Microsoft Clarity: checkout shows up in a quarter of AI agent sessions.

Microsoft Clarity says checkout was recorded in 25.21% of website sessions involving AI agents, against 14.65% of human-only sessions on the same sites. The study, published on 29 October, covers more than 5.2 million sessions involving AI agents across 230,511 Clarity projects.
The data runs from 13 August to 9 September 2026. Clarity identified the sessions by detecting agentic browsers from the OpenAI, Google and Anthropic ecosystems, then compared which of its Smart Events fired. Form submissions were close (19.85% for agent sessions, 18.56% for people). Outbound clicks went the other way: 26.99% of human sessions against 14.28% of agent sessions. Appointment bookings appeared in 8.42% of agent sessions and availability checks in 2.41%, while neither made the human top ten. Microsoft says agent session volume in its data rose 88% in three weeks, and it has made a new Agent Activity Dashboard generally available for sites that get agent traffic.
Microsoft adds its own caveats. A session can include a handoff to the person, and Clarity can't split the agent's actions from the human's, so the figures describe activity in sessions involving agents, not tasks an agent finished alone. It also says the events shouldn't be read as steps in one conversion funnel.
Some of the maths is unclear. PPC Land worked out that the published counts and percentages imply a base of about 1.22 million sessions rather than 5.2 million, which the study doesn't explain. PPC Land also notes there's no breakdown by agent provider, industry, country or device, that the detection method isn't described, and that Meta's Muse agent isn't among the named ecosystems; the window closed a day after Muse's US release, according to PPC Land.
Our read: brands in Saudi Arabia and the UAE that feed checkout, lead and booking events into Meta, Google, Snap or TikTok bidding should start checking how much agent traffic reaches those pages. The study has no regional split, so how far the pattern holds in the Gulf isn't known yet.
Why it matters
Checkout, form and booking events are what ad platforms optimise on. If agents trigger a growing share of them, conversion rates and cost-per-lead figures start mixing two kinds of visitor.
Related trends.

Meta tests two AI apps: a virtual pet and a wardrobe stylist
Meta has released two AI apps in a few Asian markets: Waddle World, a virtual pet game for children, and Style Sketch, which suggests outfits from clothes people own. Both come from Meta's push to ship more test apps using its own AI tools.
Why it mattersStyle Sketch turns Instagram photos into a wardrobe inventory, a data source that could later shape fashion ads and shopping features.

StackAdapt lets brands retarget ChatGPT ad viewers on TV and display
Programmatic platform StackAdapt says advertisers buying ChatGPT Ads through it can now retarget people who saw those ads across connected TV, display and its other channels. It is one of the first moves to carry ChatGPT ad exposure beyond OpenAI's own product.
Why it mattersIt is one of the first partner offers to take ChatGPT ad exposure out of ChatGPT, which changes how that spend can be planned and judged.

Zuckerberg says shopping is the top use for Meta's Muse agent
Meta's Muse AI agent has been downloaded more than 10 million times since its US and Canada launch last month, and shopping is its most common use. Mark Zuckerberg said average purchases through Muse are above $200.
Why it mattersMeta wants its AI agent to sit between shoppers and brands, which would shift commerce from websites to agent-to-agent deals.


