TLDR;
OpenAI launched Shopping Research in ChatGPT. It asks customers questions and builds buying guides around their needs. Retailers need clear compatibility details and variant differences to support accurate recommendations, and should still check that suggested prices and stock are current.
What happened
OpenAI launched Shopping Research, a conversational tool that asks clarifying questions, researches product options and produces tailored buying guides. It broadened shopping discovery beyond a simple results carousel, with product information drawn from available sources and the usual need to verify current prices and stock.
Why it matters
Clarifying questions move shopping research beyond a broad category match. The assistant may need to distinguish materials, dimensions, compatibility, use cases and trade-offs across a range. Catalogue pages that repeat the same promotional language give it little basis for choosing the right product, particularly when variants differ in commercially important ways.
Shopping research moves product selection towards a dialogue about requirements. A customer may begin with a use case, answer clarifying questions and receive a tailored set of candidates. That makes suitability information more valuable than a broad category description. A product that looks competitive on headline specifications may fail the buyer's actual constraint, such as compatibility, space, maintenance or an ongoing cost. For brands, the commercial opportunity is to explain those details clearly enough that the product can be considered for the right reasons. The risk is a persuasive guide that overlooks a limitation. Retailers should care about the quality of the shortlist and the eventual purchase, rather than celebrate inclusion regardless of fit. A recommendation that attracts unsuitable demand can increase returns and support costs instead of profitable growth.
How your brand can benefit / be affected
Audit high-value product families against the questions customers actually ask before buying. Make suitability, limitations and variant differences explicit, backed by current specifications and supporting evidence.
Test representative buying scenarios and check the final guide for incorrect prices, unavailable products and mismatched variants. Use those errors to prioritise catalogue improvements. Recommendation frequency is useful only alongside the accuracy of the recommendation and the quality of the resulting customer journey.
Build product information around the questions that change a buying decision. Explain who the item suits, what it works with, what is included and where it is a poor fit. Use precise identifiers for variants that differ materially. For example, two products with the same family name may have different connection requirements or service entitlements; those differences should not be buried in an image or a downloadable document alone. Keep the explanation understandable for a non-specialist buyer. The purpose is to make accurate matching possible, while also helping customers who reach the product page without using an assistant.
Test research-style questions with several constraints, including cases where your product should not be recommended. Compare the result with the approved catalogue and note which missing details led to a poor match. Correct those details at source, then check whether the destination page helps the buyer confirm the choice. Measure engagement with suitable products, completed purchases and expectation-related returns. Keep attribution limits clear because observing a guide does not show how often customers receive it. Use the exercise to improve merchandising quality and identify genuine information gaps. The brand benefits when the assistant helps a customer understand suitability, not merely when its name appears in another list of products.
News date: 24 November 2025. Editorial review: 16 September 2026. Analysis includes subsequent developments where stated.