• AI visibility

Guide · AI search

How do skincare brands get their products found when shoppers ask AI about their skin?

By publishing clear, substantiated answers to the concern and ingredient questions shoppers ask, backed by real expert involvement and honest reviews. Skincare is a category where shoppers cannot judge a product before using it, so assistants and buyers both lean on familiar names and trust signals. This guide covers where AI sits in skincare discovery, what research shows about how assistants pick skincare products, and how to compete without crossing the lines the FDA and FTC draw on claims. It is not legal or medical advice; check claims with your regulatory counsel.

By Underneath Team Guide Updated

The short version

  1. Dermatological skincare is where the growth is: L’Oréal’s Dermatological Beauty division, home to CeraVe and La Roche-Posay, grew 10.6% like-for-like in the first half of 2026, with all three of its main skincare brands up in double digits (L’Oréal (opens in a new tab)).
  2. Shoppers already ask AI about their skin: in a survey of 1,238 US women by skin-analysis vendor Haut.AI, 62% were interested in or already using AI for skincare or bodycare advice (opens in a new tab).
  3. Familiarity wins ties in skincare: when ten products had identical ratings, prices, reviews and ingredients, three AI models picked the one real brand, such as CeraVe, in all 670 valid trials (Chu and Hou, 2026 (opens in a new tab)). A rating edge of just +0.075 stars was enough for an unknown brand to win half the time.
  4. Trust is contested online: 21% of Americans rely on Instagram or TikTok influencers for skincare advice, 36% of Gen Z, and more than 16 million adults cut back or stopped using sunscreen because of online claims (American Academy of Dermatology (opens in a new tab)).
  5. The rules apply to what assistants read: the FDA (opens in a new tab) says claims “on the Internet” help decide whether a product is a cosmetic or a drug, and the FTC’s final rule on fake reviews (opens in a new tab) allows civil penalties, including for AI-generated fake reviews.

Who buys skincare, and what is a customer worth to a brand?

Concern-led shoppers who build routines, so a customer won on one product is worth a regimen and its repurchases.

US skincare is large and steady. Circana reports (opens in a new tab) that prestige skincare grew 3% in dollars in 2025 and was the fastest-growing prestige category by units sold, while skincare at mass retail grew 6% in both dollars and units, driven by facial cleansers and moisturizers.

Dermatological brands are the standout. At L’Oréal, Dermatological Beauty grew 10.6% like-for-like in the first half of 2026, with growth accelerating for a third consecutive quarter and divisional profitability of 28.4%. La Roche-Posay, CeraVe and SkinCeuticals all grew in double digits, and the company credits launches such as CeraVe Sun.

For a skincare brand, value comes from the routine. A shopper who trusts one product, often a cleanser, moisturizer or sunscreen, tends to try a serum or treatment from the same range and to repurchase what works. That is our inference from how regimens are sold rather than a published figure, but it explains why being named for one hero product matters beyond a single sale.

Where does AI sit in skincare discovery today?

Between a skin concern and the shortlist, alongside social media, Reddit and, for some, a dermatologist.

Shoppers start from a concern: acne, redness, dryness, dark spots, aging, sun protection. Haut.AI’s survey found that 74% of US women believe AI could make their routines easier or more effective. Qualitative research by the consultancy 8th Day (opens in a new tab) describes people asking ChatGPT about conditions such as rosacea and perioral dermatitis, and buying what it recommends. The Nod Mag (opens in a new tab) describes a writer uploading a selfie and receiving a routine of named products; a dermatologist it interviewed warned that a photo “cannot reliably determine skin hydration, sebum production, collagen content, barrier function, inflammation or biological skin age.”

Assistants are already naming brands for plain skincare questions. Chu and Hou reproduce a real ChatGPT answer from May 2026 to the question “I would like to buy a good face moisturizer, which is the best?” that named CeraVe as “Best overall.”

AI arrives in a crowded field of advice. The American Academy of Dermatology’s 2026 survey of 1,132 US adults found that nearly half of Americans, and 64% of Gen Z, report encountering sunscreen misinformation online. That makes credible, accurate brand information more valuable, not less.

Which skincare questions do shoppers ask AI?

Questions about concerns, ingredients, compatibility and comparisons, often with skin type and sensitivities attached.

We wrote these sample prompts in a shopper’s voice; none come from real chat logs:

  • Concern: “Moisturizer for sensitive, acne-prone skin that won’t clog pores.”
  • Ingredient: “Niacinamide or azelaic acid for redness, and what’s the difference?”
  • Compatibility: “Can I use retinol and vitamin C in the same routine?”
  • Comparison: “Which mineral sunscreen for oily skin leaves no white cast?”
  • Credibility: “Which cleansers do dermatologists recommend most?”
  • Sensitivity: “Fragrance-free body lotion for very dry skin.”

Two features set skincare apart. First, many questions are framed around an ingredient rather than a brand, so a brand appears only if its product is clearly connected to that ingredient and concern. Second, some questions edge into medical territory. Brands should answer the cosmetic part well and point to a dermatologist for the rest, which is also what regulators expect.

How does an AI answer turn into skincare sales?

Through the hero product: an assistant names it for a concern, and the shopper checks, buys and builds a routine.

  1. The concern question. The assistant names a handful of products and explains why, often by ingredient.
  2. The check. Shoppers look for reviews, dermatologist involvement, ingredient lists and price. OpenAI documents (opens in a new tab) that ChatGPT’s product results draw on structured product data, other third-party content and review summaries built from “reviews from public websites,” which it does not verify.
  3. The purchase. The shopper buys at a retailer, a marketplace or the brand’s own site, depending on where the assistant’s links and the shopper’s habits lead.
  4. The routine. A product that works earns repurchase and a chance to sell the rest of the range.

The most valuable answers are the ones where a brand is named for a clear concern it can genuinely serve. A reasonable expectation is that hero products named in AI answers carry more of a skincare brand’s new-customer acquisition over time, though no public data yet measures this.

What decides which skincare products an assistant recommends?

Research shows familiarity, ratings and authority claims all matter; platforms document only general product signals.

Documented by the platforms. OpenAI says ChatGPT considers structured metadata such as price and description, other third-party content and public reviews, and that labels such as “Most popular” are model-generated, not verified.

Observed in a study of skincare specifically. Chu and Hou tested three AI models on moisturizers, exfoliants, sunscreens and cleansers. With nothing to separate products, the real brand won every time. But the advantage was fragile: a +0.075-star rating edge, 1.6 times the reviews or a 7.3% lower price let an unknown brand win half the time. Authority language, such as clinical-trial or dermatologist endorsements, was worth the equivalent of +0.17 rating points. In the tests, even fabricated clinical claims worked, which the authors themselves label “potential false advertising.” And when every competing brand used the same authority language, the gain from it fell from +0.802 to +0.007 in their payoff measure. Our article on whether AI assistants favor big brands covers the brand side of this research.

Our inference for brands. Real, specific, verifiable evidence is the durable advantage. Vague “clinically proven” copy is easy for competitors to copy and, as the study shows, loses value when everyone uses it; fabricated evidence is illegal and dangerous. See our article on legitimate GEO versus manipulation.

Trust factors specific to skincare. Substantiated claims, published clinical testing, genuine dermatologist involvement, complete ingredient lists, fragrance and allergen information, and reviews that describe real skin types and results.

What claims can a skincare brand make, and why does it matter for AI?

Only claims that keep the product a cosmetic, unless it is a drug, and only claims you can substantiate.

AI makes compliance more important, because assistants repeat what they find. The FDA decides whether a product is a cosmetic or a drug by its intended use, which can be established by “claims stated on the product labeling, in advertising, on the Internet, or in other promotional materials.” Claims that a product will “increase or decrease the production of melanin (pigment) in the skin, or regenerate cells” are among the FDA’s examples of claims that can make a product a drug. The FDA also says the term “cosmeceutical” has no meaning under the law.

Sunscreens are regulated differently. The FDA regulates them as nonprescription drugs, and in June 2026 it added bemotrizinol (opens in a new tab) as a permitted active ingredient. Under the Modernization of Cosmetics Regulation Act (opens in a new tab), cosmetic companies must keep records supporting safety substantiation and report serious adverse events to the FDA within 15 business days.

The FTC polices advertising. Its Health Products Compliance Guidance (opens in a new tab) says health-related benefits generally need “randomized, controlled human clinical testing” to meet its standard of “competent and reliable scientific evidence,” and that advertisers are liable for misleading endorsements, including expert ones. Its fake reviews rule bans buying or creating fake reviews, including AI-generated ones, and was approved by a 5-0 vote.

The practical point for GEO: everything a brand publishes to be found by AI is advertising. Content that overstates results may get repeated by an assistant and still be a violation. Mattress brands face a similar line on sleep claims, covered in how mattress brands win AI shoppers.

What does a skincare brand lose when assistants leave it out?

The new shopper for that concern, and often the routine that would have followed.

In Chu and Hou’s tests, brands that did nothing while competitors improved their descriptions received no recommendations at all. Those were controlled experiments with invented brands, not live markets, but they show how quickly a small, well-documented advantage moves AI recommendations in skincare. For an indie brand without the recognition of CeraVe, the upside is the reverse: clear, comparable evidence can overcome the familiarity edge.

There is also a reputational cost. Our study of Reddit citations found that Google’s AI Overviews cited Reddit in 17.9% of answers, and that 20.8% of sentences citing Reddit were not supported by the thread. If forum anecdotes are what assistants find about your product, they may describe it inaccurately. We have not seen a study measuring how much skincare revenue this costs, so treat it as a risk to check.

How does GEO work for a skincare brand?

It makes accurate, compliant answers about your products easy to find and verify; it cannot guarantee recommendations.

Generative engine optimization (GEO) for skincare usually covers six pieces of work:

  1. Concern and ingredient pages. Explain which concerns each product is designed for, the key ingredients and how to use them, in cosmetic terms your regulatory team has approved. Our article on the product content AI shopping assistants prefer explains why specific facts work better than adjectives.
  2. Evidence summaries. Publish what your testing actually showed: study design, number of participants, duration and results, so assistants and shoppers can see the basis for a claim.
  3. Real expert involvement, disclosed. If dermatologists helped formulate or test a product, name them and their role. Avoid vague “dermatologist-approved” lines you cannot document.
  4. Honest review programs. Collect detailed reviews from verified buyers without conditioning incentives on sentiment, as the FTC rule requires. Our article on fake reviews and AI recommendations explains why review integrity matters to assistants too.
  5. Consistent product facts everywhere. Keep ingredient lists, sizes, prices and claims identical across your site, retailers and marketplaces, since assistants combine them.
  6. Monitoring by concern. Track the concern, ingredient and comparison questions for your hero products across ChatGPT, Gemini, Perplexity, Copilot and Google’s AI features, and correct errors at their source. If an assistant repeats misinformation about your product, see how to fix wrong brand information in AI answers.

What can’t the evidence yet tell a skincare brand?

It does not show how often AI answers change which skincare product a shopper buys.

The Haut.AI survey comes from a vendor that sells AI skin analysis and covers US women only. The AAD figures describe social media and sun care, not AI assistants. The 8th Day and Nod Mag accounts are qualitative. Chu and Hou’s experiments supplied products to the models rather than letting them search the open web, and used invented brands. L’Oréal’s growth figures show where skincare demand is, not what drives it. No platform documents how it chooses between skincare brands for a concern question.

Where should a skincare brand start?

Start by checking what assistants say about your hero products for the concerns they are meant to address.

A useful first step is an audit of the concern, ingredient and comparison questions your customers ask, across the main assistants, showing which products are named, what claims are repeated about yours and which sources are cited. Paired with a review of the claims on your own pages, that shows where you are missing and where you may be misdescribed. If you want us to run that audit with you and plan the work around hero-product sales and routine adoption, reach out to our team. Our generative engine optimization service page shows how the ongoing work fits a skincare brand, from approved concern and ingredient pages to evidence summaries and honest review programs.

Frequently asked questions

Do people really ask AI for skincare advice?

Yes. In Haut.AI’s survey of US women, 62% were interested in or already using AI for skincare or bodycare advice. Dermatologists caution that AI cannot diagnose skin conditions from a photo.

Can we say our product is “clinically proven” to help AI recommend it?

Only if you can substantiate it. In tests, authority language helped, but the FTC expects health-related claims to rest on competent and reliable scientific evidence.

Does dermatologist backing help with AI recommendations?

In controlled tests, authority signals such as expert endorsements shifted recommendations. Real, documented dermatologist involvement is both more defensible and harder for competitors to copy.

Can a small skincare brand beat a famous one in AI answers?

In controlled tests, yes, when it had a clear advantage such as better ratings, more reviews or a lower price. With nothing to separate products, assistants chose the familiar brand.

Sources

Free strategy call

Some questions are easier to answer about your own business.

Bring the one that matters most. On a free 30-minute call we’ll take a first look at it and send you a short written read afterward.

Ask about my businessHow we work