ChatGPT shopping: catalog access is not a recommendation

Your product can be integrated without being recommended

For a Shopify owner checking ChatGPT shopping, separate three questions: is product data integrated, was this product selected, and which seller's offer appeared?

A missing recommendation does not answer the first question. A correct product card does not prove the right price, variant, or seller appeared.

OpenAI's shopping help says Shopify product data is integrated through Shopify Catalog, with no additional work required from individual merchants. It does not promise every product will appear.

---

Check these separately before rewriting anything

OpenAI says product selection considers the shopper's query and context, structured metadata, third-party content, model responses, and safety policies. It does not describe a simple feed-versus-blog ranking formula.

Check	What to do today	Save as evidence	What it cannot prove
Catalog integration	Confirm your store platform; Shopify merchants can use OpenAI's stated Catalog integration route, while other merchants should follow its direct-feed application guidance	Platform record and current official guidance; any feed acceptance record you actually have	That a specific SKU is eligible, ingested correctly, or selected
Exact product identity	Open the result's destination and compare the product name, color, kit, and variant with your product record	Result screenshot plus destination URL and variant	That other variants are represented correctly
Price and availability	Compare the displayed offer with the same variant on the live PDP; check shipping for the intended destination	Displayed price, PDP price, stock state, destination, timestamp	That every shopper sees the same offer or that a delayed update is a ranking problem
Product selection	Run one exact-name query and two real purchase questions, recording the full wording	Which products appeared, whether cards appeared, and any cited sources	Catalog exclusion when your product is absent
Seller selection	Open a product's merchant list and check seller identity, price, availability, and destination	Merchant name and offer link	Why that product beat another product
Recommendation accuracy	Compare each factual reason in the answer against the correct PDP and policy	Exact claim, source section, match or mismatch	That editing the source will change future recommendations

OpenAI lists availability, price, quality, and maker or primary-seller status as merchant-ranking factors. Those describe the seller list, not a disclosed product-ranking formula.

These are separate records, not a combined “AI visibility score.” Leave unknown fields unknown.

---

A kit example catches errors a brand-name search misses

The public Ugmonk Analog Card Bar Pro Kit page gives us concrete checks. This is an independent page inspection, not a customer engagement or a ChatGPT result test.

Fact on the inspected page	Exact check if a shopping result appears	Error to flag
Black Pro Kit is listed at $129 with a 16-inch bar	Follow the card to the Pro Kit, not the bar-only product	Treating a bar-only price as the kit price
Pro includes Small Task, Small Dot Grid, Large Dot Grid, and Weekly card packs	Compare the included card types	Describing all Analog kits as having the same contents
The kit selector lists Essentials at $79 with a 10-inch bar and two card types	Check which kit the answer recommends	Calling the cheaper Essentials a 16-inch kit
The banner says free US shipping $200+	Check the order and destination before repeating “free shipping”	Applying the threshold benefit to a $129 item alone
Returns or exchanges are within 30 days of delivery; used or non-original-condition returns may incur a 15% restocking fee	Preserve both the timing and condition	Rewriting this as an unconditional free trial

Prices and policies can change. Reopen the source before using these facts. The independent kit-page audit follows the same product through a fuller page-level review.

---

Keep a log that separates errors from absence

Spend 25 minutes on one product: five on your product record, ten on the queries, and ten comparing results. This is a workload limit, not a statistically valid ranking study.

Copy this record once per query:

test time / country / account context:
exact query:
product cards shown: yes / no
our exact product shown: yes / no
product and variant URL:
merchant and offer:
claim checked / source URL and section:
result: match / mismatch / not observable
next action: source correction / result feedback / no factual issue found

Run the same queries again after correcting a confirmed error. Keep account context and location as consistent as possible; record changes you cannot control.

A new appearance is an observation, not proof that the edit caused it. An absence is “not observed in this test,” not “excluded from the catalog.”

---

Do not use Google checks as proof about ChatGPT

Google's AI features documentation says supporting pages must be indexed and eligible for a search snippet. It requires no special AI markup.

For Google, inspect the exact URL in Search Console, check snippet restrictions, and compare structured data with visible facts. Keep Merchant Center information current where applicable.

Those checks diagnose Google access and data. They do not establish ChatGPT eligibility or prove that Google Shopping determines ChatGPT's choices.

---

Fix the mismatch you can demonstrate

Correct a wrong variant, stale offer, or incomplete policy in the source you control. If ChatGPT still shows incorrect information, use the product feedback controls described by OpenAI.

Do not buy a rewrite to solve an unproven catalog problem. If your verified issue is unclear PDP copy, review the free preview at Bulti Instant before paying.

Neither a feed nor a rewritten page guarantees selection, placement, or accurate future answers.