AI shopping visibility: whether AI recommends your products.

AI shopping visibility is whether your products appear as product cards when a shopper asks an AI surface what to buy. The unit is the card, not the sentence: a card carries an image, a price and a buy link, so a brand-mention tracker can score well while your products never appear at all. RankX AI measures the cards on ChatGPT Shopping and Google AI Mode, matched against your own catalogue, over a 28-day window so single odd runs cannot swing the picture.

Ask ChatGPT a category question about what you sell and look for the product carousel, then ask again tomorrow: in RankX AI's own measurement the same question came back with a different carousel verdict 24% of the time. To check properly, connect or upload your catalogue, track the questions shoppers ask, and read the rate over 28 days. RankX AI reports which of your products appeared, in which slot and at what price, which merchants appeared instead, and on ChatGPT Shopping whether a card was sponsored.

Measurably, yes. Adobe measured AI-referred traffic to US retail sites up 138% year on year in May 2026, converting 54% better than traffic from other sources, and up more than fourteen-fold since it began counting in October 2024. Salesforce measured AI and AI agents influencing around a fifth of global online sales, 262 billion dollars, across the 2025 holiday season. Both surfaces now take a shopper past the recommendation: OpenAI's Instant Checkout buys inside ChatGPT, and Google's Universal Commerce Protocol puts a checkout on eligible products in AI Mode, in the US first, which is what agentic commerce means in practice. The volumes are still small next to Google, and the honest case is the direction: product discovery is moving into the answer, the buyers there convert, and early placement is cheap to measure.

ChatGPT Shopping and Google AI Mode: the product carousel in ChatGPT's answers and the shopping results inside Google's AI Mode. Google AI Mode is the conversational answer surface, not the AI Overview above a results page, and those two are the shopping surfaces a data provider can observe at industrial quality today. Each tracked prompt is checked on both.

OpenAI states that shopping results are chosen independently and are not ads, built from product data in merchant feeds and third-party sources and picked for relevance to the question, so a recommendation is earned rather than bought. Ads, where ChatGPT shows them, sit below the answer and are labelled as sponsored. What RankX AI can add is measured rather than stated: category and product questions produced a carousel four times in five, questions about a brand by name never, and the same question does not reliably return the same cards. Complete product data, in the feed and in the structured data on the product page, is the part a merchant controls.

Not today, and the honest reason is the data rather than the shopping. Perplexity does show product cards, but no industrial data provider sells a structured feed of them, and RankX AI will not run a scraping fleet of its own to guess at one. No provider exposes a shopping surface for Claude at all, and Anthropic's September 2026 commerce work is agent blueprints for retailers to run on their own sites rather than a shopping answer inside Claude. Whether any of those assistants name your brand in written answers is a different measurement, and AI Visibility tracks that across every assistant RankX AI covers. When a surface can be observed properly, it is added as a surface, not folded into an average.

The unit, the surfaces and the window. AI visibility asks whether an assistant or answer engine names your brand in a written answer, across ChatGPT, Claude, Gemini, Grok, Perplexity and Microsoft Copilot, and is measured as a mention rate over 30 days. AI shopping visibility asks whether a surface shows your product as a card, on ChatGPT Shopping and Google AI Mode, matched against your catalogue over 28 days. A brand can score well on the first while its products never appear on the second, which is why RankX AI reports them apart and never blends them.

No. AI Shopping needs a product list, not a platform, and a file is enough. Both platforms do connect, and they connect for different jobs: a connected WooCommerce store keeps this product list current on its own, while a connected Shopify store is read and fixed as a store, its SEO metadata, its image alt text and its redirects. For AI Shopping specifically, a Shopify export or a Google Merchant Center feed is read as it comes, so nothing has to be reformatted first.

5 credits per shopping prompt, per surface, per run. Ten tracked shopping prompts checked on every surface is a 100-credit sweep, and the 7-day trial's 1,000 credits cover 200 checks. Reading the results costs nothing: the Products, Merchants and Ads views are arithmetic over answers already collected.

Yes, with one honest caveat. The catalogue can be a file of a dozen products, ten prompts is a 100-credit sweep on Starter, and the questions are written from what you sell rather than from a category list. The caveat is the surface, not the tool: a niche question may never trigger a carousel on either surface, and RankX AI reports that as blank rather than zero, because it means the surface had no product opinion on that question yet. That is worth knowing before a rival fills the slot, and it costs nothing to read.

Partly, in five countries. Google's AI performance insights report in Merchant Center shows a share of voice for your free listings in AI Mode and AI Overviews, against competitors you define in Merchant Center, for English-language queries in Australia, Canada, India, New Zealand and the United States. It does not cover the UK, ChatGPT Shopping, merchants outside your defined set or sponsored placement, and it reports impressions rather than which products appeared in which answer. RankX AI reads the answers themselves, on both shopping surfaces, in the market you choose.

No, and deliberately. A price change is a revenue decision and an invented attribute would be fabrication, so no shopping finding is ever auto-applied. What you get instead is a diagnosis: whether a missing product is a feed problem or a listing problem, and an honest refusal to claim either when the evidence cannot support it. Acting on that diagnosis is a separate step you approve yourself, in Chat.

Yes, on approval. Chat reads your shopping visibility for free, and in Act mode it proposes the rewrite as a card showing the exact before and after, applying it only when you approve. It covers one WooCommerce product's SEO title, meta description and short description. It can never change the price, the stock, the status or the product URL, because those inputs do not exist on the tool, and the price is reported before and after so you can see it did not move. If the product is edited in WooCommerce while the card is waiting, the write refuses rather than overwriting that edit.

Never. An ad is bought, not earned, so sponsored cards are reported in their own view and excluded from every visibility figure. The Ads view also separates cards actually shown from cards returned but not rendered, because only a card a shopper saw is placement worth paying for. In the sample scenario that is 12 sponsored cards, counted apart.