Short answer: Since September 2026, rank-tracking company Nozzle says, Google has been returning complete, normal-looking search result pages to rank trackers with most of the usual websites swapped out. Nozzle calls them “poisoned” pages. On those pages, a tracked business's own site is missing from the top 10 far more often than it should be, so a rank report can show a drop that never reached real customers.
That is a strange thing to read when you own an HVAC company in Fort Wayne, a lawn crew in Auburn, a dental office or a restaurant. You did not change anything on your website. Your Google Business Profile looks the same. Then a monthly report from a vendor lands in your inbox with a red arrow on it. This post explains what Nozzle measured, what it does and does not prove, and how to check your own numbers before you spend money reacting to a chart.
A note on how we are treating the evidence. The main study comes from a company that sells rank tracking, and it studies pages collected for its own customers. Other rank-tracking vendors (Sistrix and Advanced Web Rankings) describe the same kind of problem, and two trade publications covered it, though both lean on Nozzle's data. We found no statement from Google. We say below where we are repeating a finding and where we are giving our own judgment.
Key Takeaways
- Nozzle analyzed 29.9 million result pages and found 22.6% were poisoned on page 1 or page 2 since September 13, 2026, against 0.28% before. The worst day was September 30, at 43.5%.
- On poisoned pages, a tracked site missed the top 10 on 60.8% of them, against 2.7% in the baseline. The page looks normal. Only the keyword's own history gives it away.
- Most of it passes quickly. Four in five spells were a single result page, 67.9% of poisoned pages had their usual sites back within three days, and 92.5% within two weeks.
- “Near me” searches were hit more often (13.6% against 8.6%), and home services ran above the U.S. average (15.0% against 8.8%).
- Nozzle says Search Console is your tiebreaker. If a rank tracker says you vanished but Search Console still shows impressions and clicks, nothing happened to the site.
What Did Nozzle Find, and When Did It Start?
Nozzle published a research report titled “Poisoned Google SERPs” on October 10, 2026, and updated it on October 21 with data through that date. A SERP is just a search engine results page. The report says that since September 2026 Google has been serving rank trackers “complete, normal-looking pages whose results have been swapped out,” and that “nothing on the page gives it away.” We read the live page on October 30.
The timeline in the report is what makes the word “poisoning” stick. On September 13, requests to Google started failing in large numbers. The requests that got through looked normal until September 20, when altered pages first appeared. Nozzle's point is that Google was not simply refusing the requests. It was answering them with a full page that was wrong. A refused request tells a scraper to try again. An altered one gets stored as if it were real.
The Numbers From Nozzle's Updated Report
- 29.9 million result pages analyzed since September 13, drawn from a sample of pages Nozzle tracks for customers across 53 countries.
- 22.6% poisoned on page 1 or page 2 since September 13, against 0.28% before. This counts desktop pages judged on both pages.
- 43.5% at the peak on September 30.
- 4% to 6% on October 14 and 15, then back to 25.9% on October 19 and 35.7% on October 21.
- 22.3% of the previous page's organic sites kept on a poisoned page, against 88.1% on normal ones.
How does Nozzle decide a page is poisoned? A page counts when fewer than half of the keyword's usual sites appear on it. Those usual sites are the ones that sat in the top 10 on at least 60% of that keyword's pages between August 20 and September 9. It is a comparison of each keyword against its own past, not against a universal list.
One caution on the numbers. Early coverage of the report quoted a lower figure, 8.0%, which counted only page 1. The October 21 update widened the test to page 1 or page 2 and found page 2 was poisoned about one and a half times as often as page 1. When you see a percentage attached to this story, check which version of the test it comes from.
Why Would Google Show Rank Trackers Different Results?
Nobody outside Google can say for certain, and we did not find a Google statement. What the sources describe is a pattern of Google making automated collection of its results harder, with rank trackers caught in the middle.
Search Engine Roundtable wrote that Google is blocking scrapers more successfully than it has in a long time. The write-up relayed Derek Perkins of Nozzle's description of a three-part approach: very wide IP blocking aimed at large residential proxy networks, a significant increase in captchas, and injecting bad data into suspect result pages so that tools collect the wrong data. The author said the third piece concerns them the most, because they suspect most tools will not notice it quickly. They also wrote that they do not think Google is targeting the tools SEOs use, but AI bots and SerpApi, a search-results data provider.
Search Engine Journal's analysis reaches a similar reading. It quotes Nozzle's view that the altered pages are Google's response to AI scraping and grounding, meaning the bulk collection of Google's results to feed AI answers, and that rank trackers collect pages the same way, so they are caught in it. SEO software company Sistrix says much the same on its support page: Google has been making it progressively harder to collect search results since 2025, and tools like theirs “aren't the target, but they are caught up in it.”
For you, the motive matters less than the mechanics. If it is true that tools are being fed altered pages on purpose or by side effect, then the reports that arrive in your inbox are built from partly contaminated data, and the people who build them have to filter it out. Some do. Sistrix says its quality checks caught the divergent results early, that it now publishes only results that pass additional checks, and that it re-collects keywords that fail. The cost is that some keywords are updated less often, and on-demand updates have been switched off since September 2026.
Does This Change What Your Customers See?
As far as we can tell from the sources, no. Nozzle's opening sentence is about “the SERPs it serves rank trackers.” The study measures pages collected by automated tools, not pages shown to people on a phone in Auburn or a desktop in Fort Wayne. That is an important limit, and it cuts in your favor. A rank report can show a collapse while the phone rings exactly as it did last month.
We should be careful about how far that reassurance goes. Nozzle's evidence is a comparison of tracker-collected pages against the same keywords' own history. It does not include a measurement of what human searchers saw during the same days. So the claim that real customers were unaffected rests on how Nozzle and the trade press describe the mechanism, plus the plain observation that businesses with steady calls and Search Console traffic did not collapse. We think that is a reasonable reading. We would not call it proven.
The study also looked at the parts of the page around the organic results, and the findings matter if you read AI or local data in a report. On a poisoned page the local pack kept 66.5% of its businesses from the previous page, against 85.8% on a normal one. AI Overview sources kept 37.5%, against 68.1%. Ads and Shopping results barely moved, and Nozzle's phrasing is that “only the parts advertisers pay for” hold steady. If a vendor tells you your map pack position or your AI Overview citations changed sharply in the last six weeks, that is worth a second look for the same reason.
One more detail is worth knowing. The sites that replace the usual results are not random. Nozzle found Facebook in the top 10 of 29.5% of poisoned pages against 21.9% of the same keywords' other pages, Instagram at 10.4% against 6.1%, Reddit at 23.0% against 20.7%, and Quora at 5.0% against 2.4%. Yelp moved the other way, 10.8% against 11.5%. So a poisoned report tends to show social media and forums elbowing past local business sites. If your report suddenly says Facebook beats you for your own service, treat that as a flag.
Which Northeast Indiana Businesses See the Most Noise?
Nozzle broke its U.S. pages out by industry, and the categories line up with the businesses we work with. All figures below are the share of U.S. result pages poisoned since September 13, per Nozzle's industry table.
Share of U.S. Result Pages Poisoned, by Industry
- Home services: 15.0%, with a worst day of 51.5% on September 29.
- Health and medical: 8.6%.
- Restaurants and food: 7.5%.
- Home and garden: 7.3%.
- Retail and shopping: 7.1%.
- Automotive: 5.0%.
- All U.S. pages: 8.8%, with a worst day of 31.2% on September 29.
Read that list with some humility. Nozzle does not tell us how many Northeast Indiana keywords are in its sample, and these are national industry buckets, not a Fort Wayne measurement. What the table does say is that home services, the group that includes many trades and lawn and landscape businesses, ran well above the average. Nozzle also reports that every industry peaked between September 27 and 29, so a bad stretch in late September in your report is the most suspicious one.
Then there is the “near me” finding. Nozzle reports that 13.6% of “near me” searches were poisoned, against 8.6% of everything else. Short, low-volume phrases were also hit more: 10.3% for phrases under 100 searches a month, against 6.6% to 7.6% above that. A local business that tracks a long list of small, specific phrases such as “furnace repair near me” is exactly the profile that sees noisy reports.
The one piece of good news for local owners is that local targeting did not make things worse. Nozzle says the same phrase tracked nationally and locally came out the same, 8.2% and 8.4%. Tracking from a Fort Wayne ZIP code was not a special risk in itself. Brand searches, the ones where someone types your business name, were hit least, at 5.3% for the most navigational phrases.
How Can You Tell a Fake Drop From a Real One?
Nozzle lists five tests in its report. We have rewritten them as plain-language questions you can ask about any alarming report, and we added where we would look first.
The Five Tests, in Plain English
- Does it match the keyword's history? Nozzle says to compare the results with the keyword's usual sites over the previous weeks, not with yesterday, which may have been poisoned too.
- Did it last? Four in five spells were a single result page. A real ranking change stays. If the drop vanished on the next update, it was probably the measurement.
- Is page 2 weird? Page 2 is poisoned one and a half times as often as page 1, mostly under a page 1 that looks fine. A report that shows page 1 steady and page 2 scrambled is a warning sign.
- Is the page complete? Nozzle says there is no error, no captcha and nothing in the HTML that marks the page. That means you cannot spot it by looking at one report.
- Does Search Console disagree? If the site “disappears” from the report while Search Console still shows impressions and clicks, Nozzle's answer is that nothing happened to the site.
The last test is the one an owner can run alone in ten minutes. Google's own help page for Search Console defines the numbers you would use. Impressions are “how often someone saw a link to your site on Google.” Clicks are “how often someone clicked a link from Google to your site.” Position is a relative ranking where 1 is the topmost spot. Google adds that the heuristics it describes “are subject to change,” which is a fair reminder that no single dashboard is a perfect referee.
Advanced Web Rankings, another rank-tracking vendor, makes a similar argument in a post dated September 18, 2026. It says Search Console's average position “blends every query, device, location, and AI Overview citation into one number,” so individual positions will always wobble around it. Its advice is to read positions as ranges and trends and to cross-check tracked data, Search Console trends and dated snapshots of the actual results. That vendor has its own product to sell, but the advice is sensible regardless of whose tool you use.
If a drop does turn out to be real, our earlier guide to SEO warning signs of a traffic drop covers what to check next, and our August spam update recovery guide covers one cause of genuine losses.
What Should You Measure Instead of a Rank Report?
The honest answer is to measure outcomes first and positions second. A rank report is a proxy for something you care about, which is whether the right people find you and contact you. When the proxy gets noisy, go back to the thing it was standing in for.
A 15-Minute Check Before You React to a Ranking Drop
- Start with the phone and the inbox. Count calls, form fills and booked jobs for the last four weeks against the four weeks before. This is the number that pays you, and no one can swap it out.
- Open Search Console. Compare clicks and impressions for the last 28 days against the previous 28. A real ranking loss shows up as fewer clicks. A fake one does not.
- Check your Business Profile numbers. Look at the calls, website clicks and direction requests Google reports for your listing. Our guide on how Google Maps ranks Fort Wayne businesses explains what moves those.
- Look at the update date. Ask when the tool last collected each suspect keyword. Sistrix says some keywords now show older verified data on purpose, which you can spot by the update date.
- Wait for a second reading. Given that most spells were a single result page, a drop that is gone at the next update was probably not a drop. Give it a few days before you change anything.
- Ask your vendor how they validate. Do they compare each result against the keyword's history before reporting a drop? Do they cross-check with Search Console? A good vendor can answer in two sentences.
This is also a good moment to revisit how you forecast. If your plan treats a daily ranking line as a straight path, a few weeks of altered pages will wreck it. Our guide to forecasting organic traffic in a range, not a promise fits this situation well, because ranges absorb noisy data in a way a single line cannot.
A word on what not to do. Do not rewrite your pages, buy a new service or pause a campaign because of a one-day red arrow. The reverse is also true: do not dismiss a drop that lasts. If your Search Console clicks fall and stay down for three or four weeks, that is a real change in how people find you, and it deserves a real investigation.
What Do We Still Not Know?
Plenty, and it is better to say so than to hand you a confident story that falls apart next month.
Open Questions
- What Google says. We found no statement from Google confirming or explaining any of this. The label “poisoning” is Nozzle's inference from its data, not Google's description.
- How representative the sample is. The pages come from keywords Nozzle tracks for its customers, which skews toward businesses that pay for rank tracking.
- How many Northeast Indiana keywords are in it. The industry and country figures are national. We have no Fort Wayne cut.
- Whether real searchers are ever affected. The sources describe tracker-facing pages. We found no measurement of what people saw in a browser over the same days.
- Where it goes from here. Nozzle reports the problem eased to 4% to 6% on October 14 and 15 and then climbed back to 35.7% on October 21, so we would not call it over.
- How your own vendor is handling it. Tools differ. Sistrix says top 10 rankings and its Visibility Index remain reliable, while it updates some keywords less often. Another vendor may handle it differently, so you have to ask.
The practical summary is short. Treat a sudden drop in a rank report as a question, not a verdict. Check the outcomes that cannot be faked first, then Search Console, then the tool. And if you would rather have someone else sort the signal from the noise, that is a reasonable thing to hire out.
Got a Ranking Report That Doesn't Match Your Phone?
Button Block helps Northeast Indiana businesses read their own numbers: Search Console, Business Profile activity, calls and form fills, side by side with whatever a rank tracker says. We will tell you plainly whether a drop looks real, and we will not sell you a fix for a problem that is only in a chart.
Frequently Asked Questions
- Did my business really drop on Google if my rank report says so?
- Maybe not. Nozzle found that a tracked site misses the top 10 on 60.8% of the altered result pages it studied, against 2.7% before. If your calls, form fills and Search Console clicks look normal, treat a one-day drop in a rank report as unconfirmed until it repeats.
- Are my real customers seeing the altered results?
- Nothing in the sources we read says so. Nozzle describes the altered pages as the ones Google serves to rank trackers, not to people searching in a browser. That is Nozzle's reading of its own data, and Google has not commented in anything we found, so we cannot call it settled.
- How long does a fake ranking drop last?
- Usually not long. Nozzle reports that four in five spells of altered results last a single result page, that 67.9% of altered pages had their usual sites back within three days, and that 92.5% did within two weeks. A drop that holds for weeks deserves a real investigation.
- Does Google Search Console have the same problem?
- Nozzle says that if a site disappears from a rank tracker while Search Console still shows impressions and clicks, nothing happened to the site. Search Console reports how often real people saw and clicked your links. Google's help page also warns that how those numbers are calculated can change.
- Should I fire my SEO agency or cancel my rank tracker?
- No, not over this. Several trackers have posted notices or added checks, and Sistrix says top 10 rankings stay reliable in its data. A better move is to ask your agency how they verify a drop before they report it, and to ask when each number was last updated.
- Is a local business more exposed than a national brand?
- Nozzle found that local targeting makes no difference when the same phrase is tracked nationally and locally. It did find that "near me" searches were affected more often, 13.6% against 8.6% for everything else, so local businesses that track those phrases may see more noise.
Sources & Further Reading
- Nozzle, “Poisoned Google SERPs” (Derek Perkins): nozzle.io (published Oct. 10, 2026, updated Oct. 21, read October 30): the 29.9 million result pages, the 22.6% and 43.5% figures, industry, country and “near me” breakdowns, recovery times and the five tests.
- Search Engine Roundtable, “Third-Party Google Rank Tracking Struggling Now”: seroundtable.com (read October 30, 2026): the three-part description of Google's approach and the list of rank-tracking tools that posted notices.
- Search Engine Journal, “Analysis: Google Is Poisoning Search Results For Rank Trackers”: searchenginejournal.com (read October 30, 2026): independent summary of the Nozzle report and its reading of Google's motive. Its figures come from an earlier version of the study, so we use Nozzle's updated numbers.
- Sistrix, “Why Google ranking data collection is changing”: sistrix.com (page modified Oct. 7, 2026, read October 30): how a rank-tracking vendor describes divergent results and its added quality checks.
- Advanced Web Rankings, “Why Your Tracked Rankings No Longer Match What You See in Google”: advancedwebranking.com (byline Sept. 18, 2026, read October 30): the vendor's view on position ranges and cross-checking. We read the opening sections only.
- Google Search Console Help, “What are impressions, position, and clicks?”: support.google.com (read October 30, 2026): Google's definitions of impressions, clicks and position, and its note that the heuristics may change.