Beauty’s race for AI search visibility is on.
As AI chatbots like ChatGPT, Perplexity and Google Gemini quickly grow — data from SEO agency First Page Sage shows that during the fourth quarter of 2025, ChatGPT captured 17 percent of total searches versus Google’s 78 percent— beauty companies are racing to boost their visibility on those platforms in an effort coined as GEO, or generative engine optimization.
This shift comes after a decades-long reign of SEO (search engine optimization) as the dominant practice to boost online visibility, namely through a focus on keywords and user experience. But GEO is different in that it requires much more to become a top performer.
“There’s no one way to hack visibility on these LLMs [large language models], and if there was, they’d quickly shut it down,” said Emily Rose Campbell, head of performance at social media and e-commerce agency Iced Media.
ChatGPT, which is the most prominent AI search engine with 60.7 percent share, cites, on average, more than 22 sources per prompt response. Gemini, which comprises 15 percent of AI searches and is the fastest-growing tool in the category, cites between seven to 10 sources per response.
This means that in addition to optimizing owned websites and social media channels for LLM readability, brands must ensure, perhaps more than ever, that they appear prominently on commonly cited sources like Reddit, YouTube and TikTok, as well as annual awards and “best of” articles by beauty publishers, which, after experiencing somewhat of a cultural lull during the social media era, are poised to grow once again in relevance.
“You need sources that are not you, to say nice things about you,” said Evan Bailyn, founder and chief executive officer of First Page Sage. “Product recommendation media is coming back. It’s possible that ChatGPT and the others will change their algorithms eventually, but right now, these ‘best of’ lists…they’re more influential than Reddit.”
High-research beauty categories like makeup, skin and hair care over-index when it comes to AI searches, per a September 2025 report from Spate. Within those, the three most common kinds of prompts are those which follow a “best of…” format; those seeking a comparison of the benefits of one offering versus another, and ingredient-related queries, said Bailyn.
Products named after their hero ingredients thus have a certain advantage.
“That’s one of the reasons why we see brands like Paula’s Choice or The Ordinary rank so highly; their product names are their ingredients, which provides a very clear and easily citable signal for LLMs,” said Sky Canaves, principal analyst at eMarketer. “That also carries into product reviews, because every review of a product is inherently going to mention the ingredients.”
“AI loves lists, it loves comparison tables and it loves step-by-step guides,” said Aude Gandon, global chief digital and marketing officer at The Estée Lauder Cos., which in October kicked off a six-month GEO pilot program for three of its brands, with an aim to subsequently implement learnings across the portfolio.
Though Gandon declined to specify which brands Lauder is testing, she said that a key effort of the pilot is to elevate expert voices around the brands.
“Building [trust] is one component, and that comes from experts. For a makeup brand, makeup artists are the ultimate experts in terms of how to find and use certain products; for skin care brands, that means making sure that for any product which we have developed with dermatologists — the ingredients, claims and benefits — are easily accessible by LLMs,” she said.
At E.l.f. Beauty, efforts to make sure owned content (like product pages) are readable by LLMs has meant a shift to writing in Markdown. Invented in the early 2000s by John Gruber and Aaron Schwartz, Markdown is a backend writing format that, compared to, say, HTML, is more simple and readable for computers, and thus is booming once again in the era of AI search.
“If you structure your data in a Markdown, or ‘.md’ file, that’s an easy way to get your data in a format that the LLM can read,” said Ekta Chopra, chief digital and AI officer at E.l.f.
Because users fundamentally ask longer and more detailed questions via AI search engines (the average LLM prompt is 23 words, nearly six times the length of a traditional search engine query, per Soci), product web pages also need to be much more detailed on the backend in order to be deemed as relevant to those queries.
“Products have to be discoverable; they have to be recommendable, and they have to be transactable. So the biggest challenge that all brands face is that the amount of content you need, and the amount of specificity that you need in that content is just a lot,” Chopra said.
L’Oréal’s chief digital and marketing officer, Han Wen, said that the rate of change in the AI search space, too, makes strategizing more challenging.
“The fundamental muscles around LLMs are not necessarily new or different to a marketer, but the speed at which they are moving now is much faster,” she said. “With SEO, I don’t remember thinking that we needed to refresh or monitor SEO on a weekly or daily basis, because search crawlers just didn’t move that fast; you had a fair amount of time to say, ‘here are the new trending keywords consumers are thinking about,’ and a fair amount of time to be able to create the right type of content, merchandise it properly, and then have it surface.
“My teams who are working on [GEO] now are looking at it on a daily basis to see, ‘OK, what’s new? How do we rank on a particular set of prompts we care about?’” Wen continued. “‘What content do we have that we should push more? What content do we need to produce more of?’ — these things are changing on an almost daily basis.”
Nascent and ever-changing as the space may be, it’s still advantageous to experiment in it right now.
As commerce integrations like ChatGPT Checkout and Gemini’s even more nascent partnerships with Shopify and Walmart, which will similarly enable checkout in the chatbot, grow, GEO is poised to have a more measurable impact on brand performance in the coming years.
“This is a unique moment that we’re in. There is a first-mover advantage; a year and a half from now, it’s going to be harder to catch up if you haven’t been investing in the correct GEO strategies that get you to rank higher,” said Bailyn.
