OpenAI Expands Product Discovery While Prioritising Merchant-Owned Checkout
OpenAI’s March commerce update separates richer product comparison from the checkout decision, with implications for the information brands supply and the purchase journeys they maintain.

OpenAI announced richer product-discovery experiences in ChatGPT on 24 March 2026, including visual browsing and side-by-side comparison. It also described expanding the Agentic Commerce Protocol to support product discovery.
In the announcement, the company said it would prioritise product discovery while allowing merchants to use their own checkout experiences. It described merchant feeds and platform integrations as part of the information infrastructure.
The announcement is a dated product development, not a guarantee that a specific merchant, product or market is currently supported. Current participation details should be checked against OpenAI's merchant information.
The shortlist and the transaction can sit in different places
Frontier's interpretation is that brands should examine discovery and purchase as connected but distinct experiences. An assistant may help the shopper decide what to buy, while the merchant remains responsible for completing the order coherently.
That makes the handover important. The product, variant, price and conditions should remain understandable when the customer reaches the merchant's page. A recommendation loses value if the destination creates fresh uncertainty.
It also makes product information a marketing input. A catalogue can help identify an item while still leaving the shopper's main question unanswered. Suitability, compatibility, limitations and local availability require reliable records and readable explanations.
The product-information article connects those facts to the actual buying decision.
Measurement should follow the same distinction
A brand can inspect whether its products are accurately represented and separately analyse identifiable referrals and orders. It should avoid assuming that an observed recommendation caused a particular transaction without evidence connecting the events.
The GEO measurement framework keeps answer quality, behaviour and commercial outcomes in their own units.
For Australian, New Zealand and China-market teams, availability and supported integrations need specific checks. A global product announcement does not establish identical local functionality.
OpenAI's update provides a clear example of commerce infrastructure evolving around product consideration. The durable preparation for a brand is to maintain accurate information and a purchase journey that can support the recommendation once the shopper arrives.



