Skip to content

Research 8 min read Updated September 26, 2026

What does AI cite when people ask where to shop online?

In short

On 2026-09-26 we ran five online shopping prompts through gpt-5.1 with web search on. It attached 46 citations, the most of any industry in the study, and 40 of them were to stores' and brands' own websites. A Vancouver pet supply prompt cited eleven retailers; a mattress prompt cited seven; Wikipedia appeared in four of five answers as background; review and comparison sites appeared mainly for the custom t-shirt prompt.

Key facts

  • Five prompts were run once each on 2026-09-26 through gpt-5.1 with the web search tool on: where to buy a quality mattress online in Canada, best online store for running shoes in Canada, good online furniture stores that ship to Toronto, best place to order custom t-shirts online in the US, and online pet supply stores with fast delivery in Vancouver.
  • The model attached 46 citations across the five answers, and by our sorting 40 were retailers' or brands' own sites, five were reference sites, and one was Reddit.
  • The Vancouver pet supply prompt cited eleven retailer hosts, from local pet stores to a grocery delivery service, and named them by delivery speed; the Toronto furniture prompt cited eight, including two national chains' Canadian sites and several smaller Canadian sellers.
  • en.wikipedia.org was cited in four of the five answers, used as background on brands and categories; a price comparison site was cited for running shoes and a savings site for mattresses.
  • The custom t-shirt prompt was the exception: its six citations were all review and comparison sites rather than printers, and the answer sorted printers by use case, one-off orders, bulk corporate orders and best value.

What we asked and how the study works

On 2026-09-26 we sent five online shopping prompts, one run each, to gpt-5.1 with its web search tool on, and recorded every URL the model attached to its answer. We sorted each cited host by type with our own rules: a store's or brand's own site, a directory or review site, a news, reference or government site, Reddit, or social. Our labels sit on top of hosts the model itself attached.

The prompts were written the way a shopper would ask: where to buy a quality mattress online in Canada, best online store for running shoes in Canada, good online furniture stores that ship to Toronto, best place to order custom t-shirts online in the US, and online pet supply stores with fast delivery in Vancouver.

What each prompt cited

Mattresses in Canada: twelve citations, to gotosleep.ca, hamuq.com, mattressmiracle.ca, nationalmattress.ca and mattressreviews.ca, plus save.ca and en.wikipedia.org. The answer sorted options by budget band and construction and named several Canadian brands.

Running shoes in Canada: eight citations, to ca.shop.runningroom.com, runasyouare.com, sourceforsports.ca, foveo.run, sneakerrank.ca, a comparison site, reddit.com and en.wikipedia.org. The answer grouped specialty stores with online shops, large sporting goods retailers and comparison sites. Furniture shipping to Toronto: nine citations, to article.com, cb2.ca, crateandbarrel.ca, furnberry.com, furnituredepot.ca, lumahome.ca, sixtrend.ca and en.wikipedia.org, grouped by budget, delivery speed and Canadian makers.

Custom t-shirts in the US: six citations, to printreviewer.com, thetestedhub.com, topconsumerreviews.com, podl.me and duneworld.org, all review or comparison pages, with the answer naming a printer for each use case. Pet supplies in Vancouver: eleven citations to eleven retailers, bluebarn.shop, bonesps.com, homesalive.ca, instacart.ca, kornapet.com, moonlightpetstore.com, nupetfooddelivery.com, truecarnivores.com, urbanpawpets.com, vanpetfood.com and wearemello.com, sorted by same-day, regional and Canada-wide delivery.

The pattern: the stores themselves get cited

Unlike the service categories in this study, where directories and review sites dominated, online shopping prompts were answered mostly from the stores' own websites. The model read product and shipping pages, named the retailers it read, and sorted them by price band, delivery speed or use case. Wikipedia supplied background on brands, and comparison sites filled in where the category is crowded with near-identical sellers, as with custom printing.

By our sorting, 40 of 46 citations were own sites, five were reference sites and one was Reddit. That is the highest own-site share in the study and the clearest case of a store's pages feeding the answer directly.

What an online retailer can do with this

The model cited pages that state plainly what a store sells, what it costs, and how fast and where it ships. For a Canadian retailer, a shipping page that says same-day in Vancouver or delivery to Toronto in words a model can quote is the difference between being named and being skipped. Category and collection pages that describe products by the attributes shoppers ask about, budget bands, materials, sizes, are what the model reads to sort stores.

For crowded categories where comparison sites take over, the store's presence on those review pages matters too. SearchPod builds category, product and shipping pages for ecommerce clients, sets up the product feed and tracking, and tracks which shopping prompts cite the store with a fixed prompt panel.

What this cannot tell you

Five prompts, one run each, one model, one day: that is the whole sample. The citations belong to this date, this model and this wording, and the type labels are ours. The stores and brands the model named are recorded only as what it said; nothing on this page verifies or endorses any of them or their prices.

Related questions

Sources

one model with web search, one run per prompt, one date; host types are sorted by SearchPod's own rules; names are recorded as the model wrote them and are not verified or endorsed; SearchPod sells AI visibility work.

Want to know how your business shows up?

Get a free, no-obligation proposal within one business day. We look at your site and your market and tell you plainly what we would do, and what we would not.

Get your free proposal

Keep reading

More in Research

All 37 in Research