Poisoned Google results peaked at 27% of Nozzle's SERPs on Sept 29, its study finds
Nozzle says 8.0% of the 17.8 million Google SERPs it could judge since Sept 13 were poisoned, against 0.17% in an Aug 20 to Sept 9 baseline. A tracker can log a ranking loss on a page that never moved, so a drop needs a second source.
The Engineer · Build desk
Nozzle's study: a poisoned page keeps 20.7% of the previous page's organic sites; a normal one 88.5%. A project's own site was missing from the top 10 on 58.6% of poisoned SERPs, 2.7% in baseline. AI Overview sources kept 42.3% of previous sites, 68.8% in baseline.
| Figure | Value | Note | Claim |
|---|---|---|---|
| Previous page's organic sites a poisoned page keeps | 20.7% | a normal page keeps 88.5% | 5 |
| Own site missing from top 10, poisoned SERPs | 58.6% | baseline: 2.7% | 6 |
| AI Overview sources kept, poisoned pages | 42.3% | baseline: 68.8% | 9 |
What happened
- After the late-September peak, the daily poisoned rate has run at about 5 to 11% since October 1.
- A poisoned page keeps 20.7% of the previous page's organic sites, while a normal page keeps 88.5%.
- Where a project's own site usually appears, it was absent from the top 10 on 58.6% of poisoned SERPs; the baseline figure was 2.7%.
- Most spells are short: 77.9% of poisoned SERPs had their usual sites back within three days, and 80% of spells lasted a single fetch.
- Nozzle reads the pattern as Google pushing back on bulk scraping for AI answers.
Why it matters
- Who has to decide decision Teams that open tickets or rewrite pages from tracker alerts have to decide whether an alert waits for Search Console impressions and clicks before anyone acts.
- What it limits constraint The spam update overlaps the peak days, so tracker data alone cannot say whether a September 29 drop was poisoning, the update, or both.
- What it costs cost Agencies tracking many client sites repeat the service-account grant on every Search Console property they want checked.
- Who's exposed exposure Dashboards, docs-ranking scrapers and AI-visibility trackers read the same pages, and on poisoned ones the AI Overview appeared 34.9% of the time against 54.2%.
A poisoned page gives a tracker nothing to flag. Nozzle says these pages throw no error or captcha and are built on a real results page's template, but most of the keyword's usual sites are missing [2]. We'd expect a tracker to log that as a lost ranking.
At 8.0% of 17.8 million judged SERPs, the study implies about 1.42 million poisoned pages [16]. That rate is roughly 47 times the 0.17% baseline [17].
Both figures come from Nozzle's own sample, and Nozzle is an enterprise rank tracker [14]. They transfer to another tracker only if Google treats that tracker's fetches the way it treats Nozzle's. Nozzle's premise is that trackers get caught because they all collect SERPs the same way [19]. Google has not confirmed any of it [20].
Nozzle's data also stops on October 9 [2]. TrueRanker's write-up, updated October 15, says most of the obviously irrelevant results are fixed. It adds that Google is also serving pages whose results fit the query and still are not the real ones, and those are much harder to spot [21].
Search Console is the control because it counts clicks and impressions from actual searches, so a page served to a bot never shows up in it [10]. Nozzle's checklist closes on the same advice: when a site "disappears" in a tracker but keeps getting impressions and clicks in Search Console, the site itself was not hit [11].
The DEV Community author automated the comparison. The script reads a tracker CSV with date, query and position, with position left empty when the site was not found. It flags every date on which a keyword fell out of the top 10, fetches Search Console data for those dates and outputs a verdict [12].
Setup is a one-time cost. Create a Google Cloud project, enable the Search Console API, make a service account with a JSON key, then add the account's email under Settings > Users and permissions. Read access is enough. Skip the last step and every call fails with a 403 even though the key is valid [13].
The check has a limit. The September 2026 spam update started on September 24 and ran 13 days and 16 hours [8]. The September 27 to 29 peak falls inside that window [18]. A drop on September 29 could be the update, bad data, or both [8].
In Search Console, impressions that held up point to bad data in the tracker. Impressions that fell point to something real, with the update as the obvious suspect. Because most spells are short, a drop that reverses within a few days fits the poisoned pattern better than it fits a penalty.
In our view the check is cheap for a site you own. It covers only properties you can add a service account to. The post does not offer a comparable check for AI-citation data.
What to watch
- Whether Google confirms or rejects Nozzle's reading that it is pushing back on bulk scraping for AI answers.
- Whether anyone measures the query-fitting poisoned pages TrueRanker describes, since Nozzle's data stops on October 9.
- Whether Nozzle's daily poisoned rate stays inside the 5 to 11% band it has held since October 1.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence50
- Adoption
- Not scored
- Hype gap+10
- Incentives60
- Confidence50
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
According to a study Nozzle published on October 10, Google has been serving some rank trackers search result pages that look completely normal but have the results swapped out.
- [2]
In its Poisoned Google SERPs study (data through October 9, 2026), Nozzle describes pages with no error, no captcha, and the same template as a real results page, with most of the keyword's usual sites gone.
- [3]
8.0% of the 17.8 million SERPs Nozzle could judge since September 13 came back poisoned, against 0.17% in the August 20 to September 9 baseline.
- [4]
Poisoned SERPs peaked on September 27 to 29 at roughly a quarter of all SERPs (27.0% on September 29), and have run at about 5 to 11% a day since October 1.
- [5]
A poisoned page keeps 20.7% of the previous page's organic sites. A normal one keeps 88.5%.
- [6]
For keywords where a project's own site is one of the usual results, that site was missing from the top 10 on 58.6% of poisoned SERPs, against 2.7% in the baseline.
- [7]
77.9% of poisoned SERPs had their usual sites back within three days, and 80% of spells lasted a single fetch.
- [8]
The September 2026 spam update started on September 24 and ran for 13 days and 16 hours, so a drop on September 29 could be the update, bad data, or both.
- [9]
On poisoned pages, the AI Overview showed up 34.9% of the time instead of 54.2%, and its sources kept only 42.3% of the previous page's sites, against 68.8% in the baseline.
- [10]
Search Console's clicks and impressions come from real searches, not from a page served to a bot.
- [11]
Nozzle's own checklist ends with the advice that if a site 'disappears' in a tracker while Search Console still shows impressions and clicks, nothing happened to the site.
- [12]
The author's script reads a tracker CSV with date, query and position (position empty when the site was not found), finds every day a keyword left the top 10, pulls the same days from Search Console, and prints a verdict.
- [13]
Setup requires a Google Cloud project with the Search Console API enabled, a service account with a JSON key, and the service account's email added under Settings > Users and permissions on each property; the API needs only read permission, and skipping the add-user step makes every call fail with a 403 even though the key is valid.
- [14]
Nozzle is an enterprise rank tracker.
- [15]
An internal dashboard that pulls positions from a SERP API, a scraper that checks where docs or landing pages rank, and an 'AI visibility' tracker that records which sites Google's AI Overview cites all consume the same data.
- [16]
About 1.42 million SERPs were poisoned since September 13 (8.0% of 17.8 million).
- [17]
The 8.0% poisoned rate is roughly 47 times the 0.17% baseline.
- [18]
The September 27 to 29 poisoning peak falls inside the September spam update's run, which began September 24 and lasted 13 days and 16 hours.
- [19]
Nozzle's read is that this is Google pushing back on bulk scraping for AI answers, and that rank trackers get caught because they collect SERPs the same way.
- [20]
Google has not confirmed Nozzle's explanation.
- [21]
TrueRanker's write-up, updated October 15, says most of the obviously irrelevant results are fixed, but Google is also serving pages whose results fit the query and still are not the real ones, and those are much harder to spot.
Get this ledger inside ChatGPT. Ask about any of our stories and the answer brings its figure, each claim’s verdict and the sources. Free, no account.
Sources
1 independent publisher whose own reporting we read for this story.
- dev.toRank Trackers Are Getting Poisoned Google SERPs. Here's How I Check Drops With the Search Console API
1 article · October 15, 2026
Topics and entities
Follow any of these and your For You feed starts watching them — no settings page required.
Topics
- Rank trackingFollow
- SERP scrapingFollow
- Search Engine OptimizationFollow
- AI OverviewsFollow