In AI answers, Sephora is a shopping and returns destination, not a name tied to product results

This snapshot, measured on 2026-09-29, covers 250 benchmark questions across ChatGPT, Claude, Gemini, Perplexity and Grok, none of which named a brand. Sephora appeared in 321 of 1,250 answers, the most of any beauty brand in the set, just ahead of Ulta Beauty (301) and L'Oréal (300). Grok named it most often (91 of 250 answers). Perplexity named it least (50 of 250).

The context is retail. Visibility by question intent was highest for shopping experience (76%), service (63%) and employer questions (45%). It was lowest for product efficacy (2%), ingredient transparency (5%) and reliability (8%). The brand the answers describe is a place to buy, return and get matched, not a name credited for what is in the bottle.

The stored answers show the pattern. In the returns and guarantee questions, Claude's answer on satisfaction guarantees pointed a first-time luxury skincare buyer to Sephora and Dermstore for their return windows. Grok's hair care refund answer described Sephora as allowing refunds on new or gently used products within 30 days, next to Ulta and Nordstrom. In the app question, ChatGPT put Sephora and Ulta at the top and tied Sephora's app to its Beauty Insider profile and an AI-enhanced experience in ChatGPT. Perplexity's shade-matching answer recommended Sephora Color iQ as a starting point for matching across brands.

Not every framing was favorable to a retailer's brand. Perplexity's employer answer said employee reviews flag concerns at Sephora around sales pressure and management culture, and it cautioned that these are warning signs, not verdicts. Perplexity's data breach answer listed a 2019 Sephora breach affecting online customers in several Asia-Pacific markets. In the largest-retailer question, Grok named Ulta Beauty the largest beauty retailer in America and called Sephora a prestige-focused competitor that trails on US scale.

The credited-for tags were thin. Customer Experience (5) and Not Substantive (5) led, with single mentions for Safety & Compliance, Labour & Workplace, Financial Performance and Price & Value.

PeakMetrics' news index recorded 5,755 mentions of Sephora over the last 30 days, 1,126 of them with the brand in the headline. That is the highest coverage among the beauty peers in this set, ahead of Ulta Beauty (5,135) and L'Oréal (4,250).

The recent headlines are about retail moves and store experience: inclusive 'Quiet Hours' across Europe, a launch on TikTok Shop through 'Sephora Drop Shop', a Korean beauty editor's Olive Young x Sephora picks, and Saie's UK store debut at Sephora. One retail-trade headline pairs Kohl's turning to Babies Are Us with the phrase 'As Sephora Falters', which is a different note from the launch coverage.

Press and answers largely agree on the subject: both are about Sephora as a retail environment. The difference is in what the answers do with it. The press covers launches, new channels and store initiatives. The answers turn Sephora into practical advice: return windows, shade matching, app personalization.

The answers also carry material the recent headlines do not show. The 2019 breach and the employee-review concerns appeared in Perplexity's answers, and the Ulta comparison on scale appeared in Grok's. Neither appears in the five recent headlines supplied. The volume of coverage is about the same as Ulta's, and so is the volume of appearances in answers, so heavy coverage does not by itself explain what the assistants say.

For a comms team, the snapshot points to three checks. First, the return and guarantee story is carried by the answers, and the windows they quote differ. Claude's guarantee answer said 60 days, while its returns answer and Grok's hair care answer said 30 days. Confirm which policy language the assistants are drawing from.

Second, the low visibility on efficacy and ingredient questions means those conversations are being held without Sephora in them. Whether that matters depends on how much the brand wants to be part of them.

Third, older items such as the 2019 breach and employee-review themes surface in specific answers, and Perplexity's answer to the employer question is one example. These are single snapshots of live answers that vary between runs, so treat them as prompts to monitor rather than a fixed picture.