The Maze: An AI shopping answer often leaves the buyer with more to check. Two studies describe the activity around the answer, using different measures. That follow-on activity matters commercially: the place a shopper discovers a product can differ from the place that resolves enough uncertainty to win the sale.

🔎 AI users encounter more information. Peter Buckley reports 12.1 average touchpoints for AI users, versus 4.2 for other shoppers in a 25-market Google/Ipsos study. A touchpoint is an encounter with shopping information. The work is described as covering roughly 36,000 shoppers with December 2025 fieldwork. People who already research more may choose AI; the comparison cannot establish what happened because of AI.
👣 The answer leaves work to do. A separate 2025 US study found 3.8 online steps after consulting AI, versus 1.6 beforehand. These are journey stages, so they cannot be added to the touchpoint counts. The US work combines more than 450 recorded sessions and a survey of 600 people. Several quick checks could take less time than one slow search, so the step count alone cannot tell us whether AI saved time.
Our take: Treat the assistant’s answer as a handoff and test whether the next step helps shoppers decide.
🛒 Make the handoff usable. Buckley describes checks of prices, options and reviews on retailer pages. Give merchants and assistants consistent model, stock and price information so a shopper can act on the recommendation. If an assistant names an in-stock model and the merchant shows a different version, the shopper has to start checking again. Test whether fixing product matches improves first orders, rather than assuming another product-page visit means success.
🎯 Test what earns the sale. The IAB’s recorded journeys around an AI consultation make clear that other steps can intervene before a purchase. Test whether earlier exposures change orders before assigning all the credit to the final visit. A last-click report can tell a merchant where the order arrived. A controlled test across comparable audiences can answer whether an earlier exposure generated orders that would otherwise have been lost.
Sources: Peter Buckley; IAB and Talk Shoppe


