Clarity finds checkout pages in 25% of AI agent sessions, versus 15% for humans

Clarity has published a session-level comparison that puts a number on agentic shopping: checkout pages appeared in 25% of AI agent sessions, against 15% of sessions driven by humans. PPC Land reported the finding, which comes from observed session behaviour rather than user surveys.

The metric records whether a session reached the checkout stage at all, not whether a purchase was completed. Read that way, it is a funnel indicator: agents appear more likely than humans to push a task all the way to the payment entry point.

The ten-point gap matters because it moves the agentic commerce debate from demos to measurement. For most of the past year, agent shopping has been discussed through product launches and proofs of concept, with little public behavioural data to compare against.

For retailers, the implication is a possible new layer in traffic structure. Agent sessions tend to be shorter and more goal-directed, which weakens traditional signals such as time on page and browse depth. Reaching checkout, checkout failure rates and payment conversion become the more useful observations.

Analytics and attribution vendors face a definition problem. It is still unclear whether agent traffic should count alongside human sessions, how a single task split across multiple requests should be stitched together, and whether different vendors are measuring the same thing. Ratios published without that context are hard to compare directly.

Caution is warranted. The report offers two percentages but no sample size, time window or industry breakdown. Without those details, claims that agent shopping has already gone mainstream are premature.

Two things to watch next: whether platforms publish their own agent-traffic shares to cross-check the figure, and how fraud controls, risk scoring and return policies adapt once agents routinely reach checkout. Those developments will say more about maturity than any single ratio.