How to get your products recommended in ChatGPT

In 2026, ChatGPT product discovery can use merchant-supplied product data through the Agentic Commerce Protocol (ACP), Shopify Catalog for Shopify merchants, and information available across the web.

OpenAI says a product can appear when ChatGPT considers it relevant to the user's intent and conversation context. It may consider factors including availability, price, reviews, ease of use and other context provided by the shopper. That means product discovery is not simply a keyword-title competition; the ranking factors guide sets out what OpenAI documents for each decision and what two 2026 studies measured.

A query such as “comfortable black heels I can wear to a wedding all day” contains several decision criteria that may not exist as neat fields in a standard catalogue. SO-ME, the consumer shopping platform Sighted powers, meets that from the shopper's side: requests arrive in those words, and matching them to the product information in the retailer feeds it receives is difficult. Think in terms of information coverage, not keyword repetition.

1. Shopify Catalog

Shopify merchant product data is already integrated into ChatGPT through Shopify Catalog, with no additional work required from individual Shopify merchants for that catalogue integration, provided the store and the product meet Shopify Catalog's requirements.

That does not mean every Shopify product will automatically be recommended for every relevant request. Integration solves part of the availability problem; it does not guarantee the underlying product information is complete, differentiated or well matched to a shopper's constraints.

2. Product feeds from approved OpenAI partners

OpenAI also takes product feeds from approved partners, and supports delivery through third-party commerce providers. For larger merchants, this creates a more direct path for product representation than relying only on web crawling. The feed's columns, how it is delivered and the Ads Manager setting that keeps it in the organic results are on the ChatGPT Shopping feed page.

3. The open web

ChatGPT shopping and shopping research can also use web information, read by OAI-SearchBot, OpenAI's search crawler. Product pages therefore still matter, particularly when they contain important details that are missing from a feed or catalogue source, and a page the crawler is blocked from reading is absent from search answers; the troubleshooting guide runs that check first.

The question to ask is whether the same product is represented accurately, consistently and usefully across the sources the platform may use, rather than “feed or website?”; the feed, schema and product page guide sets out what each one carries.

Make the facts complete

Start with the basics: title, variant, category, price, availability, size, material, colour, identifiers, images, delivery and the other core attributes relevant to the category.

Add supportable decision context

Look at how real shoppers ask for the product. If customers routinely care about crease resistance, packability, skill level, compatibility, washability, room size, pet age or gifting occasion, work out where those facts can be accurately represented.

Do not invent attributes because an LLM says they sound plausible. Enrichment should be grounded in manufacturer information, retailer knowledge, product documentation or other defensible sources.

Keep sources consistent

Conflicting price, stock, colour, product names or specifications across the feed and the product page create unnecessary uncertainty. Product data hygiene still matters.

Test recommendation, not just presence

Searching for your own brand and finding yourself proves very little. Use realistic, non-branded shopping requests where your product is genuinely eligible, and record:

• whether the brand or product appears;
• how it is described;
• why it is recommended;
• which merchant is selected;
• whether the result is stable across repeated tests and query variations;
• whether changes improve the outcome over time.

Sighted's pilot with The Game Collection reported a share of a defined set: 200 products were enriched and 17 collections created, AI citation coverage went from 0% to 32% of the priority products tested, and the store's organic conversion rate rose 64%. A share of a defined product set can be measured again next quarter, which a brand search that finds you cannot.

OpenAI deprioritised Instant Checkout, its native checkout inside ChatGPT, in March 2026. Its commerce model is still evolving, and since then it has put its emphasis on product discovery and merchant flexibility rather than on native checkout.

For UK retailers, the immediate commercial priority is to make sure your products can be discovered and evaluated correctly now, while keeping your catalogue ready for deeper commerce integrations as they expand, rather than waiting until checkout is everywhere.

Do I need Shopify to appear in ChatGPT shopping?

No. Shopify has a direct catalogue integration, but OpenAI also takes product feeds from approved partners and uses web information in shopping experiences.

Does appearing in ChatGPT cost money?

Organic product results are separate from advertising. OpenAI says ads do not influence ChatGPT answers or organic product-result selection.

Should I create an llms.txt file for ChatGPT shopping?

Do not make an llms.txt file your priority. The more important work is accurate product representation in supported commerce and product sources, and useful crawlable product information. Platform support for auxiliary files can change, so test based on documented behaviour rather than assuming an “AI SEO” file is a ranking lever.

1. Confirm your route in: on Shopify, that the store and product meet Shopify Catalog's requirements; otherwise a feed through an approved OpenAI partner, or product pages OAI-SearchBot is allowed to read. The requirements checklist goes through them.
2. Make the core facts complete: title, variant, category, price, availability, size, material, colour, identifiers, images and delivery.
3. Add the decision context shoppers ask for, grounded in manufacturer information, retailer knowledge or product documentation, never invented.
4. Remove conflicts in price, stock, colour, product names and specifications between the feed and the product page.
5. Test with realistic, non-branded shopping requests and record whether the product appears, how it is described, why it is recommended and which merchant is selected.
6. Repeat the same tests over time to see whether your changes improve the outcome.

The free AI visibility audit reads up to 50 of your product pages, checks whether OAI-SearchBot is allowed in, and reports which facts each page states and which it leaves out. Inside the product, Sighted enriches the catalogue at scale for this kind of discovery: the attributes a product's category needs, written from your own descriptions, options and photos for your approval.

On Shopify the approved values are written back to the product, so Shopify Catalog carries them, and for approved OpenAI partners the same records build the feed. SO-ME runs on the same enrichment.