Google Tests ?st_source=ai_overview Parameter On AI Overview Links
Google is testing ?st_source=ai_overview URL tracking on AI Overview and AI Mode links, first spotted by Tom Pool. If shipped, it would split AI-generated clicks from organic search traffic.
- Google is testing a ?st_source=ai_overview parameter on links inside AI Overviews and AI Mode.
- SEO practitioner Tom Pool first flagged the test in a LinkedIn post and shared multiple screenshots.
- Search Engine Roundtable's Barry Schwartz attempted to replicate the test across several browsers but could not.
- The parameter appears on inline response links and on the source citation list.
- Google has not publicly confirmed the test, which is consistent with a limited A/B rollout.
Google is testing a new URL tracking parameter — ?st_source=ai_overview — appended to links inside AI Overviews and AI Mode, a change that could finally let publishers separate AI-generated clicks from traditional search clicks in their analytics.
The parameter appears on links inside both the inline AI Overview copy and the source citation lists, according to Tom Pool, an SEO practitioner who flagged the test in a LinkedIn post. Pool wrote: "Just started seeing the ?st_source=ai_overview parameter on links mentioned within AI Overviews — is this the start of actual proper attribution reporting?"
What does the parameter actually do?
The string is appended to the end of any link Google surfaces inside its generative search surfaces. In practice that means:
- Every URL inside an AI Overview response carries the tag
- Links inside the "sources cited" panel carry it too
- The destination URL is otherwise unchanged from the publisher's perspective
If Google ships the test broadly, webmasters and analytics teams will have a clean signal to isolate traffic originating from generative surfaces rather than the traditional ten blue links. Today, publishers see only a generic Google referral, with no way to distinguish a click from an AI Overview, a Featured Snippet, or the main web results.
Who saw it — and who couldn't replicate it?
The signal comes from a single practitioner, not from Google. Pool documented the parameter across multiple screenshots and said he saw it "in links mentioned in the source list too."
Search Engine Roundtable's Barry Schwartz attempted to reproduce the test in several browsers without success. The mismatch between the two observers is consistent with a limited A/B rollout — a small slice of users or sessions receive the parameter while the rest continue to get clean URLs.
That distinction matters. Tool-signal observation should be treated separately from a confirmed Google statement. At this stage, Google has not commented publicly on the test, and the company has a long history of running quiet URL experiments that never reach general availability.
Why publishers have been asking for this
Since AI Overviews launched broadly, publishers have struggled to quantify the traffic impact of generative surfaces. Google Search Console does not break out clicks from AI Overviews as a distinct property, and most web analytics platforms only register a generic google.com referral with no surface-level detail.
A standardized parameter such as st_source=ai_overview would:
- Allow server-side filtering of AI Overview traffic in raw log files
- Enable direct comparison against the same page's organic click volume
- Feed cleanly into existing Google Analytics 4 segment workflows
Until now, the only ways to approximate the split involved reverse-engineering referrer strings, watching user-agent shifts, or buying third-party clickstream datasets — all fragile and incomplete.
Who is most affected?
Publishers with the highest exposure to AI Overview citations stand to gain the most from clean attribution. That group typically includes:
- News outlets and reference sites that Google cites most often
- E-commerce product pages that appear in shopping-adjacent Overviews
- How-to and recipe publishers competing for top-of-page placement
- B2B SaaS blogs cited for technical definitions
Sites that rarely surface in AI Overviews will see a smaller absolute change in their analytics, but a relative one — they will finally be able to prove the absence of generative referrals.
What to watch next
Three signals will indicate whether this is a real rollout or a one-off experiment:
- A Google Search Central post or documentation entry confirming the parameter
- Wider replication across geographies, devices, and logged-in versus logged-out sessions
- Adjacent parameters surfacing for other surfaces, such as AI Mode standalone results or Featured Snippets
For now, the parameter sits in the "test observed by one practitioner" bucket. SEOs running log-file analysis should already check their servers for st_source=ai_overview referrals and tag them as a separate channel. The string is stable enough to build a permanent filter around if Google ships it.
via seroundtable.com (Original)
More from Nathan Brooks
Show full bio
Senior reporter covering consumer brands and retail at SERP Journal.
63 articles