Beauty Brands Index 2.0: L'Oréal Paris, Dior Lead AI Answers

L'Oréal Paris was named in 100 of 288 four-engine answers (34.7%). Dior Beauty was named in 74 of 288 (25.7%), and the two intervals overlap. Chanel Beauty was named first most often, in 36 of 288. Google AI Overviews appeared on 57 of 72 searches, and e.l.f. was named most often within them, in 16 of 57.

Appearance rate across 288 four-engine answers. The bar is the 95% confidence interval; overlapping bars cannot be told apart at this sample size.

The full list

Own-domain citation is the share of 288 grounded answers that cited the entity's own domains (subdomains included). Citation share is of all 2,421 citations.

The top eight, side by side

All four engines returned grounded answers to all 72 prompts, but they named different brands. ChatGPT named retailers much more often than the others: Ulta Beauty in 30 of 72 answers (41.7%) and Sephora in 22 of 72 (30.6%). It named e.l.f. in only 5 of 72 (6.9%). Perplexity named e.l.f. in 18 of 72 (25%). Gemini named L'Oréal Paris in 35 of 72 (48.6%).

ChatGPT

72 asked · 72 returned · 72 grounded

Claude

72 asked · 72 returned · 72 grounded

Gemini

72 asked · 72 returned · 72 grounded

Perplexity

72 asked · 72 returned · 72 grounded

Google AI Overviews were captured through 72 U.S. desktop searches and scored only where an overview appeared, so five-engine figures use 288 + 57 = 345 answers.

Google AI Overviews appeared on 57 of 72 searches (79.2%, CI 68.4–86.9%). The other 15 searches showed no overview. Within the 57 overviews, e.l.f. was named in 16 (28.1%), Sephora in 15 (26.3%), Maybelline in 13 (22.8%) and L'Oréal Paris in 12 (21.1%). Chanel Beauty was named in 6 of 57 (10.5%) and Estée Lauder in 0 of 57. allure.com was the most-cited source, with 29 citations (18.4%) and an appearance in 26 of 57 overviews (45.6%). youtube.com was next with 20 citations, then byrdie.com with 14.

Who each question surfaces first

Six questions from the set. Each tile uses that question's own four-engine answers as the denominator.

The four engines produced 2,421 citations in total. whowhatwear.com was cited most often, with 134 citations (5.5%) across 121 answers. Next came allure.com (96), marieclaire.com (68), aol.com (68), youtube.com (66) and ulta.com (65). reddit.com was cited 41 times. Of 2,421 four-engine citations, general news and business press supplied 1,109 (45.8%), company and brand sites 522 (21.6%) and review and comparison publishers 439 (18.1%).

Four AI engines · share of 288 grounded answers

Google AI Overviews · share of 57 overviews

Citation share by source type

Share of all citations by the type of site cited. Source types were reclassified on October 29, 2026. 'Company / brand sites' now covers only the official sites of companies, brands and schools; publishers, directories, trade bodies and other sites have their own types, and sites that could not be placed are 'Other / unclassified'.

How can brands improve their AI visibility?

These steps follow from the measured results above. They describe what the engines cited in this study and are not a guarantee of results.

How this study was measured

24 questions × 3 runs each × 4 AI engines = 288 answers + 72 Google searches · 57 AI Overviews

24 questions, each asked 3 times on each engine, measured October 26, 2026. Market: United States, English. The destination site was not mentioned to any engine. Entities count as named when they appear in the answer text or the extracted entity list; products, imprints and research arms roll up to their parent.

Own-domain citations count a domain plus all its subdomains, with the same rule in every table.

Definitions

• Appearance rate: answers naming the entity ÷ answers returned, with Wilson 95% intervals.
• First mention: average order in which the entity first appears (1 = first).
• Own-domain citation: grounded answers citing the entity's domains ÷ grounded answers.
• Citation share: citations to the entity's domains ÷ all citations.

Limits

• Measured on one day, October 26, 2026. Answers came from each engine's API (model IDs above); the consumer apps can answer differently.
• Only 28 tracked brands were scored. Untracked names that appeared often were not scored, including Maison Francis Kurkdjian (28), Jo Malone London (22), Creed (20), Tom Ford (19) and Neutrogena (18).
• Names are counted when they appear in an answer's text, so longer forms such as legal company names (for example "Meta Platforms, Inc." or "Ralph Lauren Corporation") count toward the brand.
• The source-type classification puts several magazine sites, including whowhatwear.com and marieclaire.com, under 'Company / brand sites'. That inflates the share for that type.
• AI Overview figures are based on only 57 overviews, so the estimates for individual brands are less precise than the four-engine figures.
• One question, "Sephora vs Ulta: where should I buy beauty products?", names two retailers, so their counts include up to 12 answers to it. The leading brands are the same without it.
• Intervals treat each answer as independent, so they understate uncertainty for the 24 repeated questions.
• Source types were reclassified on October 29, 2026: 'Company / brand sites' now means only official company, brand or school sites. Name counts did not change.

Appendix: The 24 questions

Questions marked "Core question" are the original study questions, asked word for word. The others are added neutral questions.

About 5W

5W is an independent PR, creative and GEO agency founded in 2003, and we get you in the chatbox. All of our work is done in house, including media relations, site building and content. Learn more about our GEO practice or write to info@5wpr.com.