Similarweb: ChatGPT recommendations drive site visits

Being recommended by ChatGPT translates into real website visits, according to a Similarweb study published on 23 June 2026. Users who saw a brand recommended by ChatGPT were about 2.5 times more likely to visit that brand’s site within a week than a competitor’s. Most of those visits never showed up as AI referral traffic, which means standard analytics are undercounting AI’s influence. The study covered US desktop activity, but the measurement gap it exposes applies to any brand tracking AI impact.

What happened

Similarweb tracked US desktop users who asked ChatGPT an industry-relevant question, received a specific brand recommendation, and then visited either that brand or a competitor within seven days, across finance, travel, and beauty. It excluded users who had visited the brand recently or named it in their prompt. Recommended brands saw roughly 2.5 times the visit rate of competitors. The visits were also higher quality: AI-influenced visitors viewed about 12 pages and stayed 11.8 minutes, against 6.5 pages and 5.6 minutes for others. Notably, 55.9% of AI-influenced visits arrived through search rather than a visible AI referral, so they look like ordinary search traffic in most analytics.

Why it matters

Brands recommended by ChatGPT were about 2.5 times more likely to win a site visit within seven days than brands that were not.

Similarweb, The Downstream Impact of AI Visibility

The hidden-attribution point is the sharp one. If AI recommendations mostly convert into branded searches, the standard dashboard credits Google, not the assistant that started the journey. Brands measuring AI impact by referral traffic alone are seeing a fraction of it. Similarweb notes the data shows correlation, not that AI directly caused the longer visits.

What this means for multi-location brands

For a brand with many locations, the takeaway is that AI assistants are now an early step in local discovery, and being the recommended name pays off downstream. That recommendation rests on the same foundation as local search: accurate, consistent, machine-readable presence across every location and platform. Make sure assistants can resolve your brand and locations correctly with clean business location data and Places AI, and treat ranking in AI search as a measured channel rather than a side effect. Watch for branded-search lifts after AI exposure instead of waiting for an AI referral line that may never appear.

The bottom line

AI recommendations move real, engaged traffic, and most of it hides inside branded search. For multi-location brands the job is twofold: be the brand AI recommends, and measure it by the branded-search lift it creates, not the referral line it rarely leaves.