ChatGPT shopping results are concentrated in a few retailers, and the study behind that finding observed no consistent minimum review count among the products ChatGPT shows. In a 5W study of 34,637 U.S. ChatGPT prompt runs between August 3 and November 1, 2026, four retailers held 85.2% of cosmetics shopping results, and 13 of the 50 most visible cosmetics products had fewer than 500 reviews. Here is what I think that means for PR and marketing teams.
Disclosure: 5W conducted this research, and I am Founder and Chairman of 5W. I am also Publisher of Everything-PR, which is published by Everything-PR News LLC and owned by a shared interest group that includes me.
In this article
What did 5W measure in its ChatGPT studies?
5W measured which retailers, brands and products appear when ChatGPT answers shopping questions, using 1,114 prompts across cosmetics, skincare and makeup, baby gear and furniture, and golf equipment. My firm ran the prompts about 31 times each on average in the United States and analyzed the results with Profound’s tools. The first study, the 5W ChatGPT Shopping Index, covers retailers and brands. The second, the 5W ChatGPT Review Study, covers star ratings and review counts.
ChatGPT showed a shopping result, meaning a carousel or product-comparison panel, in 56.7% of the 34,637 runs. The rate was highest in cosmetics at 66.8% and lowest in golf equipment at 43.4%. I find that encouraging for brands, because more than half of the shopping questions we tested gave a company a place to appear.
The summary below covers the findings I think matter most for communications teams.
- Four retailers hold 85.2% of ChatGPT cosmetics shopping results, and the top three retailers hold 59.3% to 79.5% in every category.
- Questions about a product type trigger shopping results far more often than questions that name a brand.
- Three of the 50 most visible cosmetics products have fewer than 100 reviews.
- No single product appears in more than 12.1% of shopping answers in any category.
A handful of retailers fill most of ChatGPT’s shopping results in each category the study tested. In cosmetics, Ulta Beauty and ulta.com together hold 48.4% of merchant share, ahead of Walmart at 15.1%, Sephora at 11.1% and Target at 10.6%. Merchant share is a retailer’s slice of all shopping results in a category.
Table 1. The three leading retailers in each category by merchant share. Source: 5W’s analysis using Profound’s tools, U.S. ChatGPT prompt runs, August 3 to November 1, 2026. Listings under two names are combined.| Category | First | Second | Third | Top three combined |
|---|
| Cosmetics | Ulta 48.4% | Walmart 15.1% | Sephora 11.1% | 74.6% |
| Skincare and makeup | Ulta Beauty 42.5% | Sephora 24.4% | Target 12.5% | 79.5% |
| Baby gear and furniture | Target 27.6% | Nordstrom 22.3% | Walmart 15.0% | 64.9% |
| Golf equipment | SCHEELS 24.4% | DICK’S Sporting Goods 24.0% | Walmart 10.8% | 59.3% |
Two results surprised me. Nordstrom ranks second in baby gear, ahead of Walmart, and SCHEELS edges out DICK’S Sporting Goods in golf by 0.4 points. For context on how Ulta approaches its range and rewards program, see our profile of Ulta Beauty’s marketing strategy, and for Nordstrom’s positioning, see our Nordstrom PR strategy profile.
I would add one caution. The 48.4% figure is Ulta’s merchant share inside ChatGPT’s cosmetics shopping results for the prompts 5W tracks. It is not 48.4% of cosmetics searches or sales.
Why is visibility different from merchant share?
Visibility and merchant share measure different things, and the gap between them tells a brand how it appears. Visibility is the share of shopping answers that mention a retailer or brand at least once, while merchant share is the slice of all results. Sephora appears in 66.0% of skincare and makeup shopping answers but holds 24.4% of merchant share, because ChatGPT usually lists several retailers for one product.
I read that as an opportunity rather than a problem. A retailer or brand that already appears in two thirds of answers has strong presence, and moving from one option among several to the main listing is a concrete goal that a team can measure.
The same pattern shows up for brands. Callaway Golf appears in 70.7% of golf shopping answers, e.l.f. cosmetics leads cosmetics brand visibility at 32.3%, and Stokke leads baby gear at 37.6%. For a different measurement of the same category, see our baby brand AI visibility analysis.
Do products need thousands of reviews in ChatGPT?
No consistent minimum review count was observed among the 50 most visible cosmetics products in ChatGPT shopping results. Three of those products have fewer than 100 reviews, 13 have fewer than 500, and the median is about 2,000. The lowest star rating among them is 3.8.
Table 2. Cosmetics products with the fewest reviews among the 50 most visible. Source: 5W’s analysis using Profound’s tools, U.S. ChatGPT prompt runs, August 3 to November 1, 2026. Visibility is the share of shopping answers that feature the product.| Product | Reviews | Star rating | Visibility | Rank among the 50 |
|---|
| Juvia’s Place I Am My Skin Hydra-Tint Foundation Stick | 19 | 4.5 | 1.4% | 35th |
| Essence Radiant Light Face Palette | 21 | 4.2 | 1.9% | 19th |
| e.l.f. Cosmetics Soft Glam Cream Bronzer | 27 | 4.4 | 1.8% | 21st |
| Essence Foundation Stick | 147 | 3.8 | 2.5% | 8th |
I want to be careful about what this shows. The rank correlation between reviews and visibility among these 50 products is 0.13, which is very weak, and the study did not compare them with products ChatGPT rarely shows. It cannot say that reviews have no effect, and it cannot say how many reviews a product needs to appear at all. What it does show is that a young product with a small review base can still reach the most visible group.
Ratings and review counts were also recorded unevenly. Profound recorded them for all 50 top cosmetics products, but for only 3 of the top 50 golf products and for none of the top 50 baby gear or skincare products. Profound does not say which retailer’s reviews the counts reflect, so I treat them as the numbers attached to ChatGPT’s product listing.
ChatGPT shopping gives communications teams four practical jobs, and each follows from the data. I would start with the questions a brand is able to influence.
First, treat retailer relationships as part of the communications plan. Because the top three retailers hold 59.3% to 79.5% of results in each category, a brand that is not carried by those retailers has fewer listings that ChatGPT can show, which is why I count retail buyer outreach and retailer editorial placements as AI visibility work as well as sales work.
Second, build content around product-type questions. Setting powder prompts triggered shopping results 81.4% of the time, compared with 30.7% for cosmetics brand and retailer prompts. A brand that publishes clear product pages, comparisons and guides for the questions shoppers ask gives ChatGPT more relevant material to match.
Third, check how the brand’s name appears. ChatGPT listed Ulta Beauty and ulta.com as separate merchants, and e.l.f. appears as both “e.l.f. cosmetics” and “e.l.f.” Visibility cannot be added across name variants, so a brand’s real footprint can be larger than any single row suggests. A quarterly check of how each product and brand name is listed is inexpensive and revealing.
Fourth, collect reviews without waiting for a target. The Review Study found products in the top 50 with as few as 19 reviews, so the review count is one input among many and not an entry requirement. Those same top products still hold strong ratings, with 49 of 50 at 4.0 or higher, so I would protect the rating while the count grows.
I think these jobs also suit PR teams unusually well. Earned coverage, retailer relationships and consistent naming are work that communications professionals already do, and now there is a public scoreboard to measure it against. For background on how AI engines choose what to show, our explainer on generative engine optimization and analysis of which sources AI engines cite most are good companions to this research.
What are the limits of this research?
The research describes where ChatGPT sends shoppers in a defined set of prompts and does not test why. 5W tracks the 1,114 prompts in the study, so they are not a random sample of every shopping question, and the results cover ChatGPT in the United States between August 3 and November 1, 2026. Rankings describe that window and will move.
The product analysis examines only the 50 most visible products in each category, and the review statistics apply to cosmetics only. Merchant shares for retailers listed under two names were combined where both names appeared in the top 50 listings, so combined shares are a floor. The studies measure visibility in ChatGPT, not sales, price or checkout.
I would rather publish the limits next to the findings, because they tell readers which questions to test next. For a related example of testing one AI engine across a category, see my analysis of hair tool brands.
Where can I read the full studies?
The full tables, definitions and method for both studies are on 5wpr.com. The 5W ChatGPT Shopping Index includes retailer and brand rankings for all four categories, trigger rates by topic and the complete methodology. The 5W ChatGPT Review Study includes the ten most visible cosmetics products, the review and rating statistics and a chart of reviews against visibility.
My next step for any brand reading this is simple. Pick one category, list your five most important products, and check which retailers carry them, how each product name is listed and how many reviews each has.
