Bazaarvoice, jointly with retail media analyst Kiri Masters, released a Brand Equity Survey of more than 3,600 shoppers across eight categories in the US and EMEA finding that 94% do outside research after receiving an AI shopping recommendation. The headline is less the AI hype than the friction: the recommendation is a shortlist candidate, not a verdict.

Click behavior confirms it. When an AI tool presented several options, 65% of shoppers clicked the brand they already knew, 24% clicked the AI’s labeled top match, and 11% chose the cheapest. Brand equity, not machine ranking, is doing the sorting.

The verification step is expensive. One-third of respondents spent over an hour vetting an AI-recommended product; 76% wouldn’t buy an unfamiliar brand without customer photos or videos, and 75% required written reviews. Even price barely closes the gap: 57% said an AI-recommended generic would need to be at least 50% cheaper before they’d switch.

“An AI recommendation gets a brand on the radar, but real customer photos, ratings, and reviews are what close the deal,” said Jo Callahan, VP of Brand Strategy and Experience at Bazaarvoice.

The distribution problem is technical too. Bazaarvoice noted on September 1 that most review widgets render through JavaScript that AI crawlers skim past, and reported clients using its structured-data infrastructure saw a median 40% lift in AI-driven referral traffic. That echoes findings that AI referrals convert at higher rates but arrive in smaller volumes, which makes each visitor’s trust check decisive.

The Agile Brand Guide flagged that the Influenster-sourced sample skews toward active review readers and writers, so direction is more reliable than magnitude. Even so, the operational read is clear: AI search visibility and photo-review depth are one budget line, not two. Sinch and Cordial released their own holiday-behavior research the same day, offering adjacent benchmarks to test the arithmetic against Q4 traffic.

Sources