Back to blog

How to measure how much traffic I lost to Google AI

Discover the step-by-step guide to calculate the impact of Google's artificial intelligence features on your website's traffic and understand the real metrics.

September 7, 2026
9 min read
7 views
How to measure how much traffic I lost to Google AI

How to understand the drop in visits to your site

To understand how to measure how much traffic I lost to Google AI, you need to cross-reference Search Console data with Google Analytics 4 (GA4) metrics, as the platform's official report displays impressions but hides direct clicks generated by synthesized answers. Many small and medium-sized Brazilian companies have observed recent fluctuations in organic traffic and are unsure how much of this change stems from user behavior regarding the new artificial intelligence blocks in search.

Although there are widely publicized international percentages regarding click loss on the web, this data does not always reflect the reality of national websites. In Portuguese search in Brazil, there is no primary study published with open methodology, which makes it essential to know how to extract and interpret the numbers from your own domain to make strategic decisions based on facts and not generic estimates.

In this practical guide, you will learn how to use the available resources to map the impact of AI on your traffic, understanding exactly what data Google provides, what it omits, and how to circumvent these analytical limitations in your daily operations.

Key points

  • The new Search Console report counts impressions in AI Overviews, AI Mode, and Discover, but does not show direct clicks.
  • Pew Research Center data shows a significant increase in no-click sessions when automated summaries appear.
  • The absence of the report in your account usually indicates an insufficient volume of impressions, not a technical error in the registration.
  • The actual impact calculation requires exporting two Search Console reports from the same period and cross-referencing them with the final traffic in GA4.

The new Search Console AI report and its limitations

On June 3, 2026, Google announced the first tests of a performance report focused on generative AI features in Search Console, in a publication signed by Mrinalini Loew, general manager of the Google Search ecosystem. A few months later, on August 11, 2026, access to the feature was expanded to a larger number of properties in the tool. This update allows tracking impressions of pages displayed in AI Overviews, AI Mode, and AI blocks in Discover, grouping records by page, country, device, and date.

However, the report has severe analytical restrictions that hinder direct diagnosis. It does not show clicks generated from AI responses, omits data at the query level (the exact keywords typed), and does not separate traffic coming from AI Overview from that coming from AI Mode. If you notice that the report is not available in your property, the explanation is simple: John Mueller indicated that the absence of the report is usually linked to an insufficient volume of impressions in these formats, and not a technical system failure. Learn more details in the Conversion article on Search Console and AI search reports.

Another crucial point is that the data provided is not retroactive. It only starts to be counted from the date Google formally began separating these types of searches on its servers. Furthermore, the same impression can be counted simultaneously in the AI segment and the web search segment for the same query. The panel also does not distinguish when your link was directly cited within the text of the AI-generated response or if it merely appeared in the traditional list of results positioned below the synthesized block, as detailed in Hedgehog Digital's analysis of the Google Search Console generative AI report.

The real impact of AI on click-through rate

International studies indicate profound changes in how users interact with search results when a synthetic answer is displayed at the top of the page. On July 22, 2025, the Pew Research Center measured an 8% click-through rate on traditional links when the page presented an AI summary, compared to 15% clicks on pages without the summary. The same survey revealed that 26% of sessions ended with no clicks when AI was present, compared to 16% of no-click sessions in conventional searches.

The drop in traffic affects even websites that achieve the first organic positions. A study by Ahrefs, conducted with approximately 300,000 keywords, showed a drop in click-through rate (CTR, which measures the percentage of impressions that result in access) in position 1 of 34.5% in April 2025 and 58% in February 2026. Understand this dynamic better in Digital Applied's analysis of AI Overviews monitoring.

Metric analyzedSearch without AI summarySearch with AI summary
Click-through rate on traditional links (Pew Research)15%8%
Sessions ending with no clicks (Pew Research)16%26%
CTR drop in 1st position (Ahrefs - Apr/2025 vs Feb/2026)-34.5% (Apr/25)-58% (Feb/26)

It is fundamental to highlight that there is no primary study published with open methodology focused exclusively on Portuguese search in Brazil. Therefore, reproducing percentages from the North American market without testing your own domain can lead to incorrect diagnoses about the behavior of your local audience.

Geometric illustration representing the overlap of web search data and AI features.
Measurement requires cross-referencing separate reports to identify impression overlap.

Step-by-step: how to measure how much traffic I lost to Google AI in practice

Given the absence of direct click reports in Search Console, the solution for identifying real losses involves cross-referencing impression data with session behavior in Google Analytics 4. Follow the steps below to perform the measurement on your site:

  1. Export web search data in Search Console: Access the performance menu and select the Web search type. Choose a comparative period (e.g., the last 90 days compared to the previous 90 days) and export the table containing URLs, impressions, clicks, CTR, and average position.
  2. Export the AI features report: In the same panel, filter the specific view for generative AI features within the same date range and export the data by page. You will have a list of URLs with accumulated impressions in the AI block.
  3. Group URLs by content type: In a spreadsheet, categorize your pages by format or intent (e.g., blog articles, product pages, categories, or service landing pages). This prevents mixing drops in institutional terms with loss of informational traffic.
  4. Cross-reference impressions, clicks, and position page by page: Compare whether your URL's entry into the AI segment generated an increase in impressions accompanied by a sharp drop in CTR and clicks in the traditional web segment. If the average position remains high (between 1 and 3), but clicks have fallen, the presence of the AI block at the top is the most likely factor.
  5. Validate commercial impact in GA4: Open Google Analytics 4 and check organic sessions and conversions directed to these same specific URLs during the period. If the number of active users dropped in the same proportion as the CTR seen in Search Console, you have isolated the traffic loss caused by automatic responses. To deepen this type of behavioral analysis, see our guide on how to find out why people leave your site without buying.
Abstract block graph representing the variation in click-through rate.
The decline in click-through rate varies according to the position and type of content displayed.

What to do when you discover how to measure how much traffic you lost to Google AI

Identifying which pages experienced a drop in access does not mean that traffic was irreversibly lost. Once you know which URLs were affected by artificial intelligence blocks, the work shifts to content readjustment and information architecture.

Manual measurement of impressions and clicks is the only reliable way to differentiate a traditional SEO ranking loss from a traffic displacement absorbed by Google AI.

When an informational article loses clicks because the direct answer appears in search, the best strategy is to reformulate the text to offer in-depth data, original tables, or analytical perspectives that a synthesized answer cannot deliver. If your site operates in e-commerce, it's worth checking out the article on how to prepare your catalog for the age of artificial intelligence, adapting the information structure to remain relevant. Also read the recommendations from the Meu Redator portal on the drop in organic traffic with AI Overviews.

Frequently asked questions

Why doesn't the AI report appear in my Search Console?

The absence of this report in your panel is usually linked to an insufficient volume of impressions in Google's new AI features. As indicated by John Mueller, it is not an error in the tag or property installation, but rather a system limitation, which only allows display when the account reaches a minimum threshold of impressions in these formats.

Does the Search Console AI report show how many clicks my site received?

No, the report displayed by Google only shows page impressions in AI Overviews, AI Mode, and Discover. It does not count clicks generated within AI responses nor does it allow knowing how many people clicked on the links cited within the automatically generated text block.

Is AI traffic data in Search Console retroactive?

No, the data in the Search Console AI report does not work retroactively. Information only begins to be recorded from the date the tool started separating this type of search in the system for your specific property.

Conclusion

Measuring traffic loss to Google AI requires looking beyond the basic reports of a single dashboard. Since the official tool provides impressions but hides clicks, the only reliable way for small and medium-sized businesses in Brazil is to manually cross-reference URL performance data in Search Console with the volume of sessions recorded in GA4.

As a practical recommendation to apply today, choose the 10 pages with the highest access history on your blog or online store, export the reports from the last three months, and identify which ones experienced a sharp drop in CTR even while maintaining top positions on Google.

Share:
Lee Sugano

About Lee Sugano

Lee Sugano

Digital solutions agency based in Japan, serving clients in 10+ countries. We share insights on development, design and digital marketing for companies that don't settle for generic.

Enjoyed this content?

Receive exclusive insights about web development, design, and digital marketing straight to your inbox.

No spam. Unsubscribe anytime.