{"id":111348,"date":"2026-10-09T13:05:52","date_gmt":"2026-10-09T11:05:52","guid":{"rendered":"https:\/\/ga.agency\/?post_type=blog&p=111348"},"modified":"2026-10-09T13:05:57","modified_gmt":"2026-10-09T11:05:57","slug":"chatgpt-try-on-ecommerce-brands-2","status":"publish","type":"blog","link":"https:\/\/ga.agency\/en\/blog\/chatgpt-try-on-ecommerce-brands-2\/","title":{"rendered":"ChatGPT’s Try On button moves the fit decision before the click"},"content":{"rendered":"\n
<\/div>\n\n\n\n

OpenAI just added a Try on<\/em><\/strong> button to clothing and accessory listings in ChatGPT’s shopping results. A shopper selects it, takes or uploads a selfie and ChatGPT generates an image of them wearing the item. The same release lets shoppers save products to Favorites and folders in their ChatGPT Library.<\/p>\n\n\n\n

Reading this as just a visual feature would miss the commercial point. The first judgement about how a garment looks on a shopper, and whether it is worth a closer look, now happens before the click. What informs that judgement is the brand’s product data.<\/p>\n\n\n\n

<\/div>\n\n\n\n

<\/span>What actually changed<\/strong><\/span><\/h2>\n\n\n\n

The Try on<\/em> button appears on clothing and accessory listings within ChatGPT’s shopping results, the image is generated by ChatGPT Images and the shopper’s reference photo is saved so it can be reused for later try-ons. Both Try on<\/em> and Favorites<\/em> are available on mobile and web and TechCrunch reports the launch as global.<\/p>\n\n\n\n

Two details are worth reflecting on: first, the button only appears on products that ChatGPT surfaces. OpenAI states that product results are selected by ChatGPT and are not ads and that selection draws on structured metadata such as price and description from first-party and third-party providers. Second, shoppers can also upload a screenshot of an item found elsewhere and ask ChatGPT to try it on. A product can therefore be evaluated inside ChatGPT whether or not the brand has made it available there.<\/p>\n\n\n\n

OpenAI is clear about the limits, which are quite intuitive. Its help centre says try-on images may not represent the product or the shopper exactly, do not guarantee fit or size and that shoppers should check the merchant’s measurements, product details and return policy before buying.<\/p>\n\n\n\n

On the merchant side, there are three routes in. Shopify merchants are already included through Shopify Catalog, with no additional work required. Other merchants can apply to OpenAI for a direct product feed. Salesforce B2C Commerce now documents a native OpenAI product feed integration. In most cases the purchase itself still completes on the merchant’s site.<\/p>\n\n\n\n

<\/div>\n\n\n\n

<\/span>Why it belongs on the agenda<\/strong><\/span><\/h2>\n\n\n\n
<\/div>\n\n\n\n

1. The fit decision moves upstream and the image cannot carry it alone<\/strong><\/h3>\n\n\n\n

A generated image can show colour, silhouette and roughly how a piece drapes. It cannot say that a jacket runs small across the shoulders or that a pair of trousers has no stretch. OpenAI’s own caveat sends shoppers back to the product details for exactly this reason.<\/p>\n\n\n\n

That makes fit information written in plain language more important. The model’s height and the size they are wearing, whether an item runs true, small or large, fabric weight and stretch and the one measurement that matters most for the item: these lines now inform a decision taken earlier in the journey, on a surface the brand does not control. Fit copy is the part of the listing that a generated image cannot replace.<\/p>\n\n\n\n

<\/div>\n\n\n\n

2. Product data quality decides who gets the button<\/strong><\/h3>\n\n\n\n

Only products that appear in ChatGPT’s shopping results carry the Try on<\/em> button and appearing depends on the quality and consistency of the data ChatGPT receives. Incomplete variants, mismatched prices between the feed and the page or missing product identifiers all weaken a product’s chances of being surfaced accurately.<\/p>\n\n\n\n

The screenshot route makes the gap more visible. A shopper can still try on a brand’s product, but without a listing there is no price, no stock status and no link back to the merchant. Favourites add a second test: a saved item that shows an outdated price or a sold-out size when the shopper returns is a broken promise and OpenAI acknowledges that merchant price updates can take time to appear. This is the same feed-readiness logic that Google’s Universal Commerce Protocol introduced earlier this year. For brands selling across several markets, it has to hold in every language and size convention the catalogue uses.<\/p>\n\n\n\n

<\/div>\n\n\n\n

3. The result shows up in returns<\/strong><\/h3>\n\n\n\n

A shopper who arrives from ChatGPT after a try-on is further along than one arriving from a generic search. Judging that traffic on sessions or click-through alone will understate what is happening.<\/p>\n\n\n\n

The more useful measures are conversion rate, size-exchange rate and full returns for ChatGPT-referred visits to product pages, compared with the same products sold through other channels. GA4’s AI Assistant channel is a starting point for isolating this traffic, but returns sit in order data rather than in analytics, so the two need joining. If ChatGPT-referred orders come back more often on specific items, those listings need better fit information and imagery first. If they come back less often, those products are strong candidates for prospecting in other channels.<\/p>\n\n\n\n

<\/div>\n\n\n\n

<\/span>What to do about it<\/strong><\/span><\/h2>\n\n\n\n
    \n
  1. Audit how top sellers appear.<\/strong> Search ChatGPT for best-selling products by name and by plain description, in each priority market. Check whether each product appears, the Try on<\/em> button shows and how the generated image compares with the real fit.<\/li>\n\n\n\n
  2. Write fit into the product data.<\/strong> Add model height and size worn, sizing tendency, fabric and stretch and the key measurement to descriptions on site and in the feed.\u00a0<\/li>\n\n\n\n
  3. Align the feed with on-page structured data, then keep it fresh.<\/strong> Price, availability and product codes should match between feed and page. Use clear, front-facing product images, since the generated image works from what the listing shows.\u00a0<\/li>\n\n\n\n
  4. Read ChatGPT traffic through returns.<\/strong> Segment ChatGPT-referred visits to product pages, connect them to order data and review conversion, exchanges and returns against other channels.<\/li>\n<\/ol>\n\n\n\n

    None of this requires a new discipline. It sits across feed management, structured data, multilingual content and analytics – it rewards brands whose product data was already in good order. What Try on<\/em> changes is the cost of weak product data, because the first fitting room a shopper visits may now be one the brand never sees.<\/p>\n\n\n\n

    If you would like to understand how your catalogue appears in ChatGPT’s shopping results, get in touch with our team<\/a>.<\/p>\n\n\n\n

    <\/div>\n\n\n\n

    Sources<\/strong><\/h3>\n\n\n\n