We logged every source 5 AI engines cited for buying questions. The store never came up.

For three weeks we pointed ShelfSight at a snowboard store (our own test store, disclosed below) and asked ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews the questions its shoppers would ask: things like "what's the best snowboard for a beginner?" Four buying questions, five engines, scanned on a schedule from June 12 to July 2.

Then we counted every source the engines cited while answering.

The numbers

• 77 answers collected across the five engines
• 57 of them carried citations, 661 citation links in total
• 124 of those were Google's own redirect infrastructure, which we exclude, leaving

537 genuine source citations

• Citations pointing at the store being scanned: 0

Zero. Not "rarely". Never, in 537 chances over three weeks.

Where the engines actually looked

The top product-relevant sources, with verbatim counts from the scan rows (we also saw facebook.com at 14 and google.com at 13, left out of the table as generic platform domains rather than product sources):

| Source | Citations | | --- | --- | | evo.com (retailer with a huge buying-guide library) | 43 | | whitelines.com (snowboard magazine) | 42 | | youtube.com | 37 | | snowboardingprofiles.com (review site) | 22 | | outsideonline.com (outdoor publisher) | 21 | | thegoodride.com (snowboard reviews) | 16 | | burton.com (brand site) | 13 | | gearjunkie.com | 11 | | tactics.com (retailer with guides) | 11 | | rei.com (retailer with guides) | 10 |

Three patterns worth staring at:

1. Engines cite the sources they trust, not the stores they recommend. Review sites, magazines and YouTube dominate. When an engine names a product, the evidence it shows the shopper is almost never the product page. It is somebody else's writeup.

2. The one retailer that dominates the list earned it with content. evo.com is not cited 43 times because it sells snowboards. It is cited because it publishes one of the best buying-guide libraries in the category. It made itself the source the engines reach for, so it gets pulled into answers it never paid for.

3. The answers had no room for a store the sources ignore. Every one of the 537 citations went to a page that reviews, ranks, demonstrates or explains products. Product pages and store homepages are close to absent from the list, and not just ours. If the sources in your category have never mentioned you, the engines assembling answers from those sources have nothing to cite you for.

The honest caveats

This is one store, one category, four questions, three weeks. The store was our own test store running a standard demo snowboard catalog, so the engines had no history with it and its zero is the floor, not a prediction for your store. Snowboarding also has an unusually strong review-site ecosystem, and other categories will skew differently. What we stand behind is the shape of the leaderboard: the engines visibly assemble buying answers from a short list of trusted sources, and either you are on that list or someone else is. We are not claiming your store is cited zero times. We are claiming you should find out.

What to do with this

The worklist follows directly from the leaderboard:

1. Find out which sources engines cite for your category. Not in general. For

your actual buying questions. That list is your PR target list, your guest-content target list, and your "get reviewed here" list.

1. Look at whether a retailer-as-publisher play is open. If nobody in your

category is doing what evo.com does, that slot is available.

1. Fix your feed first. Citations decide who gets talked about; feed data decides

whether AI shopping surfaces can read your products at all. GTINs, specific product types and real descriptions are the entry ticket.

We run this exact citation analysis for any Shopify store, free, on your public product feed. No install, no signup, about a minute: Run the free AI visibility report.