Loading blog content...
Loading post title...
About the Author
Understanding how much traffic Google AI Overviews send to websites is still difficult.
Google Search Console does not provide a dedicated signal that clearly separates AI Overview traffic from other organic traffic. That makes it harder for SEO teams to determine which pages are receiving clicks from AI Overview citations and how that traffic is being reported in analytics.
To better understand the impact, one transportation-industry brand tracked AI Overview referral activity for nine months.
From September 2025 through June 2026, the tracking captured 51,200 events across 1,661 cited snippets.
The results showed that AI Overview traffic can fluctuate significantly, citation activity is concentrated among certain snippets and a portion of the traffic may be incorrectly classified as Direct.
How the AI Overview Tracking Worked
The tracking relied on a URL fragment that Google sometimes adds when someone clicks a cited snippet inside an AI Overview.
The fragment is:
#:~:text=
A custom dimension was created in Google Analytics 4 to identify sessions arriving with this fragment.
This provided a way to surface potential AI Overview referral activity without a dedicated AI Overview traffic report in Search Console.
The tracked snippets were then grouped into different topic categories to identify patterns in citation and traffic activity.
However, the method has limitations, which are discussed later in the article.
A Small Number of Snippets Generated Most of the Activity
The data showed that AI Overview traffic was not evenly distributed across the 1,661 tracked snippets.
The highest-performing snippet generated 2,276 events.
Across all 1,661 snippets, the average was 31 events per snippet.
This difference shows that some snippets generated considerably more activity than others.
The tracking also revealed different snippet lifecycles.
Some snippets performed strongly during a particular period before declining, while others began gaining activity months after publication.
The source links these changes to factors including seasonal shifts in query intent and content becoming less aligned with what Google prefers to cite.
Which Content Was Being Cited?
The tracking showed different citation patterns across content categories.
Transfer time and pricing content
Transfer time and pricing pages were receiving frequent citations and showed upward momentum.
The data suggested that these existing assets could benefit from being refreshed and expanded.
Destination guides
Destination guides were performing below what the data suggested they could potentially contribute.
Transport comparison tables
Structured transportation comparison tables stood out in the dataset.
Tables presented as actual HTML tables were receiving citations at a relatively high frequency.
Together, these findings point toward a preference for content that provides specific and structured information.
Examples in the dataset included:
- Times
- Prices
- Named routes
- Comparisons
- Direct answers to specific questions
AI Overview Traffic Was Not Always Reported as Organic
One of the most notable findings came from analyzing traffic attribution.
Some traffic arriving through AI Overviews was being classified as Direct in GA4 instead of Organic Search.
Across the full dataset, the average misattribution rate was 22.4%.
That amounted to 11,468 AI Overview events being attributed to Direct rather than Organic.
The percentage changed from month to month.
Highest misattribution rate
May 2026: 29.3%
Lowest misattribution rate
April 2026: 16.8%
For SEO reporting, this creates an important consideration.
If AI Overview traffic is being placed into the Direct channel, the organic channel may not reflect the full amount of traffic generated through Google Search.
AI Overviews Accounted for 7.53% of Organic Sessions
The same dataset was used to calculate the proportion of organic sessions associated with AI Overviews.
Across the full tracking period, AI Overviews accounted for 7.53% of organic sessions.
But the figure was highly variable.
During February and March 2026, the share reached approximately 16%–17%.
More recently, at the time of writing, it had fallen to approximately 2%–4%.
This means the 7.53% figure should not be interpreted as a permanent share of organic traffic.
The data shows that AI Overview visibility can change substantially between periods.
AI Overview Visibility Can Rise and Fall
The tracking showed that individual citations can have different lifecycles.
A snippet may perform well for a period and later lose activity.
Another snippet may begin gaining visibility months after the content was published.
The source identifies several factors that may influence these changes, including:
- Seasonal changes in query intent
- Content freshness
- What Google prefers to cite
- Query type
- Google’s confidence in available content
- Broader algorithmic changes
This makes ongoing monitoring useful when evaluating AI Overview performance.
Why Content Specificity Matters
The content patterns in the dataset suggest that highly specific information can perform well in AI Overview citations.
The strongest examples included content focused on concrete information such as:
Times. Prices. Routes. Comparisons.
This differs from broad editorial content that may not directly address a clearly defined question.
The transportation comparison tables were another example.
Their strong citation frequency suggests that the way information is structured can matter alongside the information itself.
There Are Two Important Measurement Limitations
The tracking approach provides useful data, but it is not a perfect measurement system.
1. The URL fragment is not unique to AI Overviews
The #:~:text= identifier can also be used by Featured Snippets and People Also Ask results.
This means some of the tracked activity could come from those search features.
The research checked Featured Snippet exposure in Ahrefs and found fewer than 30 inclusions at the time of checking.
Based on the volume and tracking period, the source believes AI Overviews were the dominant driver, but the measurement cannot be considered completely clean.
2. Events and sessions are different measurements
The custom GA4 dimension operates at the event level.
The AI Overview share calculation compares those events with sessions.
That is not an ideal comparison.
The source describes the result as directionally accurate but notes this limitation for anyone attempting to replicate the approach.
What This Means for SEO Reporting
The data highlights a potential problem with relying solely on standard analytics channel reporting.
If some AI Overview referrals are classified as Direct, an SEO team could see less organic traffic than the site actually received from Google Search.
In this dataset, the average misattribution rate was 22.4%.
The rate also changed significantly between months, ranging from 16.8% to 29.3%.
That makes it useful to investigate how AI Overview referrals are being classified rather than assuming all Google-generated traffic is being accurately categorized.
What This Means for Content Strategy
The tracking also provides a way to connect citation activity with content decisions.
The data showed strong activity around transfer times and pricing content, while destination guides were underperforming relative to what the data suggested they could potentially drive.
This can help identify where existing content may need attention.
The source specifically points toward:
- Refreshing content that is already receiving citations
- Expanding relevant content
- Monitoring snippets that are gaining or losing activity
- Paying attention to specific information
- Using structured comparison formats where appropriate
The goal is to make content prioritization more grounded in observed citation activity.
AI Overview Traffic Should Be Viewed Over Time
The 7.53% average across the full period does not tell the entire story.
AI Overview traffic reached approximately 16%–17% of organic sessions during February and March 2026.
It later declined to around 2%–4%.
That range shows why a single month should not necessarily become the benchmark for future performance.
Tracking the data across multiple periods provides a better view of how AI Overview activity changes.
Conclusion
Nine months of AI Overview tracking provide a useful look at how Google’s AI search experience can affect organic traffic and reporting.
The dataset recorded 51,200 events across 1,661 cited snippets, with AI Overviews accounting for 7.53% of organic sessions across the tracking period.
But that percentage varied considerably, reaching approximately 16%–17% during February and March 2026 before falling to around 2%–4% more recently.
Attribution was another major finding. 22.4% of tracked AI Overview events were attributed to Direct rather than Organic, representing 11,468 events.
The data also showed that citation activity can change over time and that specific, structured content including times, prices, routes and comparison tables performed strongly in this dataset.
For SEO teams, the takeaway is not to treat AI Overview traffic as a fixed number. Track it over time, examine attribution and use citation patterns to inform content decisions.
Frequently Asked Questions (FAQs)
How much AI Overview data was tracked?
The project tracked 51,200 events across 1,661 cited snippets between September 2025 and June 2026.
What percentage of organic sessions came from AI Overviews?
AI Overviews accounted for 7.53% of organic sessions across the nine-month tracking period.
When did AI Overview traffic reach its highest share?
The share reached approximately 16%–17% of organic sessions during February and March 2026.
What was the more recent AI Overview traffic share?
At the time of writing, AI Overview traffic had fallen to approximately 2%–4% of organic sessions.
How much AI Overview traffic was attributed to Direct?
The average misattribution rate was 22.4%, representing 11,468 events attributed to Direct instead of Organic.
Which month had the highest misattribution rate?
May 2026 had the highest rate at 29.3%.
What types of content received frequent AI Overview citations?
Transfer time and pricing content received frequent citations with upward momentum. Structured transportation comparison tables also showed strong citation frequency in the dataset.
Is the AI Overview tracking method completely accurate?
No. The #:~:text= fragment is also used by Featured Snippets and People Also Ask. In addition, the calculation compares event-level and session-level measurements. The source considers the approach directionally useful but not perfect.
Built to scale from MVP to market dominance
Ready for a Partner That Gets It?
Stop coordinating between your web developer, SEO consultant, and PPC manager. Get one team that owns the entire funnel from code to conversion.
Book a CallBuilt to scale from MVP to market dominance
Ready for a Partner That Gets It?
Stop coordinating between your web developer, SEO consultant, and PPC manager. Get one team that owns the entire funnel from code to conversion.
Book a Call