ChatGPT is adding a virtual fitting step to product discovery. OpenAI’s 1 October 2026 release notes announce a virtual try-on feature for clothing and accessories, available on web and mobile. For ecommerce teams, the significance is that shoppers can move from finding a product to visualising it on themselves within the same conversation, before opening a retailer’s website.
Users select the “Try on” option on a product listing, then take or upload a selfie. ChatGPT Images generates the visualisation. The reference photo is retained for later attempts and can be replaced or removed through Settings, Personalization and Reference photos. The announcement also introduces saving products to Favorites and organising them in Library folders, connecting the immediate shopping session with later consideration.
A new step between discovery and the merchant click
The change adds another decision point inside the assistant. A shopper comparing jackets could assess an imagined look before deciding which retailer to visit. For brands, that makes presence in a product listing more valuable than a simple mention in a text answer. The commercial opportunity is to be represented accurately at the moment the user starts evaluating the item personally.
That is an interpretation of the new workflow, not evidence of increased conversion. OpenAI has not published performance figures for this launch, and the release notes do not establish a new ranking advantage for products used in try-ons. Teams should avoid treating the feature as proof that assistant interactions will replace product-page visits or reduce returns.
The limitations matter to merchandising. OpenAI’s shopping help page explains that generated try-on images can differ from the actual item or person and do not guarantee fit or size. Size charts, material descriptions and return terms therefore remain essential. A convincing visualisation should lead to better-informed consideration, with the retailer’s factual details available to resolve what the image cannot establish.
Catalog quality becomes a practical AI Search priority
Merchants should review the relationship between each product image and its corresponding catalog record. A blue jacket photograph associated with a black variant, or a discontinued item still presented as available, can undermine the shopping experience regardless of how polished the generated image looks. Audit stable identifiers, variant relationships, image URLs, prices and stock information across the website and any catalog distribution channels.
OpenAI’s product-feed documentation describes structured catalog data as an input for accurate product discovery, including pricing, availability and seller context. Feed hygiene is therefore a supported operational priority. It does not follow that submitting a feed guarantees inclusion, that every product receives a try-on button, or that higher-resolution photography is a newly disclosed ranking signal.
For SEO and AI Search teams, the useful work spans content and merchandising: keep product facts consistent, make the correct variant easy to identify, and provide images that clearly represent what is sold. Treat descriptive copy, image management and catalog maintenance as connected responsibilities rather than optimising a product title while leaving its underlying records inconsistent.
Measure the journey without inventing attribution
Retailers can monitor the quality of their product representation in sampled shopping conversations and inspect identifiable visits and purchases in their own analytics. Those observations should remain separate: a product appearing in a response, a virtual try-on and a tracked merchant visit are different events. The launch does not document a merchant dashboard exposing every try-on interaction or its downstream conversion.
The immediate response is a focused catalog audit and a baseline of observable results. Virtual try-on brings visual consideration closer to conversational discovery, while saved products may extend that consideration beyond one session. The opportunity is real as a new user experience; its effect on revenue, ranking and click behaviour still needs measurement.