Brand or retailer: who gets recommended when your products sell through other stores

A brand that sells through Home Depot, Amazon and a dozen specialty retailers has the same product on a dozen pages. When a shopper asks ChatGPT, Google AI Mode, Gemini, Perplexity or Copilot for a recommendation, the assistant reads the pages and feeds it can reach and picks the ones that answer best. Three things tilt that choice toward the retailer.

• Completeness. A large retailer's PDP usually has the specification table, the Q&A block, the review volume and the comparisons the question needs.
• Eligibility. Google groups offers for the same product by its identifiers and shows prices from several merchants in AI Mode product listings. ChatGPT reads the OpenAI product feed from merchants who submit one. A retailer with a feed in every program is eligible everywhere.
• Trust signals. Review counts, return policies, shipping speed and domain history favour the retailer when the shopper has named no brand.

The assistant has chosen your product and a place to buy it. The question is whether that place is right for your business. For retail partners as a shelf, see digital shelf and AI.

The brand is the source of record for its own products, whoever sells them. Four things carry across every listing.

• Identifiers. The GTIN is assigned by the manufacturer, and Google uses it to match offers to one product and warns that products with missing or incorrect GTINs may have limited visibility. The brand attribute names the manufacturer on each listing. Consistent GTIN, brand and MPN across every feed is what lets an assistant see one product in many stores.
• Titles and attributes. The title, specifications, materials, dimensions, compatibility and use case start in your product record and travel to each retailer through its item setup. If the record is thin, every listing is thin.
• Brand-supplied content for Google's catalog. Google's Manufacturer Center lets brand owners submit titles, descriptions, images, key features and videos that enrich Google's product catalog, whether or not the brand sells direct.
• Q&A. The fit, compatibility, care and delivery answers are the same on every listing. Written once, they go on your PDPs, to retailers through their content programs and into Google Merchant Center conversational attributes on your own feed.

If you also sell direct, your own GMC listing, PDP structured data and OpenAI product feed carry the same record, so a fix at the source reaches your site and, through syndication, the retailers.

There are two versions of the question. For a direct-led brand: the goal is the brand site showing up, so is a big-box retailer showing up instead fine? For a wholesale-led brand: the same SKU is listed on many sites and the sale can go to any of them.

Both are reasonable, and they lead to different work. A retailer winning is fine when the wholesale margin is acceptable, when the alternative was a competitor's product, and when the retailer listing is accurate: the right variant, current price, your images and your answers.

The brand site has to be the link when a line is direct-only, when the direct margin or the first-party relationship is the point, when bundles or subscriptions exist only on your site, or when the retailer listing is wrong and the assistant repeats the error. Most brands land on a mix: brand presence everywhere, site presence for specific lines and branded questions. See branded vs unbranded questions for the split.

One number hides the decision. Track these separately for the same questions shoppers ask, per assistant:

• Brand presence. Your brand or product is named, carded or cited in the answer, whichever domain the link points to. The number for the wholesale business.
• Site presence. The product card or citation links to your domain. The number for direct sales.
• Who got the link. When it was a retailer, which one, and whether the listing was correct.

GMC's AI performance insights describes how your brand is discovered in shopping questions on Google AI Mode and AI Overviews, with share of voice against competing brands: brand presence. Search Console's generative AI report counts impressions for your pages: site presence on Google surfaces. For ChatGPT, Gemini, Perplexity and Copilot, the per-question record of appearances by type and destination is the measurement, using the appearance types. Paz tracks all three per assistant against your baseline and names the retailer that took the links.

The work starts in the record every listing inherits. Our experts find the questions shoppers ask where your product is missing or a retailer's listing answers badly, then complete the identifiers, titles, attributes and Q&A from your own sources within the guidelines you approve. Paz publishes the result to your PDPs, your GMC feed and your OpenAI feed through supported connections, and prepares the same content for you to send to your retail partners. Where a retailer listing shows the wrong variant or an old price, our experts bring it to you with the evidence for that partner.

Appearances are tracked per assistant as brand presence, site presence and who got the link, so you can decide where the brand site has to win. This sits across the product content service and the AI visibility service.