Poisoned Google SERPs

What it looks like

Each pair below is the same keyword, device and target location, collected one to three days apart. Pick one, then drag the divider across the page.

By the numbers

A SERP is poisoned when fewer than half of its keyword’s usual sites appear anywhere on it, and they come back on a later SERP. A keyword’s usual sites are the ones in its top 10 on at least 60% of its SERPs from August 20 to September 9.

The SERPs come from a representative cross-section of the datasets Nozzle collects for its customers: 49 countries, desktop, iPhone and Android, and targeting from whole countries down to postal codes. Four in five of those datasets had at least one poisoned SERP.

On September 13, requests to Google started failing in large numbers, and for two weeks fewer SERPs came through. The ones that did were normal until September 20, when poisoned SERPs first appear. They reached a quarter of all SERPs from September 27 to 29, and have run at about 5 to 11% a day since October 1.

That sequence is why we call it poisoning. Google is not refusing these requests; it answers them with a complete page whose results are wrong, and nothing on the page says so. A refused request tells a scraper to try again, while a poisoned one gets stored as real. 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.

Since September 13, 8.0% of SERPs were poisoned, against 0.17% in the baseline. Of the markets in the chart, Australia (11.5%), India (10.9%) and Canada (10.5%) ran highest, the United States at 8.5%, and Germany, France and Spain at 3.4 to 4.3%, with the United Kingdom lowest at 3.0%. The device barely matters: 7.7% on desktop, 8.3% on iPhone and 7.1% on Android. A SERP counts as poisoned once its keyword’s usual sites come back, so the last days of the chart are still filling in.

What takes their place

Social platforms and forums. Each keyword’s poisoned SERPs against its own other SERPs since September 13: how often a site is in the top 10.

Reddit gains the most, 8.0 points, while YouTube, Yelp and MapQuest lose ground. Three or more social sites share the top 10 of 7.2% of poisoned SERPs, against 2.7% of the others, and on US keywords 14.2% of poisoned SERPs have a foreign country-code site, such as a .ca or .co.uk, in the top 10, against 6.1%.

Where it hit

The share of SERPs poisoned since September 13 in each country with at least 10 datasets and 10,000 judged SERPs. Hover a country for its number, or open the table.

Western Europe ran lowest: the United Kingdom, Denmark, Spain, France, Italy, the Netherlands, Germany and Portugal were all under 5%. Their neighbors were not: Austria ran at 18.8%, Belgium at 13.2% and Switzerland at 12.8%, with Norway at 13.1%. South Africa (18.4%), the UAE (14.4%), New Zealand (13.8%) and Saudi Arabia (12.9%) complete the group above 12%.

It also depends on the route a request takes to Google. On the same days, from September 30 to October 9, one of the routes Nozzle collects through returned 1 to 7% poisoned SERPs and another 9 to 30%.

What kind of searches

• “Near me” searches more than most. 11.4% of them were poisoned, against 7.8% of everything else.
• The long tail more than the head. 9.4% for phrases searched fewer than 100 times a month, 6.1 to 7.1% above that.
• Brand searches least. 5.3% for the most navigational phrases, 6.6 to 9.0% for the rest.
• Local targeting makes no difference. City and postal-code SERPs look worse, at 8.4% and 10.7% against 7.8% for whole countries, only because they lean toward mobile and “near me”. The same phrase tracked nationally and locally, in the same country on the same device, comes out the same: 7.5% and 7.4% across 21,508 such pairs.

By industry

Each ridge is one industry’s US SERPs, day by day from August 20, all on one scale and ordered by the share poisoned since September 13; the grey ridge is every US SERP. Every industry peaked between September 27 and 29; what differs is how high. Legal SERPs reached 51.3% on September 27, pet SERPs 13.6% on September 28.

Beyond the organic results

A poisoned SERP changes more than its organic results. For each part of the page, the chart compares a SERP with the keyword’s previous one: of the sites, questions or searches that part showed, how many it shows again.

The organic results keep 20.7% of the previous SERP’s sites on a poisoned SERP, against 88.5% in the baseline, and “People also search for” keeps 30.7% of its searches, against 87.9%. Only the parts advertisers pay for hold steady: shopping keeps 92.2%, against 92.8%, and ads keep more than usual, 83.7% against 70.6%. What is shown changes too: the AI Overview appears on 34.9% of poisoned SERPs against 54.2% in the baseline, and “People also search for” on 52.7% against 87.2%, while image packs nearly double, 44.2% against 23.0%, and forum blocks rise from 20.5% to 27.0%. On desktop keywords that usually run to page 8 or beyond, 50.9% of poisoned SERPs stop at page 4 or earlier and 10.9% show no page numbers at all, against 0.2% and under 0.1% in the baseline.

What it does to tracked sites

For keywords where a project’s own site is one of the usual sites, that site was out of the top 10 on 58.6% of poisoned SERPs, against 2.7% in the baseline and 10.2% on the same keywords’ other SERPs since September 13. If a tracked site fell out of the top 10 for a day in late September and came straight back, this is the likely reason.

How long it lasts

Most spells of poisoning are a single SERP: 80% last one, 15% two and 5% three or more. The next SERP has the usual sites back 55.9% of the time, and 77.9% are back within three days, two days at the median.

How to tell it from a real ranking change

• The page is complete. No error, no captcha, the same template, the same build of Google’s frontend as a normal page. Nothing in the HTML marks it.
• The result set does not match the keyword’s history. Comparing against yesterday alone is fragile, because yesterday may have been poisoned too. Comparing against the keyword’s usual page-1 sites over the previous weeks separates a data problem from a ranking drop.
• It does not last. Four in five spells of poisoning are a single SERP, and most are back within three days. A real ranking change stays.
• Poisoned pages are thinner on average. An AI Overview and “People also search for” less often, more image packs and social sites, shallower pagination. None of that is reliable on its own.
• Search Console disagrees. If a site “disappears” and Search Console still shows impressions and clicks, nothing happened to the site.