We'd like to use analytics tools to see how this site is used and what visits from AI assistants lead to. Not for ads. Decline: only a cookie-free page count runs. What we store.

Arbling

Which retailers AI links to for your products, and how to check it yourself

Published Nov 21, 2026Intermediate · 7 min read

What you’ll learn

  • What ChatGPT, Perplexity and Google each say about how a seller gets linked
  • How to record the sellers an AI answer shows for your buyer questions
  • How sample data separates a product that appears from a sale you actually get

Which retailers AI links to for a product is decided by each platform's own process, and the platforms describe only part of it. ChatGPT's shopping help page says it may show a list of merchants offering a selected product, built from merchant and product metadata from third-party providers or from merchants directly.

The same page says ChatGPT ranks merchants on factors like availability, price, quality, and whether they are the maker or primary seller of that item. Read that as examples, not a complete list. It also says the prices in the first response typically reflect the first merchant listed, which may not be the lowest available price.

The shopper matters too. The page says ChatGPT weighs the shopper's query and context, such as Memory or Custom instructions, and that not all available products will necessarily be shown. On the data side, OpenAI's product feed spec makes seller_name, the name of the seller supplying an offer, a required field on each row. A brand that sells through retailers and through its own store should check what that field says for each offer.

Perplexity describes its own side. Its blog post says its shopping product cards are not sponsored and are tailored to the search by its AI. It lists increased chances of being a recommended product as a benefit of its free Merchant Program, because members' products will be in its index. Increased chances is not a guarantee of placement, and nothing here says otherwise.

Google adds a third view. Its Search Central guidance says AI Mode and AI Overviews may use different models and techniques, so the responses and links they show will vary. It lists keeping Merchant Center and Business Profile information up to date among its SEO practices for AI features. In Merchant Center, open Analytics, then Products, to find the AI performance insights report. It covers English-language shopping queries in AI Mode and AI Overviews for accounts in Australia, Canada, India, New Zealand and the United States.

None of these pages is a ranking formula, and none promises that a feed, a program or a setting will win the link.

The practical way to find out which retailers AI links to for your category is to ask the questions your buyers ask, over and over, and write down every seller each answer shows. Because the platforms say results depend on the shopper's query and context and that links vary between surfaces, any single answer is one sample, not a verdict. Repeat the same questions over time and compare.

A simple record has four columns:

  • The exact buyer question, worded the way a shopper would type it.
  • Which assistant and which surface answered.
  • Which product, and which variant, appeared.
  • Which seller the answer linked to, whether that was your own store or a retailer.

Keep the wording of each question fixed between rounds, so a change in the answer reflects the answer and not a rewrite of the question. Do the same for the question phrasings you expect to matter least.

This is the work Arbling does at catalog scale: it turns a catalog into buyer questions and records which products, variants and sellers AI assistants show for each. If you are weighing how to measure this more broadly, our guide on how ChatGPT picks products to recommend covers what a shopper's question changes.

A worked example with sample data: one duvet cover, three sellers

A sample record shows why the seller column matters as much as the product column. Arbling's page shows sample data for a king linen duvet cover across 12 buyer questions. In that sample data, a competitor's cover appears in 7 of 12 answers, the brand's product sold by another store appears in 3, and the brand's product sold by the brand appears in 2. This is sample data, not a real brand, retailer or measured result.

Read the three outcomes separately. The 7 answers with a competitor's cover are a product-level gap: the brand's cover was not the one named. The 3 answers where the brand's product shows up but another store sells it are a seller-level gap: the right product, a link the brand does not control. The 2 answers where the brand sells its own product are the outcome a brand with its own store is usually after.

A tally that stops at "did the product appear" would count the second and third outcomes together and hide the difference. That is why every product in AI answers is matched to its own family, variant and seller in this sample. For a brand that sells through retailers, a retailer link is not necessarily a loss, but you cannot decide that without seeing it.

It also helps to know your catalog's own gaps before you read the answers. Our guide on product visibility in AI walks through which products get named and why a feed can disagree with a product page.

What to do next

Start with what you can see today. Run a free scan to see how AI assistants read your store as it stands, then list the buyer questions for the products you sell through retailers, the ones where a retailer link and your own link would both be plausible. Ask those questions repeatedly, record the seller each answer shows, and treat each round as one sample.

When a gap points to a product fact, Arbling has the product facts behind each gap fixed in the catalog; the merchant previews each change and publishes it on confirmation. If you would rather talk it through first, book a call with the founder.

For the structured fields behind this, see our guide to product structured data for AI shopping agents, and for the broader picture, agentic commerce explains the shift this question belongs to.

The sources above were checked on 21 November 2026.

Sources

See what AI assistants say about your business. Free scan, about a minute.

See who AI recommends
Share
Was this guide helpful?

Get new guides in your inbox.

No spam — one email when we publish something worth reading.

Yevgeniy Makhmutov
Founder, Arbling

Founder of Arbling. Arbling makes product catalogs readable, trusted, and buyable by AI shopping agents across regulated verticals.

Related concepts