The Maze: Retailer websites were named nearly four times as often as AI assistants in a summer 2026 survey of 1,000 UK adults. IMRG and Columbus asked which sources people were most likely to use to help decide on an online purchase. Respondents could name several. The answers describe where people say they would look for help, with room for a search engine, a shop and a retailer’s product page in the same journey.

The details
🛒 Retailer sites led the list. Retailer websites drew 62.1%, search engines 55.7% and customer reviews and ratings 45.8%. A product page can serve two jobs in that mix. It gives shoppers a place to confirm the price or size they saw elsewhere, and it can host reviews that help them choose. The survey did not measure how often those jobs happened on the same page.
🤖 About one in six named AI assistants. AI assistants reached 16.3%, online marketplaces 16.7% and physical stores 22.5%. The AI category includes general chat tools and a retailer’s own assistant. The 0.4 percentage point gap with marketplaces is too small to build a strategy around without evidence that it holds beyond the sample. About one in five respondents named a physical shop as a place they would look for help deciding on an online purchase.
Our take
A retailer can improve the places where shoppers settle a decision and still learn which channels bring them there. Those two jobs call for different tests.
🔗 Check where the recommendation lands. The report author says the research did not trace how shoppers reached retailer websites. An assistant might narrow a shopper’s choice to a specific laptop configuration while the link opens a page that defaults to a different memory size. For a small set of products, check whether assistant recommendations resolve to the correct, in-stock variant and whether visitors can confirm the quoted price. Use identifiable referrals to spot common failures. The survey cannot tell you what share of site visitors first consulted an assistant, and referral logs may miss visits where a shopper returns independently.
🏬 Make a store visit easier to plan. Physical stores were named by 22.5% as a resource for deciding on an online purchase. If a customer wants to try two sizes, knowing which branch has both can save a wasted trip. In categories where fit or feel matters, test variant-level stock and a clear way back to the same product page. Judge the test on completed orders and returns across channels. The survey offers a reason to test the handoff and leaves its sales and margin effects open.


