Skip to content
RankX AI
RankX AI Docs
AI Shopping

AI Shopping

Whether your products are recommended when a shopper asks an AI assistant. Two surfaces, product-level rather than brand-level, with honest denominators.

AI Shopping answers a different question from AI Visibility: not "is our brand named" but "do our products get recommended". It asks shopping prompts on two consumer shopping surfaces, records which product cards were shown, and matches them against your catalogue.

It is product-level, and that is why it is its own section rather than a fold of AI Visibility. Every AI-visibility item is brand-level; this is a table of products with its own catalogue, its own cadence, its own allowance and its own competitor set.

The four screens

ScreenThe question it answers
Shopping OverviewDo our products appear at all, per surface
ProductsWhich of our products appear, on which questions
MerchantsWhose products appear instead of ours
AdsWhere advertising is observable, and where it is not

Two surfaces, and they are not equivalent

RankX AI observes ChatGPT Shopping and Google AI Mode. They differ in ways that change what a number means, and shopping surfaces sets out the detail. The one difference to carry everywhere: only one of the two exposes whether a result was an advertisement. A zero in the ads column on the other surface means "not observable here", never "no advertisers here".

Shopping prompts are a separate allowance

A shopping prompt is a shopper's question, and it is a different object from an AI-visibility prompt:

AI visibility promptAI Shopping prompt
Asked onThe five AI assistantsThe two shopping surfaces
MeasuresWhether your brand is namedWhether your products are recommended
ChargedPer prompt, per assistant, per runPer prompt, per surface, per run
Plan allowanceIts own capIts own separate cap

The two allowances do not draw on each other. Using all your AI-visibility prompts does not reduce what you can track here. Figures are in plans and limits.

The denominators, which decide what every figure means

Shopping visibility divides by answers that showed products at all, not by checks run. An answer that returned no product carousel has no opinion about your catalogue, so counting it against you would be inventing a loss.

Every check therefore falls into one of several states, and the screens keep them apart:

  • Checks run, the total.
  • Checks that produced an answer.
  • Checks that showed products, which is the denominator.
  • Checks that could not run.
  • Answers that included your products, which is the numerator.

A blank visibility figure means nothing showed a carousel in the window, so there is no denominator. It is never rendered as zero. See null is not zero.

An incomplete catalogue reads as a loss

The one measurement error that is yours rather than RankX AI's.

Matching works by comparing the product cards a surface showed against the catalogue RankX AI holds for your Website. A product missing from that catalogue cannot be matched, so an answer that recommended it is recorded as an answer that recommended somebody else.

Two consequences: your visibility reads lower than it is, and Merchants reads higher, because unmatched cards are what that list is built from.

Before drawing a conclusion from a low number, check the product catalogue.

Cadence and cost

Shopping checks are charged per shopping prompt, per surface, per run, so your bill is:

shopping prompts x 2 surfaces x runs per month

The cadence floor is your plan's, or 7 days while on trial, whichever is slower, on the same basis as every other check family.

Prices are on the credit cost reference.

Where to go next

  • Shopping surfaces, for what each surface can and cannot tell you.
  • Products, including why the catalogue matters more than it looks.
  • Merchants, for who wins the slots you do not.
  • Ads, and the zero that means "not observable".