Index No. 01 · Facial Sunscreen · United States · September 2026

Who does AI recommend when you ask for a sunscreen?

We asked five AI engines 49 real shopping questions, three times each, and counted who they named. One clear leader, one absent tier, and five different personalities.

By Vivian Qu · Published 27 September 2026

735answers analysed

5engines

49prompts × 3 runs

78brands named

Findings

Six things the answers showed

EltaMD is the consensus AI sunscreen.

Named in 42% of all answers and named first in 20%. Every one of the five engines names it first more often than any other brand. It is a dermatologist-office brand with little consumer advertising. La Roche-Posay is named almost as often (39%) but rarely first (7%).

Dermatologist and pharmacy brands take 40% of all mentions.

Prestige 16%, mass 15%, Korean beauty 12%. Independent and direct-to-consumer brands get 6%. Australian brands, the most talked-about SPF makers on social media, get under 1%. Ultra Violette appears in five answers out of 735.

Legacy prestige is absent.

Estée Lauder: not named once. Chanel and Lancôme: twice each. La Mer appears ten times, mostly as a "not worth it" aside. Heritage carries no weight in an AI answer, because an AI answer is built from what has been written about you.

Each engine has a personality.

Gemini is the Korean-beauty engine: Beauty of Joseon in 26% of its answers, SKIN1004 in 20%. ChatGPT is the drugstore engine: Neutrogena 30%, Australian Gold 23%. Perplexity names the fewest brands. Grok leans on CeraVe (45%) and Eucerin, which the others largely ignore. A brand "recommended by AI" is recommended by one AI at a time.

The same question does not get the same answer.

Asked three times, ChatGPT names the same brand first only 41% of the time. Perplexity, 78%. Any index that runs each prompt once is reporting noise. This one runs each prompt three times and says so.

On safety questions, brands vanish.

Across 120 answers to prompts like "Is sunscreen toxic?", the engines named a brand twice. They cite the WHO, the American Academy of Dermatology and the FDA, and they correct the myth. Authority sources own the safety conversation. Brands are not in it.

The map

Named often, named first

Two numbers per brand: how often it is named at all, and, when it is named, how often it comes first. Top right is where a brand wants to be.

Mention rate against first-pick conversion, top 30 brands

Mention rate is the share of 735 answers naming the brand. Conversion is the share of those mentions where the brand was named before any other. Hover a point for its numbers.

The full data

Get the complete leaderboard

The findings above are public. The working data is for brand and marketing teams, and we ask for a company email to send it.

Top 20 leaderboard with mention and first-mention rates
Challenger view: independent, kids', European and Australian brands
Engine-by-engine breakdown for ChatGPT, Claude, Gemini, Perplexity and Grok
Share of voice by question type
The citation map: 25 most-cited domains, by type
Downloadable CSV files

Method

How this edition was made

Prompts. 49 English prompts across six clusters: texture and wear, sensitive skin and barrier, longevity, Korean and Japanese sunscreen, kids and teens (written from a parent's side), and myths and safety. No prompt names a brand.
Engines. ChatGPT (gpt-5), Claude (claude-sonnet-5), Gemini (gemini-3.5-flash), Perplexity (sonar) and Grok (grok-4.7), through each provider's API with web search enabled where the API supports it. Each prompt was run three times per engine in a fresh session, US locale, on 26 September 2026.
Counting. Every answer was matched against a list of 102 brands with their product lines and abbreviations. A second pass flagged capitalised names not on the list; those were reviewed by hand and added when they were brands.
Measures. Mention rate: the share of answers naming the brand. First-mention rate: the share of answers naming it before any other brand. Share of voice: a brand's mentions divided by all brand mentions in a cluster. Stability: whether the first pick agreed across the three runs.
Sponsored placements were flagged and excluded. None appeared in API responses during this run.
Limits. API answers can differ from what a signed-in consumer sees in an app. ChatGPT returned no web citations on 78 of 147 answers, which suggests it often answered from memory. Google AI Mode has no API and is not included. The Index reports what the engines said. It does not judge the brands.

Questions

Straight answers

Which sunscreen brand does AI recommend most?

EltaMD. Across 735 answers from ChatGPT, Claude, Gemini, Perplexity and Grok, EltaMD was named in 42% of answers and named first in 20%. La Roche-Posay was named almost as often, in 39% of answers, but named first in only 7%.

Do ChatGPT, Gemini, Claude, Perplexity and Grok recommend the same sunscreens?

Only at the top. All five name EltaMD first more than any other brand. Below that they differ: Gemini names Korean brands such as Beauty of Joseon far more often, ChatGPT names drugstore brands such as Neutrogena, Perplexity names the fewest brands, and Grok leans on CeraVe and Eucerin.

Does AI give the same sunscreen answer every time?

No. Asked the same question three times, ChatGPT named the same brand first only 41% of the time. Perplexity was the most consistent at 78%. This is why the VIBEAU Index runs every prompt three times per engine.

How visible are independent sunscreen brands in AI answers?

Not very. Independent and direct-to-consumer brands received 6% of all brand mentions. Dermatologist and pharmacy brands received 40%. Australian brands, including Ultra Violette, received under 1%.

How was the VIBEAU Index measured?

49 English shopping prompts, none naming a brand, each run three times on ChatGPT, Claude, Gemini, Perplexity and Grok through their APIs on 26 September 2026, with web search on where supported. Every answer was matched against a list of 102 brands. The full method is published on this page.