The 2026 State of AI Visibility
How ChatGPT, Claude, and Gemini answer buyer questions about face serums — why a single product (SkinCeuticals C E Ferulic) pulls 2.3× the runner-up, how CeraVe places four SKUs in the top 12, and where the citation graph actually leads.
Executive summary
Face serums are one of the most concentrated commercial categories we’ve measured. A single product — SkinCeuticals C E Ferulic — owns nearly 10% of every AI answer, roughly 2.3× the runner-up. Beyond that, AI recommends products, not brands: CeraVe places four different SKUs in the top 12.
The 60-second read
SkinCeuticals C E Ferulic pulls 69 mentions across 221 answers — 9.8% share, 2.3× the runner-up. This is the closest thing to a “default answer” we’ve seen in a commercial category since Pacvue in Amazon PPC. If your brand competes on any “best vitamin C serum” or “best anti-aging serum” query, you’re essentially competing for #2.
CeraVe places four different products in the top 12: Resurfacing Retinol Serum (#2, 30 mentions), Skin Renewing Vitamin C Serum (#4), Skin Renewing Retinol Serum (#6), Hydrating Hyaluronic Acid Serum (#12). The Ordinary places two SKUs (Niacinamide 10% + Zinc 1%, Hyaluronic Acid 2% + B5). AI is doing SKU-level segmentation before it does brand-level — buyers see specific products, not brand landing pages.
Every single top-12 product hits 100% positive sentiment except SkinCeuticals at 99% (one critical mention across 69). When AI describes serums, it uses cleaner language than in any category we’ve measured. Sentiment is a wedge in email marketing and Amazon PPC; here it’s a wash. Share of voice is what actually differentiates.
ChatGPT, Claude, and Gemini all put SkinCeuticals C E Ferulic at #1. Unusual — in most categories the top pick diverges by model. This is a signal that the professional-dermatology reputation of C E Ferulic has permeated every AI’s training data uniformly. Where models diverge is at #2 and #3: Claude picks TruSkin and The Ordinary; Gemini picks CeraVe and Maelove; ChatGPT sticks with CeraVe SKUs.
The top 5 citation domains are all brand-owned DTC sites: skinceuticals.com (78 cites), cerave.com (74), theordinary.com (72), laroche-posay.us (66), paulaschoice.com (60). No Amazon. No Sephora. No Reddit. The AI is going directly to product pages. And aad.org (American Academy of Dermatology) at #6 is the one authority citation that surfaces — unique to health-adjacent categories.
Beyond the top 12, 89 additional products show up 1–3 times each — La Roche-Posay Hyalu B5, Medik8 Crystal Retinal, Maelove The Glow Maker, Timeless Vitamin C, TruSkin Vitamin C, and dozens of niche dermatologist-recommended formulations. AI knows the ingredient stack well enough to name specific products for specific concerns.
Methodology
We ran 75 buyer-intent prompts — the specific questions skincare shoppers, dermatologist-recommended-product hunters, ingredient researchers, and DTC founders actually ask — against three AI surfaces: ChatGPT (gpt-5.6), Claude Sonnet 5, and Gemini 3.6 Flash. That’s 221 completed AI answers, parsed by an LLM extractor for: which products were mentioned, in what position within ranked answers, with what sentiment framing, and citing which source URLs.
Buyer-intent prompts spanned the archetypes AI models tend to segment against for face-serum queries:
Mentions were tagged for sentiment (positive / neutral / critical), position in ranked answers, and the specific URLs the AI referenced. Everything below is derived from that dataset.
Scan cycle: week of 2026-08-26. Google AI Overview and Perplexity aren’t in this iteration — the batch pipeline for those was still completing when we generated this cut. We’ll include both in the next cycle. Category defined by Foxish’s benchmark taxonomy: “face-serums.”
The winners
Unlike email marketing, this leaderboard has a clear #1. Below SkinCeuticals, three distinct challenger patterns emerge: brands that spread their share across multiple SKUs, brands that punch above their weight on a single hero product, and brands that ride ingredient-specific queries.
Every AI puts it at #1. Dermatologist-recommended for decades, ingredient stack (15% vitamin C + 1% vitamin E + 0.5% ferulic acid) is the professional benchmark other serums are compared against. AI has canonized this SKU as the default in a way we’ve only otherwise seen with Pacvue in the Amazon PPC category.
CeraVe doesn’t have a hero product, but it has four different serums in the top 12: Resurfacing Retinol (#2), Skin Renewing Vitamin C (#4), Skin Renewing Retinol (#6), and Hydrating Hyaluronic Acid (#12). Total combined mentions exceed the #1 single product. When AI answers a “best drugstore retinol” or “affordable vitamin C” query, it’s picking a specific CeraVe SKU for the specific ask.
A single SKU riding a single ingredient-specific query pattern (“best niacinamide serum,” “serum for acne-prone skin”). At $6, it’s an order of magnitude cheaper than SkinCeuticals. AI treats the two as different answers to different questions rather than direct competitors — the model rarely names them in the same response.
Every AI agrees on SkinCeuticals for #1. What differs is who they pair it with. ChatGPT stays inside the CeraVe portfolio. Claude reaches for TruSkin Vitamin C (a mid-tier DTC alternative) and The Ordinary. Gemini reaches for CeraVe Resurfacing Retinol and Maelove The Glow Maker (a DTC C E Ferulic competitor). If you’re a challenger brand, which AI you’re optimizing for shifts your positioning language.
Face serums are the sharpest example of “AI recommends products, not brands” that we’ve measured. Buyers get answers at the SKU level. If you sell serums and your website is organized around brand-level PDPs instead of ingredient-specific landing pages, you’re fighting AI’s taxonomy instead of using it.
The full leaderboard
Product-level, not brand-level. SkinCeuticals C E Ferulic pulls more than 2× the runner-up. Below that, CeraVe holds four of the top 12 slots and The Ordinary holds two — the portfolio pattern is unmistakable. Sentiment is universally strong (100% positive across the whole top 12), so ranking here is about mention count and ranking position, not sentiment differentiation.
| Rank | Platform | Share of voice | Mentions | Sentiment |
|---|---|---|---|---|
| 1 | SkinCeuticals C E Ferulic | 69 | 99% positive | |
| 2 | CeraVe Resurfacing Retinol Serum | 30 | 100% positive | |
| 3 | The Ordinary Niacinamide 10% + Zinc 1% | 29 | 100% positive | |
| 4 | CeraVe Skin Renewing Vitamin C Serum | 24 | 100% positive | |
| 5 | Timeless 20% Vitamin C + E Ferulic Acid | 19 | 100% positive | |
| 6 | CeraVe Skin Renewing Retinol Serum | 17 | 100% positive | |
| 7 | TruSkin Vitamin C Serum | 14 | 100% positive | |
| 8 | The Ordinary Hyaluronic Acid 2% + B5 | 14 | 100% positive | |
| 9 | Maelove The Glow Maker | 13 | 100% positive | |
| 10 | La Roche-Posay Hyalu B5 | 13 | 100% positive | |
| 11 | Medik8 Crystal Retinal | 13 | 100% positive | |
| 12 | CeraVe Hydrating Hyaluronic Acid Serum | 10 | 100% positive |
Also in the long tail: 131 additional products show up in the dataset, with 89 of them cited 1–3 times each — SkinCeuticals Phloretin CF, Drunk Elephant C-Firma, Kiehl’s Powerful Wrinkle Reducing Cream, Naturium Niacinamide, Good Molecules Discoloration Correcting, Beauty of Joseon Glow Serum, Bio Jouvance Vitamin C, and dozens of niche or clinical-grade formulations.
Per-model breakdown
Every AI puts SkinCeuticals C E Ferulic at #1. What differs is who they pair it with. ChatGPT sticks with CeraVe SKUs; Claude reaches for TruSkin and The Ordinary; Gemini brings in Maelove. Which AI you optimize for shifts your positioning.
The consensus at #1 is striking. In email marketing, Amazon PPC, and hospital categories the top slot flips between models. In serums, every AI agrees SkinCeuticals C E Ferulic is the answer. The professional-dermatology reputation of C E Ferulic has permeated all three training corpora identically. Google AI Overview and Perplexity data for this category will land in the next cycle.
The citation graph
Every brand-owned .com is a top citation source — but the more interesting story is what’s not here. No Amazon. No Sephora. No Reddit. No dermatologist review blogs. AI is going directly to brand product pages and one authority source (aad.org) for its recommendations.
Two things worth flagging. First: Paula’s Choice ranks #5 by citations (60) despite being outside the top 12 by mention count — their ingredient-education content is over-indexing them relative to product-level recommendations. Second: aad.org (American Academy of Dermatology) at #6 is the only non-brand authority in the top 8. That’s a health-adjacent signal AI models weight uniquely — and something skincare brands can influence by getting cited in dermatologist-authored AAD content.
Reading the tea leaves
Add specifics. “Best face serum” gets you SkinCeuticals C E Ferulic ($185) and CeraVe from every AI. Add “under $30 for hyperpigmentation” and you get The Ordinary Niacinamide 10% ($6) or CeraVe Skin Renewing Vitamin C ($20). AI is doing sophisticated ingredient- and price-level segmentation — but only if you give it the grip to work from. Vague queries default to the ingredient benchmark; specific queries surface real fit.
You own the category default. The 2.3× lead over the runner-up is durable because it’s ingredient-anchored: AI cites C E Ferulic’s specific formulation as the professional benchmark. Your risk is the price gap. Timeless 20% Vitamin C + E Ferulic (#5, a direct dupe at ~1/10 the price) is already appearing in the same responses on ingredient-specific queries. If AI starts pairing you with a dupe on every ingredient prompt, the “professional benchmark” framing weakens.
Your strategy is working. Multiple SKUs in the top 12 means AI is picking your specific product for the specific query — not sending buyers to a generic brand landing page. Keep publishing dedicated product pages for each hero ingredient combination. Your risk is SKU cannibalization: if AI can’t distinguish Skin Renewing Vitamin C from Resurfacing Retinol on ingredient content alone, it’ll compress you back to one SKU per query.
Two paths our data supports. One: build an ingredient hero — The Ordinary at #3 and Timeless at #5 both won that way. Pick one ingredient combination (niacinamide + zinc, 20% vitamin C, retinal) and own it at a specific price point. Two: get cited by aad.org. It’s the only non-brand authority in the citation graph, and dermatologists writing there are already surfacing named products for AI to pick up. Reach the dermatologists writing there, and their recommendations start showing up in AI answers with your name attached.
The playbook
Adapted for the face-serum category specifically. Ordered by cost and effort.
AI recommends products, not brands. CeraVe wins by having four different serum SKUs each with their own dedicated page. If your brand has multiple serums but only a “shop serums” category page, AI has nothing to distinguish them from your Vitamin C from your Retinol. Every SKU needs its own comprehensive PDP with ingredient percentages, use cases, skin types, and price positioning.
Expected lift: additional SKUs into the top 30 mentions. Cost: 2–3 weeks per SKU page.
The Ordinary owns niacinamide 10% + zinc 1%. Timeless owns 20% Vitamin C + E ferulic (as a Skinceuticals dupe). Maelove owns the Glow Maker Vit C stack. Pick one specific ingredient combination at a specific price tier and publish comprehensive content around it — not just a PDP, but ingredient-education blog posts, before/after content, and a comparison against the ingredient benchmark (SkinCeuticals). This is how AI starts naming you when the query is ingredient-shaped.
Expected lift: entry into the top 15 for that ingredient query set. Cost: 4–6 weeks of content build.
aad.org is the one non-brand authority in the top 8 citations (44 mentions). It’s the American Academy of Dermatology — and dermatologists who publish there are already surfacing specific product recommendations that AI picks up. Reach the dermatologists who write for AAD (or contribute their content); reach the Board-certified dermatologists whose bios AAD cites; get quoted with your brand named. This is health-adjacent SEO and the highest-leverage single citation opportunity in the category.
Expected lift: 15–30 authority-tagged citations over 6 months. Cost: PR + dermatologist relationships, ~3 months to seed.
Every AI names SkinCeuticals C E Ferulic first. Timeless made #5 by explicitly comparing themselves to it at 1/10 the price. That comparison shape is what AI cites when a “affordable vitamin C serum” query lands. Publish a comparison against the category benchmark for each of your hero SKUs — not marketing puffery, ingredient-percentage side-by-side plus price. AI models discount one-sided pages but reward accurate comparison content.
Expected lift: closes ingredient + budget-shaped queries. Cost: ~1 week of writing per comparison.
Face serums are the sharpest example we’ve seen of AI recommending products, not brands. Watching “CeraVe’s share of voice” hides that CeraVe Skin Renewing Vitamin C is climbing while CeraVe Hydrating Hyaluronic Acid is slipping. Track each SKU separately — that’s where the strategic decisions live.
Expected lift: attribution clarity — you know which fix moved which query. Cost: 15 minutes a week.
About this report
Foxish is an AI visibility monitoring platform. We track how ChatGPT, Claude, Gemini, Perplexity, and Google AI Overview answer buyer-intent questions across categories — product mentions, sentiment, share of voice, and the citation graph behind every answer. This report is a snapshot of what Foxish sees for face serums. If you’d like the same view for your specific competitor set (or a different DTC skincare vertical), the underlying platform runs your custom prompt set on a schedule and generates a live version of every chart in this report.
Data collected via Foxish’s benchmark pipeline: batch API calls to OpenAI, Anthropic, and Google (Gemini). Google AI Overview and Perplexity data for this category will land in the next cycle — both batches were still completing at report generation. Category defined by Foxish’s benchmark taxonomy: “face-serums.” Report generated August 27, 2026.
Get your segment
The overall face-serum leaderboard is directionally correct but the concern or ingredient or price tier your customers are asking about is what matters most. Send us your brand’s SKUs and your top 3–5 competitors — we’ll run Foxish’s benchmark pipeline against your specific prompt set for 7 days and generate a report in this exact shape with your SKUs’ positions included.
Or try Foxish yourself at foxish.ai/signup — the same pipeline that produced this report, watching your brand across the surfaces and segments you care about.