Conversion suggests a later-stage visit
In Similarweb's June 2025 ecommerce estimate, AI-referred visits converted about 23% above paid search and a little more than twice the rate of organic search. Those are BGraph calculations based on the published channel rates.
A plausible explanation is that some comparison and recommendation work happens inside the assistant before the visitor reaches a merchant. The public figures do not prove that mechanism, but they show why AI referrals should not be judged by traffic volume alone.
Engagement improved as the channel scaled
Adobe reported a twelvefold increase in AI-driven retail traffic from July 2024 to February 2025. In February, AI referrals had a bounce rate 23% below non-AI referrals. Adobe also reported that the conversion shortfall narrowed from a much wider gap in 2024 to 9% below other traffic in February 2025.
This is a different dataset with different definitions. Its value is directional: more AI traffic did not automatically produce weaker engagement.
Keep scale and quality on separate axes
A channel can have a high conversion rate and still contribute little total revenue if its visit volume is small. It can also influence journeys without receiving the final click. A useful market view needs both the size of the channel and the behavior after arrival.
For category analysis, compare AI, organic search, paid search, direct, and referral traffic using the same outcome window. Keep source-specific estimates separate until their populations, devices, geographies, and conversion definitions can be aligned.
- Referral volume
- Landing-page mix
- Engagement depth
- Conversion and repeat behavior
- Assisted journey position