How do I compare landing pages for AI assistant visits in GA4?

The Citead team, updated

In short. Compare AI assistant landing pages in GA4 by adding Session source / medium and applying comparisons to recorded Session source values.

Definition: A landing page is the page associated with the first pageview in a session, a period of interaction with a website. Comparing AI assistant visits in Google Analytics 4 (GA4) means examining these entry pages with session-source breakdowns or comparisons that select recorded assistant sources.

Key facts

Where is the landing-page source breakdown?

Google's Landing page instructions give this report path:

Reports > Engagement > Landing page

Adding Session source / medium as a secondary dimension displays sources for each landing page. The landing-page dimension groups the path and query string associated with the first pageview.

Google's Traffic acquisition documentation defines Session source as the publisher or inventory origin of traffic. Session medium describes the acquisition method.

How are assistant sources compared side by side?

Google's comparison documentation provides the following workflow, applied here to assistant-source analysis:

  1. Open the comparison control at the report's top right and select + Create new.
  2. Choose Session source and exactly matches, then select an assistant source value recorded in the property.
  3. Select Apply, then create another comparison for a different recorded source.

Each comparison shows its source across the landing-page rows. Multiple values in one condition use OR logic, allowing a combined assistant subset; separate conditions use AND logic.

The contains match is case sensitive. Only dimensions with recorded values appear; incompatible dimensions are grayed out.

How does the AI Assistant channel differ from a source?

A channel groups traffic using rules. Google's default channel definitions classify AI Assistant traffic when the medium equals ai-assistant. Analytics sets this medium when a referrer matches its assistant list. A comparison on Session default channel group, with AI Assistant selected, therefore provides a channel-based subset alongside source-specific comparisons.

Google's scope documentation explains that Session source describes acquisition for new sessions, including returning visitors. First user source retains the original acquisition value when a visitor returns. Session source therefore answers the question about the visit's origin.

Which figures describe the selected visits?

Google's Traffic acquisition documentation defines Sessions as visits beginning on the website or app. Average engagement time per session measures time when the website has browser focus. These metrics also appear in the Landing page report and describe volume and engagement within each comparison.

A key event is an interaction marked as important to the business. The same documentation defines Key events as the count of those occurrences and Session key event rate as the percentage of sessions containing one. The Traffic acquisition report offers this rate for source-oriented outcome analysis.

FAQ

Can I start with sources instead of landing pages?

Yes. Google's Landing page documentation describes the reverse layout: in Traffic acquisition, select Session source / medium as the primary dimension and Landing page + query string as the secondary dimension.

Why does my comparison disappear after changing reports?

Google's comparison documentation says that moving between reports unselects comparisons other than All users. You can reactivate previously applied comparisons using the control beside All users.

Sources

  1. Google Analytics Help, "[GA4] Landing page report", undated
  2. Google Analytics Help, "Apply comparisons to reports", undated
  3. Google Analytics Help, "Default channel group", undated
  4. Google Analytics Help, "Scopes of traffic-source dimensions", undated
  5. Google Analytics Help, "[GA4] Traffic acquisition report", undated

Check how a page reads to answer engines. Open the GEO score analyzer

Test a page and see what to fix first

The analysis usually takes less than 10 seconds and needs no account.

400,000+ pages scored in the public benchmark