Brand Mention, Product Card and Citation in AI Answers

Buyers ask to be "mentioned and cited" and ask who scores "which position you are in within those cards." Those are different events, and one answer can contain any mix of them. Paz records four:

• Brand mention. The answer names your brand in text, with no specific product and no link. It builds recognition and often sends a shopper to search your brand later.
• Product mention. The answer names a specific product in text, with or without a link. The shopper now has something to search for or click.
• Product card. The answer shows the product as a card with image, title, price and a merchant link. In ChatGPT this comes from the product feed or Shopify Catalog, in Google AI Mode from Google Merchant Center. Cards carry the highest click intent and depend most on feed quality.
• Source citation. The answer links to a page on your site as a source for a claim it made. Citations are what the Search Console generative AI report and Bing AI Performance count.

A shopper asking ChatGPT for the best waterproof boots under 200 dollars sees a shortlist. If your brand is mentioned, you are on the list. If your product has a card in the top set, you are one click from the product page, where the Shopify agentic report and the GA4 AI Assistant channel record the visit. If your sizing guide is cited for "runs half a size large", the shopper arrives with the question already answered.

Position matters inside each type. Third card of three is a different outcome from first of eight, and a citation at the top of an answer is read more than one in a footnote list. Paz records position where the assistant exposes it, with the engine, locale, date and the underlying answer, so a change can be traced.

How each type is produced also differs. Cards depend on feed quality and eligibility. Mentions and citations depend on the product page, the product Q&A it carries and the third-party pages that talk about you. Fixing one may move one type before another.

Rolled into a single "visibility" number, the four types tell you very little about what to do next. Split out, they point at the work:

• Brand mentions up, product cards flat. The assistant knows the brand and has nothing eligible to show. Look at Google Merchant Center and OpenAI feed eligibility, identifiers, availability and attributes.
• Cards present, citations absent. The product is eligible, and the page itself is giving the assistant nothing worth quoting. Look at product Q&A, specification detail and visible claims.
• Citations to third-party pages, none to yours. Reviews, forums or retailers answer the question better than your page does. Write the answer on your page.

Two aggregates sit on top of this split. AI product found rate is the share of configured questions where the merchant appears in any of the four types. AI share of voice compares your appearances with the brands that appear for the same questions. Both are only as useful as the appearance types underneath them. The measurement methodology defines the recording rules.

Paz runs a configured set of the questions shoppers ask across ChatGPT, Google AI Mode, Google AI Overviews, Gemini, Perplexity and Copilot on a schedule, built from your buyer personas and selling priorities. Each run records, per question and assistant, which of your products appeared, as which appearance type, in which position, with the underlying answer saved. Competitors appearing for the same questions are recorded the same way.

Our experts then read the pattern. A collection with mentions and no cards gets feed and attribute work. A product with cards and no citations gets product Q&A on its page. Fixes ship within the guidelines you approve, exceptions come to you first, and the next runs show whether the appearance type moved. See the AI visibility service.