A brand's own page was the cited source in just 31% of cases where an AI tool had already recommended that exact brand by name. A lot of AEO advice collapses "get cited" and "get recommended" into the same goal. They aren't. A citation tells you an AI tool found and read a page. A recommendation tells you it named your brand as the answer. New ecommerce research shows exactly how far apart those two outcomes can be.
A brand can be the answer and still not be the source
Only 2.8% of AI citations pointed to a brand's own page, even inside the cases where that brand was the one being recommended. Shero Commerce, a Shopify agency, examined purchasing questions across 60 product categories and gathered every source cited by Google AI Mode, ChatGPT, and Perplexity in response, covering 1,851 cited sources drawn from 8,573 product descriptions across 883 Shopify stores. Inside the narrower set of 159 cases where an AI tool named a specific brand as the recommended option, the brand's own page was still the cited source in only 31% of them.
Both ChatGPT and Perplexity recommended Gymshark, Alo, and Beyond Yoga by name for a "squat-proof leggings" query, and none of the three brands supplied the source cited underneath that recommendation - credit went to third-party sites instead. A shopper reading that answer sees the brand's name in the recommendation and a stranger's website in the footnote.
Most of the buying research happens somewhere else
Review and editorial sites accounted for 59% of all citations in Shero's data, the largest single source category by far - the study names Good Housekeeping, Verywell Fit, and Reviewed.com as examples. That's the content actually doing the work of convincing an AI tool a product is worth naming, even when the brand itself never gets the link.
Google AI Mode named or cited a sampled brand in only 9.5% of the 60 categories tested, and roughly a third of those categories returned no sampled brand at all. Getting recommended isn't guaranteed just by showing up in a category. On Google's newest AI product specifically, most categories checked produced no brand recommendation whatsoever.
What this changes about where the effort goes
Off-site mentions correlate with AI visibility at 0.66 to 0.71, against just 0.19 to 0.29 for backlinks, according to Ahrefs' analysis of 75,000 brands. Shero's data adds the missing half of that picture: third-party mentions aren't only a stronger lever than links, they're often the entire citation, even in the exact moment your brand is the one being recommended.
Most AI visibility tools already separate a brand mention from a source citation, and this is exactly why that distinction earns its place in a report. A brand can win the recommendation and lose the citation in the same answer. Track whether your brand gets named. Track separately whether your own page ever gets cited. The two move independently, and treating them as one number hides which lever actually needs pulling.
A few common follow-up questions
If our own site is rarely the cited source, is there any point improving it?
Yes, since a page still has to be strong enough for a third-party reviewer to check it, quote it, and recommend it in the first place. The site stops being the finish line and becomes one input among several.
Does this apply outside ecommerce?
Shero's data is specific to purchasing questions across Shopify stores, so the exact percentages won't transfer directly to a B2B or service business. The underlying pattern, that a citation and a recommendation are separate outcomes, holds regardless of category.
How do we get onto the review and editorial sites that are actually getting cited?
Start with whichever sites already appeared as citations for your own product category during a manual check of a handful of buyer questions in each platform. That tells you which specific publications to pitch, rather than guessing at PR coverage in the abstract.