Prepare product data for ChatGPT shopping discovery without guessing.

What “ChatGPT shopping visibility” should mean for a merchant

There are two different jobs: preparing trustworthy product data for OpenAI commerce surfaces, and measuring whether a product actually appears in a ChatGPT shopping experience. Those should not be mixed together. Lumen audits the product and feed evidence it can access and labels third-party search evidence separately; it does not relabel that evidence as a direct ChatGPT result.

Three evidence checks belong in the conversation

1. Feed readiness: check whether the required product facts are present and usable in the published feed format. OpenAI’s Stable file-upload discovery specification lists nine required fields; Ads and checkout have separate requirements.
2. Storefront consistency: compare a product row with the page it points to. The selected variant, product identity, price, currency, image, and availability should describe the same offer.
3. Observed placement: record whether the product was actually returned for a specific shopping query, in which experience, and when. If a direct ChatGPT result was not observed, report it as “not measured,” not as a ranking or a failure.

Start with a representative product sample

Include ordinary products and important variants. For a size or color variant, the product identifier, page URL, price, and stock status should resolve to the purchasable option being checked. A parent page that opens on a different variant can make good feed data look inconsistent to a buyer or automated checker.

Lumen establishes a baseline from the storefront and published product evidence available to the assessment. Missing fields and mismatches can then become prioritized implementation work instead of a vague visibility score. See the related ChatGPT product-feed checks and feed-to-page mismatch guide.

Remeasure the same evidence after a fix

Keep the sample products and checks consistent between the baseline and follow-up. Compare which fields are now present, whether the relevant page agrees, and which issues remain. This makes the impact of a catalog or template change reviewable. It does not prove that ChatGPT will recommend a product, and no assessment provider can guarantee placement.

Questions merchants should ask

Does feed readiness guarantee visibility? No. It checks data quality and consistency, not placement. Does Lumen submit a merchant feed to OpenAI? No; the assessment identifies evidence and implementation gaps for the merchant’s team. What if a feed or shopping result is not accessible? The report should say that the signal was unavailable or not measured rather than infer a pass.