Product feeds for AI commerce
Prepare factual, well-structured catalog data and understand what a source audit can establish. Native feed compatibility, independent AI readiness and platform approval are separate questions.
Choose the right starting point
- ChatGPT Ads Feed Checker for the native OpenAI format.
- AI Shopping Readiness for independent coverage dimensions.
- Product Feed Quality for general source checks.
- Merchant Center feed errors for source diagnostics and their limits.
- Scoring methodology for rules, versions and limitations.
Recent ChatGPT Ads updates and rollout guides
ChatGPT Ads product feed requirements
Start with the specification for the feed format you actually submit. Product Feed Scan checks a native OpenAI product file; passing this audit is not Ads approval.
Read the guide →Practical guides
ChatGPT Ads product feed requirements
Start with the specification for the feed format you actually submit. Product Feed Scan checks a native OpenAI product file; passing this audit is not Ads approval.
Read guide →AI commerceChatGPT Ads feed vs Google Shopping feed
Treat the two feeds as separate export contracts. Shared product facts can be reused, but field mapping, applicable checks and platform decisions must be reviewed separately.
Read guide →AI commerceAI shopping product data checklist
Review six areas: identity, content, taxonomy, variants, assets and optional enrichment. These are Product Feed Scan coverage dimensions, not a universal AI platform certification.
Read guide →AI commerceProduct taxonomy for AI commerce
Category presence and category correctness are different checks. Product Feed Scan detects missing category context; selecting the right category still requires a catalog and destination-taxonomy review.
Read guide →AI commerceProduct identifiers for AI commerce
Use stable item IDs for catalog records and genuine assigned identifiers where applicable. A valid-looking GTIN is not proof of ownership, and a feed ID is not the same as a product ID.
Read guide →AI commerceProduct feed delivery errors vs data errors
First determine whether the receiver obtained the file. Delivery failure, parsing failure and product-data failure require different evidence. A successful local audit does not test SFTP or API delivery.
Read guide →A safe improvement workflow
Scan the current export, review the actual findings, repair the source mapping and rescan. Check the destination separately. No audit score guarantees advertising acceptance, AI recommendations or additional sales.