Beauty & Skincare AI Visibility Study 2026
Which beauty and skincare brands AI assistants name and recommend when shoppers ask what to buy. A point-in-time study of thousands of AI answers.
Skincare discovery has moved into AI answers. This study measures which brands assistants put in front of shoppers, how often those mentions are active recommendations, and the source patterns behind them.
What the answers showed
Citations analysed 4,903 AI answers to 500 buyer questions across 2 AI assistants, and recorded 2,101 operating brands being named in beauty and skincare.
CeraVe was the most visible brand in the sample, named in 27.2% of analysed answers, with 97.6% of its appearances presented as an active recommendation. La Roche-Posay followed at 26%.
This is a point-in-time research snapshot, not a live ranking.
Scale of the sample
Top 25 most visible brands
| # | Brand | Sampled visibility | Recommendation rate | Avg position | Questions covered | Confidence |
|---|---|---|---|---|---|---|
| 1 | CeraVe | 27.2% | 97.6% | 2.01 | 208 | Strong |
| 2 | La Roche-Posay | 26% | 98.4% | 2.47 | 217 | Strong |
| 3 | Neutrogena | 22.8% | 96.1% | 3.59 | 209 | Strong |
| 4 | Deciem | 10.9% | 97.6% | 3.3 | 132 | Strong |
| 5 | Paula’s Choice | 10.7% | 98.5% | 2.97 | 123 | Strong |
| 6 | SkinCeuticals | 9.3% | 93.9% | 2.54 | 111 | Strong |
| 7 | Maybelline | 7.9% | 97.4% | 2.96 | 65 | Strong |
| 8 | Cetaphil | 7.6% | 96.8% | 3.26 | 99 | Strong |
| 9 | COSRX | 7.2% | 98% | 2.72 | 69 | Strong |
| 10 | Kiehl’s | 7.1% | 95.4% | 3.94 | 84 | Strong |
| 11 | Vanicream | 6.2% | 99.7% | 3.16 | 72 | Strong |
| 12 | e.l.f. | 6.2% | 98% | 3.7 | 69 | Strong |
| 13 | The Ordinary | 6% | 100% | 3.28 | 112 | Strong |
| 14 | Clinique | 5.7% | 96.1% | 3.36 | 82 | Strong |
| 15 | Drunk Elephant | 5.5% | 97.4% | 3.66 | 72 | Strong |
| 16 | L'Oréal Paris | 5.2% | 100% | 2.81 | 68 | Strong |
| 17 | Ipsy | 4.9% | 92.1% | 1.9 | 38 | Strong |
| 18 | Dermstore | 4.5% | 94.6% | 3.73 | 50 | Strong |
| 19 | Laneige | 4.4% | 98.6% | 3.22 | 43 | Strong |
| 20 | Olay | 4.4% | 96.7% | 4.28 | 57 | Strong |
| 21 | Aveeno | 4.1% | 99% | 4.4 | 56 | Strong |
| 22 | EltaMD | 4% | 96.4% | 2.58 | 41 | Strong |
| 23 | La Mer | 3.9% | 91.6% | 2.14 | 40 | Strong |
| 24 | Olaplex | 3.8% | 96.3% | 1.63 | 28 | Strong |
| 25 | Birchbox | 3.8% | 93% | 3.03 | 38 | Strong |
How the assistants differ
Answers analysed per assistant
- gemini2,491 (50.8%)
- chatgpt2,412 (49.2%)
Leading brands by assistant
- CeraVechatgpt: 31.5%gemini: 23%
- La Roche-Posaygemini: 26.8%chatgpt: 25.2%
- Neutrogenachatgpt: 25.3%gemini: 20.3%
- Deciemchatgpt: 14.2%gemini: 7.6%
- Paula’s Choicechatgpt: 13.6%gemini: 7.9%
- SkinCeuticalsgemini: 13.9%chatgpt: 4.6%
- Maybellinechatgpt: 9.2%gemini: 6.6%
- Cetaphilchatgpt: 8.9%gemini: 6.3%
Each percentage is measured against the answers analysed on that assistant.
The kinds of sites the answers lean on
Aggregated across the whole study. Individual pages and answer-level citations are not published; only domains observed at least three times appear here.
| Domain | Share of observed source uses | Type |
|---|---|---|
| allure.com | 16% | unknown |
| aad.org | 4.7% | unknown |
| goodhousekeeping.com | 4.1% | unknown |
| cosmopolitan.com | 3.6% | unknown |
| byrdie.com | 3.3% | unknown |
| cerave.com | 3.1% | unknown |
| sephora.com | 3% | unknown |
| healthline.com | 2.6% | unknown |
| vogue.com | 2.4% | unknown |
| laroche-posay.us | 2.4% | unknown |
| paulaschoice.com | 2.1% | unknown |
| ulta.com | 1.9% | unknown |
| neutrogena.com | 1.7% | unknown |
| nytimes.com | 1.7% | earned |
| reddit.com | 1.7% | ugc |
| ewg.org | 1.5% | unknown |
| amazon.com | 1.5% | unknown |
| gq.com | 1.3% | unknown |
| deciem.com | 1.2% | unknown |
| menshealth.com | 1.1% | unknown |
How to read these numbers
Sampled AI visibility is the share of the analysed AI answers in this study in which a brand was named. Recommendation rate is the share of those appearances where the answer actively recommended the brand rather than mentioning it in passing. Per-engine visibility is measured against the answers analysed on that engine only. Confidence reflects sample size: Directional up to 20 qualifying observations, Moderate from 21 to 100, Strong above 100.
Directories, ranking bodies, regulators and publishers are classified separately and excluded from the operating-brand leaderboard. This is a point-in-time research snapshot, not a live ranking.
