Beauty is one of the most AI-researched categories there is, and one of the hardest to win, because the answers lean so heavily on sources you don't own. A D2C skincare or beauty brand that isn't found, described accurately, and recommended by AI is being cut from shortlists it never sees. Here's the playbook.
Why beauty is a high-stakes AI category
- Relentless research intent. Shoppers ask AI in concern-and-ingredient language: "best vitamin C serum for sensitive skin," "niacinamide vs retinol for acne," "fragrance-free moisturizer for eczema." Every query fans out across concern × ingredient × skin type × budget. Hundreds of branches per product.
- Third-party sources dominate. AI answers about beauty lean on Reddit communities, review sites, expert/derm roundups, and "best of" listicles. Exactly the sources beauty shoppers already trust. Your own PDP is one input among many.
- Claims and ingredients are scrutinised. AI will describe your actives, concentrations, and whether you're "good for" a concern, so accuracy and perception carry real risk.
The playbook
1. Map your category's fan-out
List the real questions: by concern (acne, dullness, ageing, sensitivity), by ingredient (vitamin C, retinol, niacinamide, SPF), by skin type, and the comparisons ("X vs Y," "alternatives to X"). Build intent-complete hubs that answer them, not a single product page per query.
2. Make your PDPs machine-legible
Beauty buyers, and AI. Want specifics, not vibes. On every product:
- Key actives and concentrations, full ingredient list, and what each does.
- Concerns addressed, skin types, and honest claims ("fragrance-free," "non-comedogenic").
- Structured data.
Product,Offer,FAQPage, so the facts are extractable.
This is what lets AI describe your hero products accurately instead of guessing.
3. Win the sources beauty AI trusts
Because third-party content dominates, the highest-leverage move is earned placement: get included in the independent "best [category]" roundups and reviews AI cites, and be genuinely discussed in the communities (skincare forums, reviews) that feed beauty answers. Trying to out-rank those roundups with your own self-listing page loses.
4. Manage perception and correct claims
Watch how each engine describes your hero products' efficacy and ingredients. AI working from stale or thin info can misstate a concentration, attribute a competitor's claim to you, or frame you as "basic." Fix wrong descriptions by making the correct facts reachable and corroborated.
5. Measure across engines, for your category
Beauty answers vary sharply by engine and phrasing. Track presence, share of voice, citations, and sentiment across ChatGPT, Gemini, Claude, and Google's AI surfaces, for your specific concern-and-ingredient queries. Over time.
In beauty, the buying decision increasingly happens inside an AI answer built from Reddit, reviews, and roundups. You win by being accurately legible to AI and present in the sources it trusts, not by shouting louder on your own PDP.
This is the category Buffy Intel was built around, and the work is concrete: legible product data, earned presence, accurate perception, measured across every engine.