ChatGPT users can save suggested products and places to Favorites, retrieve them in Library and organize them into folders. The feature adds a way to keep a recommendation available after its initial appearance in a conversation. For brands and local businesses, the commercial implication is a potential bridge between being discovered and being reconsidered later, although OpenAI has not documented a new ranking benefit or advertiser measurement tool tied to saves.
The official Library documentation, checked on October 3, explains that selecting a product or place's bookmark icon saves it to Favorites. On the web, saving also opens an optional folder menu, and users can move saved items between folders from Library. The page displayed a relative update date of two days earlier; it does not establish an exact launch time for the feature.
A recommendation can survive the conversation
The important change is continuity. A product encountered during a comparison or a place suggested while planning a trip can be saved deliberately and revisited. Users no longer have to rely entirely on remembering which conversation contained a particular option. This makes the idea of an AI shortlist tangible: a collection of candidates the person has chosen to retain.
Consider someone exploring several restaurants for an upcoming visit. A saved place could remain a candidate even if the person postpones the decision. Similarly, a shopper might preserve a product while checking alternatives or waiting to buy. These examples illustrate possible use, rather than documented behavior or evidence that Favorites increases purchasing.
A save also has a different meaning from a citation or click. A citation indicates that an answer referenced something; a click indicates that a user followed a route outward. A favorite reflects a decision to retain an item inside ChatGPT. Those events may occur together, but they should not be treated as interchangeable measures of demand or conversion.
What brands can prepare without inventing a new signal
For ecommerce teams, the useful response is to review whether products are represented clearly enough for a person to recognize them later. Accurate naming, meaningful images and unambiguous variant information can help customers distinguish candidates. This is a practical recommendation about product presentation, not a published formula for earning saves or appearing in ChatGPT recommendations.
For local businesses, the same principle applies to place identity. A person returning to an option needs to understand which location it refers to and where to verify current information before visiting. A saved candidate should not be assumed to preserve live opening hours, prices or availability: the documentation does not explain how every saved product or place is refreshed.
Teams should also avoid interpreting the feature as proof of recommendation memory. OpenAI documents saving and organization, but this page does not say that favoriting an item boosts its visibility in future answers or changes how alternatives are selected. User-controlled collections and personalized recommendation logic are separate questions.
The measurement gap remains
The commercial opportunity is easier to describe than to quantify. A retained item may be considered after the original session, but the Library documentation does not promise merchants or place owners access to save counts, folder activity or a Favorites attribution report. A brand cannot assume it can observe a customer's shortlist simply because the customer can create one.
Where businesses can measure visits and purchases, those outcomes remain useful. They do not reveal the entire path through an assistant, particularly when discovery, saving and action happen at different times. The next development to watch is whether OpenAI documents refresh behavior or commercial reporting around saved items. For now, Favorites expands what users can do with recommendations: keep them, organize them and return when they are ready to decide.