Ahrefs told its customers on October 22 that Google has been “poisoning” the search results it serves to suspected SERP data providers, showing them different results from the ones real searchers see. Depending on the data provider, Ahrefs said, between 5% and 20% of search results are affected — the widest range any vendor has put on the problem so far.
The post, published on the Ahrefs blog under the title “An update on Google’s poisoning of search results,” is short and gives no methodology, sample size or start date. What it confirms, from a vendor of Ahrefs’ scale, is that the swapped results other trackers have reported since September are an industry-wide problem rather than a fault in any one tool — and that the ranking history those tools stored during the affected weeks cannot simply be trusted as recorded.
What Ahrefs Told Its Customers
“Over the past few weeks, we’ve witnessed an increase in the ‘poisoning’ of search results by Google, where different results are shown to suspected SERP data providers,” the company wrote. “Depending on the data provider, we’re seeing between 5% and 20% of search results affected in this way.”
Ahrefs described a three-part response. It already runs multiple filters designed to detect and reject abnormal results and request accurate replacements. It is also going back over results it has already stored and cleaning up any that may have been affected.
The company offered no view on why Google is doing this, and Google has not commented publicly on any vendor’s account of the poisoning. Ahrefs closed with a hedge: “Search engines regularly make changes, and what works to get accurate data today may not work tomorrow.”
How The Poisoning Works, And Who Saw It First
A poisoned results page looks normal. That is what makes it different from the blocking trackers have dealt with for years: a blocked request leaves a visible gap, while a swapped page stores a confident, wrong ranking. Vendors say the pages served to suspected automated collection swap in unrelated results — often forums and social sites — while tracked sites drop out of top tens they actually hold.
Ahrefs is late to speak, not first. Nozzle dated the first poisoned pages in its data to September 20 and measured a peak of 27.0% on September 29, settling at 8.0% in its published study. DemandSphere reported spurious results on about 11% of its daily checks on October 7. Semrush, SE Ranking, SISTRIX and AccuRanker have all published statements, with AccuRanker’s measured rates — under 1% by mid-October — far below Ahrefs’ range, a gap PPC Land notes points at how the data is collected rather than what is tracked.
The timing made the damage worse. The poisoning overlapped almost exactly with Google’s September 2026 spam update, which ran from September 24 to October 8 and produced genuine ranking drops — the update whose three-wave rollout GEroundtable tracked earlier this month. A drop recorded in that window could be a real demotion or a page Google served only to a suspected bot, and the two are hard to separate after the fact.
Stored History Is Being Rewritten
The newest part of Ahrefs’ statement is the cleanup. Earlier vendors described checks applied to new data; Ahrefs is the first vendor in PPC Land’s coverage to say it is revisiting data already stored. A ranking chart exported for a client report in early October may no longer match the same chart viewed today.
Ahrefs did not say how far back the cleanup reaches, how many stored results it has replaced, or whether corrected data points will be flagged. Nor did it say which products draw on the affected data: its SERP collection also feeds keyword difficulty, traffic estimates and AI Overview tracking, in a market where Semrush’s latest study put AI Overviews on 88% of tracked commercial keywords — measurements only as good as the results pages underneath them.
The practical check vendors recommend is Google’s own data. When a site appears to vanish from a rank report while Search Console still shows impressions and clicks for it, nothing happened to the site — the tracker was served a swapped page. During episodes like the ranking volatility and chatter spike of October 16, practitioners now have to ask a second question alongside “did rankings move”: whether the tool reporting the move was shown the real results at all.
Why It Matters
Rank tracking is the measurement layer the whole search industry reports through, and poisoning attacks the layer itself rather than any ranking in it. Agencies bill, publishers plan and boards judge performance on numbers from Ahrefs and its rivals; if 5% to 20% of the underlying results pages were swapped, a share of September and October’s reported wins and losses never happened. Ahrefs joining the vendors who have gone public — with the highest ongoing range stated so far, and a cleanup that can quietly change history customers already reported — turns a tracker complaint into an audit problem for anyone who acted on that data.
Background
Google and the companies that collect its results have been in a long escalation. Google dropped the num=100 results parameter in September 2025, multiplying the requests trackers need; its goto redirect rollout in August 2026 raised collection costs again; and its lawsuit against SerpApi survived in amended form after a court dismissed the original complaint in July 2026. Ahrefs, founded in 2010 and based in Singapore, is one of the largest independent collectors of search data.
0 Comments