Short answer: you are almost certainly already in the pipeline. Shopify feeds ChatGPT shopping automatically, so when your products are invisible, something is filtering you out. There are five real suspects. Here is the diagnostic we run on our own stores, in order.
If your store runs on Shopify, it is almost certainly already inside ChatGPT's shopping pipeline. Shopify's catalog integration feeds merchant stores into ChatGPT automatically, with no app to install and no feed to submit. So when your products are not showing up in answers, the question is not how to get in. It is which of five things is filtering you out: product data quality, eligibility, corroboration, category competition, or crawler access. We hit the last one ourselves, on our own site, while selling GEO as a service. That embarrassment became this diagnostic. We run it on Pet Buckets and EMPRICS, our own Shopify stores, and now it is yours.
Almost certainly yes, and this is the fact most store owners have not caught up with. In 2026 OpenAI rolled out a commerce model where ChatGPT surfaces products inside the conversation, and when the shopper decides to buy, it sends them to the merchant's own store to complete the purchase. Discovery happens in the chat. The transaction happens on your storefront. The order lands in your Shopify admin, and the customer relationship stays yours.
For Shopify merchants specifically, entry is automatic. Shopify's catalog integration feeds stores into ChatGPT shopping at the platform level, and OpenAI's own line on it was that Shopify merchants need no additional work to be included. No feed to build, no app to install, no waitlist to sit on. Your products are, in principle, already available to be recommended.
That flips the whole question. If you were outside the pipeline, the job would be submission. Since you are inside it, the job is diagnosis: something is causing the model to pass over your products when it assembles an answer. In our experience it is one of five things, and they are checkable in an afternoon.
The pipeline runs on your catalog data. Vague titles, missing descriptions, no availability, inconsistent pricing, weak images. When the model cannot cleanly parse what a product is, what it costs and whether it can ship, it reaches for a product it understands. This is the most common cause we see, and the most fixable.
Some products are excluded by category or policy, and availability is scoped by region and currency. A product that cannot be bought by the person asking is not going to be shown to them. If your catalog is fine but a specific product line never appears, check whether it sits in a restricted category or is simply not purchasable where your buyers are asking from.
A language model recommending a store is making a trust decision, and it wants evidence beyond your own site. Reviews, mentions, a consistent brand entity, other pages on the internet that agree you exist and are what you say you are. A store no one else has ever written about is hard to recommend with confidence. This is the slowest lever to move, and the one that separates stores that get named from stores that get skipped.
ChatGPT does not list every eligible product. It picks a handful for the answer. In a category with thousands of near-identical products, being eligible is nowhere near enough, and generic products with no differentiation get averaged out of the answer. Specific beats generic here: the more precisely a product matches a specific ask, the better its odds.
The quiet killer. OpenAI's crawlers, OAI-SearchBot for search and ChatGPT-User for live visits, need to reach your store. A robots.txt rule, a firewall, or a CDN bot-protection setting can shut them out without anyone noticing. We know because it happened to us: our own CDN was silently blocking AI crawlers through a dashboard default we never chose, on the website of an agency that sells GEO. One toggle. Weeks of invisibility. Check this even if you are sure it is fine, especially if you are sure it is fine.
Receipt · Own Brand Pet Buckets · Our Own Shopify DTC StoreEvery check in this article runs on our own storefront first. This is the store we break things on so clients do not have to. Open →Data quality: open your five best sellers in Shopify admin and read them like a machine would. Is the title specific, the description factual, the price and availability current, the product type and vendor filled? Then ask ChatGPT to shop for that exact kind of product and read what it surfaces instead of you, because the products it picks show you what complete data looks like in your category.
Eligibility: check the product's category against restricted lists, and confirm it is actually purchasable in the regions your buyers ask from, in their currency.
Corroboration: search your brand name in a fresh chat and see what the model knows about you. If the answer is thin or confused, your problem is entity and reputation work, not your catalog. Reviews you genuinely have, consistent business details everywhere, and real coverage all feed this.
Competition: ask for your category ten different ways and note who keeps appearing. If the same three brands own every phrasing, study what they have that you do not, usually sharper differentiation, better data, or more corroboration.
Crawler access: read yourdomain.com/robots.txt and look for rules blocking OAI-SearchBot or ChatGPT-User. Then log into your CDN or bot-protection dashboard and find the AI-crawler controls, because these are often set to block by default. Then check your server or CDN logs for those user agents actually arriving. Clean robots.txt plus zero crawler hits in the logs means something upstream is still eating them.
OpenAI has been unusually plain about this. Shopping results are organic, not ads, and there is no way to pay for placement. Relevance to the shopper's request comes first, and alongside it OpenAI names signals like availability, price, quality, and whether you are the maker or the primary seller of the product rather than a reseller three steps removed.
Read that list again, because it is just retail fundamentals. In-stock products, honestly priced, accurately described, sold by the brand that makes them. There is no growth hack in it, which is bad news for anyone selling one and good news for any store willing to do the unglamorous data work. The stores winning this channel early are mostly winning on completeness, not cleverness.
One caution from the other direction. Anyone who tells you they can promise your products a spot in ChatGPT's answers is lying to you, the same way rank promisers always have. The pipeline is real, the levers are real, and the outcome still belongs to the model.
You are not locked out, you just have to do the work Shopify does for its merchants automatically. OpenAI runs a merchant program where non-Shopify stores, WooCommerce, custom builds, whatever you run, submit a product feed directly. The feed specification lives in OpenAI's own developer documentation, and that is where you should read it, not in a blog summary that was accurate the month it was written. Specs like this move fast, and a feed built from a stale secondhand post fails in ways that are miserable to debug.
The five reasons above still apply once your feed is in. A submitted feed with thin data, no corroboration and blocked crawlers is exactly as invisible as an automatic one.
Top to bottom, in this order, because sequence stops you polishing the wrong thing first.
We will run this exact diagnostic on your store: crawler access, catalog data, eligibility, corroboration and category position, and hand you the findings with the fixes in priority order. On our own stores first, then on yours.
Get the diagnostic→no retainer trap. just truth.