What Does Nozzle Claim?
Nozzle, a company that sells SERP tracking, says that since mid-September Google has answered some automated requests with complete, normal-looking results pages whose results have been swapped out. In its words, “Google is not refusing these requests; it answers them with a complete page whose results are wrong, and nothing on the page says so.” It adds: “It is Google’s response to AI scraping and grounding, the bulk collection of its results to feed AI answers, and rank trackers, which collect SERPs the same way, are caught in it” (Nozzle).
Three points about the source matter. Nozzle has a commercial interest in the topic, and its page names no individual authors. The data comes from its customers’ tracking, which is not a random sample of searches. And the explanation, that Google is doing this on purpose to deter AI scraping, is Nozzle’s inference. Search Engine Journal’s Roger Montti, who reported the study on October 14, added his own view that it may also confound spammers; his article contains no Google response (Search Engine Journal). PPC Land reported on October 13 that Google “has not commented publicly” on Nozzle’s or DemandSphere’s figures (PPC Land).
What Does Google Say About Automated Queries?
Google has not addressed this study, but its rules on the activity are public. Its spam policies define machine-generated traffic as “sending automated queries to Google,” including “scraping results for rank-checking purposes or other types of automated access to Google Search conducted without express permission,” and say such activity “consumes resources and interferes with our ability to best serve users” and violates its spam policies and Terms of Service (Google Search Central, last updated August 28, 2026). In other words, the automated collection that rank trackers rely on is something Google says it does not permit. That is context for the dispute; it does not confirm that Google is altering results.
The closest Google statement in the coverage concerns a related August change to result links, not this study. DemandSphere quoted Google as saying on August 26: “We have a long history of deploying technical measures against evolving forms of abuse, and we regularly take steps to protect our services and users,” according to PPC Land. We did not see the statement in a Google publication.
How Did Nozzle Measure It, and What Are the Denominators?
The denominators change from figure to figure, so they are worth setting out. Nozzle’s sample holds 63.1 million results pages collected from August 20 to October 9, 2026, of which 43.9 million were “judged.” A page is judged only if its keyword had five or more baseline pages and a set of “usual sites,” defined as those in its top 10 on at least 60% of baseline pages from August 20 to September 9. Since September 13, Nozzle collected 27.5 million pages, of which 17.8 million were judged, and “the 8.0% poisoned is a share of those” (Nozzle). Pages with no organic results are left out.
A page counts as poisoned when fewer than half of the usual sites appear on any of its pages, and at least 80% of them return on a later page within 14 days, or most missing sites return within three days. Because a page is classed only once the sites come back, the most recent days are undercounted, which Nozzle says itself.
| Measure (Nozzle) | Poisoned pages | Comparison | Denominator |
|---|---|---|---|
| Share of pages poisoned | 8.0% since Sept. 13 | 0.17% baseline | Judged pages (17.8 million since Sept. 13) |
| Organic sites kept from previous page | 20.7% | 88.5% baseline | Previous page’s sites, same keyword |
| AI Overview shown | 34.9% | 54.2% baseline | Pages in each group |
| Image pack shown | 44.2% | 23.0% baseline | Pages in each group |
| Ads kept from previous page | 83.7% | 70.6% baseline | Where both pages show ads |
| Customer’s own site out of top 10 | 58.6% | 2.7% baseline; 10.2% other pages since Sept. 13 | Keywords where the site is a usual site |
| Reddit in top 10 | 28.8% | 20.8% on the same keywords’ other pages since Sept. 13 | Weighted by each keyword’s poisoned pages |
Note the last row: Reddit’s comparison is with the same keywords’ other pages in the same period, not with the August baseline. Nozzle reports poisoning first appeared on September 20, peaked at about a quarter of pages from September 27 to 29, and has run at “about 5 to 11% a day since October 1.”
Where Is the Effect Largest?
% of judged result pages in each country
Vendor data, not independently reproduced. Nozzle's table covers 45 of its 49 countries (those with at least 10 datasets and 10,000 judged pages); the overall figure is 8.0%. The United States supplies more than 60% of all judged pages. Nozzle says results also depend on the network route a request takes.
The chart shows selected countries from Nozzle’s 45-country table, from Austria at 18.8% to the United Kingdom at 3.0%; the United States, with most of the sample, is at 8.5%. Nozzle reports that the device makes little difference (7.7% desktop, 8.3% iPhone, 7.1% Android) and that “near me” searches are hit harder, at 11.4% against 7.8%. It also tested whether local targeting matters and found it does not once other factors are matched: “7.5% and 7.4% across 21,508 such pairs.” Among U.S. industries, legal ran highest at 15.9% and pets lowest at 4.3%.
What Can the Study Not Show?
Nozzle lists its own caveats. Days with under a fifth of usual volume can swing the line sharply. The last days are still filling in. Results depend on the request route: from September 30 to October 9, “one of the routes Nozzle collects through returned 1 to 7% poisoned SERPs and another 9 to 30%.” Signals such as a thinner page are not “reliable on its own,” and comparing against yesterday alone is “fragile, because yesterday may have been poisoned too.”
Two further limits are ours. The study measures pages served to automated collection; it says nothing about what people searching on Google see. And the period overlaps Google’s September 2026 spam update, which ran from September 24 to October 8, so real ranking changes and collection problems happened at the same time; see How Long Does a Google Spam Update Take?
What This Means for Anyone Buying a Placement
Placement reports often lean on rank trackers: “the article ranks sixth for our brand name,” “visibility rose 20%.” If Nozzle is right, a one-time automated reading from late September or October could come from an altered page, and a sudden drop that reverses within days may be the tool’s data rather than the market. Our recommendations, not platform rules: ask any vendor how its rank data is collected and whether it re-checks before reporting a change; confirm important positions in a browser while signed out; and compare with Search Console, which is Google’s own record of real searches. The first-party approach is in How Should a Company Measure Whether Placement Worked? and How to Track Branded Search in Google Search Console.
Methodology and Limitations
Four limits apply. First, every figure in the chart and table is Nozzle’s; we read its page and did not reproduce its collection. Second, Nozzle and DemandSphere both sell data collected from Google, which is a conflict of interest a reader should weigh, though Nozzle publishes its definitions, denominators and downloadable aggregates. Third, Google has not commented on either study, so the cause remains unconfirmed. Fourth, DemandSphere’s 11% figure comes to us through PPC Land’s account, and PPC Land notes the method was not described.
Conclusion
On one vendor’s numbers, about one in twelve automated Google checks it could judge since mid-September returned a page that looked normal and was not. Google says the automated checking itself breaks its rules and has said nothing about the study. For a company reading a rank report, the safe habit is to treat any single automated reading as unconfirmed until it is repeated or matched in Search Console, whether it flatters a placement or not.
Frequently Asked Questions
Is Google giving rank trackers altered results?
Nozzle, a rank-tracking vendor, says so, reporting that 8.0% of the 17.8 million result pages it judged since September 13, 2026 were "poisoned." Google has not commented on the study. Q: What does "poisoned" mean in Nozzle's study? A: A page where fewer than half of its keyword's usual sites appear on any of its pages, and the sites come back on a later page. The usual sites come from an August 20 to September 9 baseline. Q: How large was the sample? A: 63.1 million result pages from August 20 to October 9, 2026, in 49 countries. The 8.0% figure is a share of the 17.8 million pages judged since September 13. Q: What does Google say about rank tracking? A: Google's spam policies say automated queries, including "scraping results for rank-checking purposes," without express permission violate its spam policies and Terms of Service. Q: Does this affect what people see when they search? A: The study measures pages served to automated collection. It does not say that people searching on Google saw altered pages. Q: How should I check a rank report? A: Our recommendation: repeat the check, confirm key positions in a signed-out browser, and compare with Search Console before accepting a change.
Sources
This guide is built from Nozzle's published research page, Google's spam policies, and trade reports by Search Engine Journal and PPC Land, all read on October 15, 2026. Recommendations on reading rank reports are ours. We did not run any tracker or reproduce Nozzle's analysis.
- vendor research Nozzle Research, Poisoned Google SERPs. Supports the method, denominators, definitions, figures, country table and caveats. Published October 10, 2026. Read October 15, 2026.
- platform documentation Google Search Central, Spam Policies for Google Web Search. Supports the machine-generated traffic policy. Page shows "Last updated 2026-08-28 UTC." Read October 15, 2026.
- trade press Search Engine Journal, Analysis: Google Is Poisoning Search Results For Rank Trackers. By Roger Montti, October 14, 2026. Supports the trade report and the absence of a Google response. Read October 15, 2026.
- trade press PPC Land, Google returns wrong results on 11% of DemandSphere's daily rank checks. Dated October 13, 2026. Supports DemandSphere's 11% figure, the quoted August 26 Google statement and that Google had not commented. Read October 15, 2026.