Rank trackers may be recording results that Google never showed a real searcher. Derek Perkins of Nozzle, a rank-tracking vendor, says Google has served suspected trackers complete, believable search pages whose listings were quietly replaced, and Search Engine Roundtable’s Barry Schwartz reported the claim on October 12, 2026.
This changes the nature of the problem we covered in September. At that point, Nozzle reported that most of its Google requests were being blocked, and the question was how much data went missing. The new claim is harder to catch. Data that arrives looks complete, so nothing signals that it is wrong.
According to Schwartz, Perkins describes a three-part approach from Google. The first is wide IP blocking aimed at major residential proxy networks. The second is a sharp rise in captchas. The third is the injection of bad data into SERPs Google suspects of being scraped, so tools sometimes collect the wrong results.
Perkins laid out the evidence in a Nozzle write-up titled “Poisoned Google SERPs.” He wrote that since September 2026, Google “has been poisoning the SERPs it serves rank trackers.” Nothing on a page gives it away, he said. A keyword’s own history does.
Nozzle’s headline measurement is that 8.0% of SERPs came back poisoned. That is the vendor’s own figure, drawn from its own collection, and no independent party has replicated it. Its other findings, as Schwartz summarized them, are that requests were blocked first and poisoned afterward, that social sites and forums filled the gaps in affected results, that tracked sites fell out of the top 10, and that the effect passes within days.
Schwartz is more worried about the third tactic than the first two. His concern is that many tools will not detect the altered data quickly and will assume everything is fine. He also gives his read on motive: he does not believe Google is going after the tools SEOs use, and expects the real targets are AI bots and SerpApi, the scraping provider Google is feuding with. That is his opinion.
Vendors have responded unevenly. Schwartz says Sistrix, DataForSEO, Advanced Web Rankings, and Morningscore have posted notices. SE Ranking said it was “missing positions for some keywords,” though most of the issue had been resolved. Several others did not reply, and some confused the older goto redirects with the new behavior. Schwartz calls the redirects the easy part for tools to handle.
Practitioners on X echoed the distrust. Lily Ray wrote on October 8 that rank trackers were “too unreliable at the moment” to share findings. Antony Deroshan wrote that Google detects bot traffic and poisons results with junk webpages instead of blocking it. Daniel Skov Jacobsen suggested some daily swings are measurement noise because tools now sample instead of counting, and advised checking the Search Console trend first. Alex Wolk said spot-checking queries in an incognito window is currently the only way to trust a movement.
Those last two suggestions are the practical core. Treat a tracker’s movement as unconfirmed until a second source agrees. Compare any sudden drop against Search Console impressions and average position. Spot-check a sample of the affected keywords by hand. Watch for the pattern Perkins describes: a tracked site vanishing from the top 10 while forums and social domains appear in its place, then reverting within days.
Search Console has a limit worth remembering. Schwartz notes it offers no click or query data for generative AI search such as AI Overviews and AI Mode, so tracker doubts hit AI visibility reporting hardest.
Add a verification step to every ranking report: no client-facing claim of a loss or gain should rest on tracker data alone. Flag keyword-level swings that reverse within days as possible poisoned collection before anyone opens a ticket or rewrites a page.
Search Engine Roundtable, Barry Schwartz, October 12, 2026, citing Nozzle’s Derek Perkins and his “Poisoned Google SERPs” write-up.
