ChatGPT is adding two steps between product discovery and a later purchase: a generated virtual try-on and a saved shortlist. OpenAI’s October 1, 2026 release notes confirm that clothing and accessories can be tried on through ChatGPT Images, while products can be saved to Favorites or organised into folders.
For brands, the significance is a longer consideration journey inside the assistant. Being recommended may now lead to a visual comparison and a return visit to a saved item. That is an editorial interpretation of the new workflow, rather than a claim that OpenAI has announced new merchant ranking factors.
What OpenAI has documented
According to the official release notes, a “Try on” button appears on clothing and accessory listings. Users take or upload a selfie, and ChatGPT Images generates a virtual try-on. The reference photo is retained for subsequent attempts and can be changed or deleted under Settings → Personalization → Reference photos. OpenAI lists the shopping update as available on mobile and web.
The separate Library guide explains that products and places can be saved using a bookmark icon. On web, bookmarking opens an optional folder menu; saved items can subsequently be moved between folders. Saving an item should be understood as a user action, rather than evidence of an order.
A product recommendation can become a shortlist
The useful shift for AI Search strategy is that discovery no longer has to end with an immediate outbound click. A shopper can examine the suggested item, visualise it and preserve it for later consideration. The product’s identity needs to remain understandable throughout those steps.
For example, a jacket with an unclear colour name or ambiguous variant information may be harder to compare confidently with another saved option. A useful merchant audit would therefore examine whether product names, images, variant descriptions and destination pages describe the same item consistently.
This recommendation does not imply that those fields have become newly confirmed ranking signals. It follows from the practical need to help shoppers recognise and evaluate a product after the assistant surfaces it.
Images carry more of the consideration work
A virtual try-on brings visual assets closer to the decision. Merchants should review whether their product photography makes the garment’s colour, construction and distinctive details easy to understand, and whether those images correspond to the variant being offered.
The generated preview should also be described carefully. A visualisation is not a measurement of fit, a guarantee of comfort or confirmation that a size will suit the buyer. Accurate size information, materials and return terms remain useful on the merchant’s own page.
Brands evaluating the experience can record whether the correct item appears, whether its variant remains identifiable and whether the onward product page supplies the information needed for an actual purchase decision.
Saved intent creates a measurement question
Favorites suggest consideration that can persist beyond the first recommendation. They do not automatically create a merchant-visible analytics event. The documentation reviewed for this update does not establish merchant reporting for saves or try-on activity, or prove that either action changes subsequent recommendation rankings.
For now, keep exposure testing separate from commercial outcomes. Track what can actually be observed, then compare available referral and purchase data without labelling every saved product a conversion. ChatGPT’s shopping experience is gaining more decision steps; measuring their commercial effect still requires evidence.