups and medical sites, and never once from the brand's own website. Every answer leaned on "best of" lists published by outlets like Wirecutter, Vogue, Elle, Allure, and Cosmopolitan, with medical sources such as Mayo Clinic, Cleveland Clinic, and the American Academy of Dermatology anchoring the health-adjacent searches on acne and sunscreen.
This is the finding that should change how a beauty brand spends. The model didn't decide CeraVe was worth recommending by reading CeraVe's site. It read a stack of "best moisturizer" articles and repeated the names that kept appearing. Your product pages, your ingredient copy, your on-site blog, none of it was the source. The roundup that mentions you is. On the K-beauty search the model said the quiet part out loud, naming its picks as "brands that repeatedly appear on best-of lists from Vogue, ELLE, Good Housekeeping, Marie Claire." That's the whole ranking logic in one sentence.
Reddit didn't show up, and that matters
Across all 12 skincare searches, Reddit was never cited, which cuts against the popular claim that Reddit now runs AI recommendations. For skincare on Perplexity, editorial and medical sources did the work, and community threads didn't make the cut once.
Worth a caveat: this is one engine, and Google's AI Overviews lean on Reddit harder than Perplexity does, so the picture shifts by platform and category. But it's a useful correction for anyone about to pour effort into seeding Reddit threads for beauty visibility. In this data, the editorial roundup was the far higher-value target, and it wasn't close.
Legacy brands own the generic searches
On broad searches, a handful of legacy brands took nearly every slot: La Roche-Posay showed up in 7 of the 12 searches and CeraVe in 6, with The Ordinary, Cetaphil, and SkinCeuticals close behind. Most are owned by conglomerates like L'Oréal and sit on essentially every editorial list ever written, which is exactly why the model keeps surfacing them.
| Brand | Searches it was recommended in (of 12) |
|---|---|
| La Roche-Posay | 7 |
| CeraVe | 6 |
| The Ordinary | 3 |
| Cetaphil | 3 |
| SkinCeuticals | 3 |
The lesson for a challenger is to stop fighting there. You aren't going to out-list La Roche-Posay on "best skincare brands," because it's on every roundup the model has ever ingested. Spending to rank your own site for that term is spending against a wall that took a decade and a conglomerate budget to build.
Where DTC brands actually break in
Challenger DTC brands did get recommended, but almost entirely in niche and values-based categories they could dominate: clean, affordable, K-beauty, and founder-led luxury. The "best clean skincare brands" search was wall-to-wall independent names, Youth to the People, True Botanicals, OSEA, Tata Harper, and Cocokind among them, pulled from specialist roundups instead of the big glossies.
The same held everywhere a challenger showed up. Supergoop won on sunscreen, Maelove and Good Molecules on affordable, Beauty of Joseon on K-beauty, Augustinus Bader on luxury. Each one led a category it could plausibly own, and each got there by being on the specialist lists for that category, not by beating CeraVe on the generic term. That's the pattern worth copying: a brand that can't crack "best skincare" can absolutely be the answer to "best clean vitamin C serum" if it owns the roundups for that lane.
What this means for a DTC skincare brand
If you want AI to recommend your skincare brand, the study points to three moves, and none of them is on-site SEO: get placed in the editorial roundups AI cites, pick a specific category you can realistically own, and earn the reviews and press that get you onto those lists. In that order.
Get into the roundups first, because that's where the recommendations are actually sourced. Pitch the beauty editors at the outlets AI reads, the Wirecutters and Allures and the niche affiliate blogs for your category, since a single placement on a list the model trusts does more than a year of your own blog. Then own a lane, because you won't win "best skincare," so define the narrower category you can lead and get on those lists specifically. And earn the proof, because reviews, third-party testing, and press are the inputs that win the placements in the first place. This is the same mechanic we've written about for other verticals in answer engine optimization for Shopify and AI search optimization for DTC brands.
There's one more piece, and it's the one most brands skip. When AI does send a shopper, your storefront has to convert them, on the first visit, on mobile, in the couple of seconds before they bounce. Earning the recommendation and then dropping that traffic onto a slow, generic store wastes the hardest-won click in beauty. That's the work we do: fast, custom storefronts built to convert, which is where a well-run conversion setup and real site speed stop being nice-to-haves. If you're a beauty or skincare brand and you want the storefront handled by a team that builds for this vertical, book a strategy call, and the best Shopify design studios for beauty and skincare brands is a good place to see who does this work.
Method: 12 skincare searches on Perplexity, logged out, US, August 23, 2026. One run per search. This snapshot is the plain-language companion to the full preregistered, cross-engine study on SSRN.
Cite this study
The full working paper, including the preregistered confirmatory design and cross-engine method, is available on SSRN:
Zam, Andrew. A Cross-Engine Content Analysis of Brand Recommendation in AI Answer Engines: A Skincare Pilot with a Preregistered Confirmatory Design (August 25, 2026). Available at SSRN: https://ssrn.com/abstract=7366498 or https://dx.doi.org/10.2139/ssrn.7366498.
