Romina Janette Riccetti
Last week, I asked ChatGPT for the best long-lasting liquid blush for oily skin. After two
follow-up questions about finish and pigment, it suggested Dior, Benefit, Rare Beauty and YSL,
citing Elle, Glamour and Insider Beauty. Claude gave a single answer, Rare Beauty’s Soft Pinch
in matte, citing PureWow and Elite Daily. Two assistants, two shortlists, and only one brand in
both: Rare Beauty, the label Selena Gomez launched in 2020. As a Rare Beauty customer, I
wanted to understand why, and the answer explains why influencer marketing for AI search now
matters so much.
AI engines recommend the brands that many independent sources describe in the same way,
for the same need, and many of those sources are creators. A Peec analysis of 30 million AI
citations found that Reddit and YouTube are the two most-cited websites across ChatGPT,
Gemini, Perplexity and Google’s AI features. Shoppers follow these answers: AI traffic to U.S.
retail sites rose 693.4% during the 2025 holidays, and it converted 31% better than other traffic.
What is generative engine optimization (GEO)?
GEO is the practice of making a brand easy for AI search engines to find, understand and cite.
The term comes from a 2024 Princeton and IIT Delhi paper, which showed that adding citations,
quotations and statistics could raise a page’s visibility in AI answers by up to 40%. SEO aimed
to rank first among blue links; GEO aims to be one of the few brands the AI names.
A few definitions help. A large language model (LLM) is the AI behind ChatGPT or Claude. A
generative engine writes one summarized answer instead of listing links. User-generated
content (UGC) is content made by customers and creators rather than the brand. A
micro-influencer has roughly 10,000 to 100,000 followers and high trust. E-E-A-T (Experience,
Expertise, Authoritativeness, Trustworthiness) describes reliable content, and Google states that
“trust is most important.” Finally, query fan-out is how AI tools split one question into several
searches, which is what ChatGPT did when it asked me about finish and pigment.
How is GEO different from SEO?
SEO targeted short keywords like “liquid blush” and measured rankings and clicks. GEO targets
long questions such as “best long-lasting liquid blush for oily skin” and measures mentions
inside answers. Google says there are “no additional requirements to appear in AI Overviews or
AI Mode,” so SEO foundations still count, but the shelf is smaller: an answer names three to five
brands, and Pew Research found users click a traditional result in only 8% of visits when an AI
summary appears, versus 15% without one.
Which sources do AI engines trust?
Each engine has favourites. According to Peec, Google’s AI Overviews and AI Mode cite
YouTube most, then Reddit, while ChatGPT leans on Wikipedia, Reddit and Forbes. My test
adds nuance: both assistants cited beauty magazines rather than Reddit or YouTube. Those
roundups, however, are shaped by months of creator buzz, so the AI reads the editorial layer
built on top of it. One test is an illustration, not proof, but the lesson holds: visibility in one
channel means visibility in one engine. A brand present on TikTok but absent from magazines
and retailer reviews may be recommended by Google’s AI and ignored by ChatGPT.
What do celebrity beauty brands teach us?
Rihanna’s Fenty Beauty (2017, launched with 40 foundation shades), Rare Beauty (2020) and
Kim Kardashian’s Skims (2019) each generated huge volumes of reviews, tutorials and articles.
Yet Fenty did not appear in my blush test, and Skims sells shapewear. AI does not recommend
celebrities; it recommends products that many sources describe in the same terms. Rare
Beauty’s team calls Soft Pinch one of TikTok’s first truly viral beauty products, and creators kept
repeating the same attributes: pigmented, long-lasting, a little goes a long way. Dior shows
another route, earning its place through editorial coverage rather than viral videos.
Fenty teaches the same lesson from another angle. The story the internet repeated about its
launch was inclusivity, so Fenty is more likely to come up in answers about shade range than
about blush longevity. A brand’s AI visibility follows the specific reputation creators have built for
each product, not the fame of its founder.
Celebrities or micro-influencers?
Both matter, differently. Celebrities make a brand a known entity with press coverage that
ChatGPT trusts, but rarely answer precise questions. Micro- and nano-influencers produce the
detailed, honest content that does, and customers add volume through reviews. Brands are
adapting: Influencer Marketing Hub found that 52.83% of brands plan to expand work with
micro-creators in 2026.
How can brands use influencer marketing for AI search?
First, audit visibility by asking AI tools twenty real customer questions each month. Second, brief
creators on specific questions, such as whether a blush lasts eight hours on oily skin, and ask
them to name the product clearly. Third, turn videos into text on the brand’s site, with transcripts, short FAQ answers and Article, Product and FAQPage markup. Fourth, be present where AI looks: Reddit, YouTube, retailer reviews and the press.
Fifth, stay honest. The FTC requires creators to disclose paid partnerships, and its 2024 rule
bans fake and AI-generated reviews and fake followers. Sixth, refresh content with every
launch, since AI favours recent information. Success is measured through citation rate, share of
voice against competitors (Rare Beauty scored two out of two in my test) and AI referral traffic in
Google Analytics 4. Brands can also compare the conversion rate of AI-referred visitors with
other channels; tools such as Peec AI automate tracking, but a monthly spreadsheet works for a small brand.
Frequently asked questions
Does Reddit and YouTube content help brands appear in AI answers?
Yes. They are among the most-cited sources across major AI engines, and detailed discussion
there becomes material the AI can quote, directly or through magazine roundups.
Is influencer marketing still worth it with AI search?
More than ever. Fewer clicks mean that the brands appearing in the AI’s answers are the ones
that matter most. And one of the main sources of information that the AI uses to determine
which brands to name in the first place is the content created by influencers.
Are micro-influencers better than celebrities for AI visibility?
Celebrities help create brand awareness and generate press coverage that ChatGPT can use to
create answers, but micro-influencers create the detailed and honest reviews that answer more
precise questions. A brand can use a combination of both to create the best answer possible.
Can a brand buy fake reviews to get recommended?
No, it is illegal to buy fake reviews in the US. Under the FTC’s 2024 rule, a company cannot buy
fake reviews and also cannot use AI to generate reviews. Many independent sources are
cross-checked by the AI engines so any fake signals would be overwhelmed by the genuine
conversation about a brand.
Conclusion: creators write the AI’s answer
My blush question showed that two AI assistants agreed on only one brand: the one creators,
editors and customers had described consistently for years. Celebrities open the door, but
creators write the answer. Brands that brief creators on real questions, structure their content
and track citations, not just likes, will be the ones AI recommends next.