Full Answer

The 15–30% conversion rate makes sense once you understand what agent traffic actually is. A human browsing a store clicks through dozens of pages, abandons carts, comparison-shops across tabs — all of which drag the conversion rate toward 2–3%. An AI agent skips that entirely. It arrives with a specific product request, a price ceiling, and payment credentials already in hand. The session is short, deliberate, and almost always converts. Here's the thing: that efficiency is exactly why it distorts your data when it lands in the same analytics bucket as human traffic.

The practical problem is measurement. GA4 files most agent visits under direct or unattributed traffic because AI agents do not carry standard UTM parameters or referrer headers. Your blended conversion rate goes up, but you cannot attribute the improvement to a channel, a campaign, or a product change. Worse, if agent traffic drops — because a model update changes purchasing behaviour — your metrics fall and you have no idea why.

Separating agent and human cohorts in your analytics is no longer optional. Server-side tracking can identify agent user-agents at the request level, before the session reaches GA4, letting you build a clean segmentation. The full breakdown of how WooCommerce 10.3 handles agent purchases explains the tracking gap in detail, and the attribution analysis shows how GA4 misclassifies the revenue.