OpenAI’s move to accept structured product data feeds marks a meaningful shift in how commerce information can power conversational discovery. Instead of products being surfaced only in response to rigid keyword queries, ChatGPT introduces a new paradigm: products that participate in natural language conversations, recommendations, and exploratory shopping journeys.
At a high level, OpenAI’s current feed specification includes familiar elements—product identifiers, pricing, availability, variants, and images. For merchants already managing feeds for platforms like Google Shopping, this may look recognizable at first glance. However, that familiarity can be misleading. These are not simply “another destination feed.” OpenAI’s product feeds introduce new data concepts, new use cases, and new expectations around how product information is structured and interpreted.
Not a Standard Product Feed
Unlike traditional shopping feeds, OpenAI’s commerce feeds are designed to support reasoning and recommendations, not just listings. The specification already supports (and strongly encourages) data beyond the basics, such as:
- Product relationships (e.g., accessories, bundles, or “often bought together” items)
- Performance and behavioral signals (such as popularity or return tendencies)
- Rich, well-normalized attributes that help models understand why a product is a good fit—not just what it is
Implementing these fields well requires more than simple field mapping. It requires judgment about which signals matter, how they should be calculated, and how they should be represented in a way that aligns with OpenAI’s evolving expectations.
A Moving Target
It’s also important to recognize that this ecosystem is still emerging. OpenAI’s product feeds are non-standard and actively evolving. Field definitions, supported attributes, and best practices are likely to change as the product matures and as OpenAI learns from early merchant integrations.
That means compliance is not a one-and-done exercise. Maintaining a high-quality ChatGPT feed will require:
- Ongoing monitoring of spec updates and policy changes
- Iteration as new optional or recommended fields are introduced
- Adjustments as OpenAI refines how product data is used in conversational contexts
This is fundamentally different from legacy feeds that remain relatively static for years at a time.
Where VersaFeed Comes In
This is where VersaFeed’s expertise becomes critical. While some merchants may be able to submit a basic feed, unlocking the real value of ChatGPT commerce—and staying compliant as the spec evolves—requires deep experience with feed architecture, enrichment, and optimization.
VersaFeed helps merchants:
- Design and build non-standard feeds that go beyond baseline requirements
- Enrich product data with AI-assisted attribute completion and normalization
- Model advanced signals like product relationships, bundles, and performance indicators
- Adapt quickly as OpenAI updates its specifications and guidance
These feeds are more complex than traditional product feeds, and they demand a higher level of technical and strategic involvement. As a result, implementations should be scoped and priced accordingly.
Getting Started
Merchants interested in participating must first register at ChatGPT. OpenAI also offers an optional instant checkout integration, which is not required but may influence approval timelines.
As conversational commerce continues to evolve, early adopters have a unique opportunity—but only if their data is built correctly. VersaFeed is prepared to help merchants navigate this new and rapidly changing landscape, turning experimental feed requirements into a durable competitive advantage.