Can Search Console Separate AI Overview Traffic?

Search Console provides a dedicated view of generative AI impressions, but its documented reports do not isolate AI Overview clicks. If your SaaS guide appears in an AI Overview and Google traffic rises, those reports cannot tell you how many visits came from that appearance. Google’s generative AI report documentation

Google introduced the dedicated reports in June 2026 and says the insights rolled out worldwide on August 31. The reporting distinction matters: you can observe AI visibility separately, then compare it with overall search traffic and website outcomes. You cannot join those numbers into a fully attributed AI funnel. Google’s reporting announcement

For a small team, the practical task is to build a comparison that supports a decision about a page. The worksheet below keeps measured results, search observations, and estimates separate.

What Search Console can separate

The Generative AI performance report for Search includes impressions from AI Overviews and AI Mode. Its documented dimensions are pages, countries, dates, and devices. The help page does not list separate clicks, CTR, queries, or a breakdown between the two AI features. Label the metric “Search generative AI impressions” so readers know what it covers. Report dimensions and coverage

Google also provides a generative AI report for Discover. Use the Search report for this comparison and keep Discover results separate. Search and Discover reporting

The standard Performance report includes AI Overviews and AI Mode within overall Web search traffic. A page’s Web clicks therefore cannot identify the clicks from either AI feature. Google’s AI features guide

Use explicit labels in your dashboard:

If the dedicated report is missing, record “report unavailable.” Google lists insufficient impressions as one possible reason it may not appear; absence of the report does not establish zero visibility. Report availability

Why impressions and position cannot reveal AI clicks

Google counts a click when someone follows an AI Overview link to an external page. An AI Overview occupies one search position, and all its links receive that position. These counting rules describe how activity enters the report; they do not create a separate traffic segment. Google’s measurement definitions

More impressions and a better average position still cannot identify which clicks came from an AI Overview. Position describes placement, not the source of each visit.

Dividing overall Web clicks by generative AI impressions also fails to produce AI CTR: the clicks include activity outside the population represented by the impressions. Multiplying AI impressions by an assumed click rate gives you a planning estimate. If that estimate helps a decision, record the assumption and keep the result separate from measured traffic.

Compare reports around one page decision

Start with a decision you can act on: “Should we spend half a day improving this guide’s worked example and next step?”

Choose one page, or a fixed group of pages serving the same customer task. For an AI support tool, that might be guides about evaluating support automation. Keep pricing pages and unrelated tutorials in separate groups so their different purposes do not obscure the comparison.

Choose two equal periods, such as consecutive 28-day windows. Record the country, device, and search-type scope, and keep those settings consistent. For a US audience, use a US country filter where available. Note any scope you cannot match across reports.

Collect the observations in this order:

1. Record Search generative AI impressions for the selected pages.
2. Record overall Web clicks and impressions for those pages in the standard Performance report.
3. Record Google organic landing sessions and the chosen outcome event in Analytics. Google’s comparison guide uses Session source = google and Session medium = organic to select Google organic sessions. Apply the landing-page scope as well. Using Search Console and Analytics together

Before exporting, note unavailable values shown as ~ or -: Google’s AI report documentation says these become zeros in downloaded data. Keep a separate missing-data field in your worksheet. The same documentation distinguishes property totals from page totals, so use the same aggregation method in both periods. Export and aggregation rules

Search Console measures activity in search results; Analytics measures activity on your website. Clicks and sessions are calculated differently. Consent, tracking setup, time zones, and canonical URL handling can also create discrepancies. Check those differences before interpreting a gap between the reports. Google’s explanation of data discrepancies

Write the decision beside the numbers. More reported visibility might prompt an accuracy check on a page that is getting exposure. Fewer signup events might prompt a check of the form or the page’s relevance to the audience. Neither observation alone identifies an AI Overview effect.

Use search samples to investigate specific appearances

A manual search check can establish whether an AI Overview appeared and linked to your page in that observation.

Create a fixed list of customer questions. Record the exact query, date, country or locale, language, device, account state, whether an AI Overview appeared, and the linked URL. Save the answer or a screenshot so you can inspect the observation later. FindVex’s guide to measuring AI citations without fooling yourself provides a broader sampling routine.

These samples show what a customer might encounter under the recorded conditions. They cannot establish how often all customers saw the same answer or which link a visitor clicked.

You can group queries where you observed AI Overviews for further investigation, but their Search Console clicks remain blended. Queries where you did not observe an Overview may differ in intent, demand, ranking, or competition. Record those differences before using the comparison to propose an explanation.

Before paying for a tracking tool, ask what its “AI traffic” number represents: measured visits, sampled appearances, or a modeled estimate. The answer determines which decisions the number can support.

Worked example: visibility rises while clicks fall

Imagine an AI support SaaS tracking one guide across two 28-day periods. The numbers below are hypothetical and demonstrate the method; they are not customer results.

AI impressions increased 50% relative to the first period, while Web clicks fell 10%. Recorded signup events increased by two. Those changes do not identify AI Overview clicks or signups.

A report could say:

The guide gained reported generative AI visibility while overall Google Web clicks declined. Recorded signup events increased from 14 to 16. These reports do not isolate the traffic or signup contribution of AI Overviews.

Check whether the page’s query mix changed, whether the signup event still fires correctly, and whether the guide or signup path changed between periods. A two-event increase is too little evidence on its own to establish a durable improvement. Keep the label “signup events” unless you have separately verified the number of distinct signups.

If the guide contains an outdated product example, correct it and record the date. The reporting uncertainty does not prevent an accuracy fix. Compare another consistent period afterward, while keeping any claim about the cause of a change separate from the observed numbers.

Copy this measurement worksheet

Use one sheet per page or group serving the same customer task. Fill in the scope before entering results.

Before sharing the worksheet, check that missing values remain distinguishable from zeros and that each percentage names its denominator. Label modeled traffic as an estimate and preserve its assumptions.

Your next task: complete one page comparison

Choose one page that helps a customer make a decision. Record the report settings and missing values, then export two equal periods. Write one sentence describing what changed and one sentence explaining what the data cannot establish. Assign a concrete action and a review date.