A new analysis from enterprise rank-tracking firm Nozzle claims that Google is showing misleading search results pages (SERPs) to rank trackers and AI companies in an apparent effort to poison the data they are attempting to collect.
Why Google May Poisoning The SERPs
Nozzle explains that the non-standard search engine results pages (SERPs) began showing up on September 20th and reached as high as 25% of the search results and since October 1st have comprised between 5 to 11% of the results per day.
That sequence of dates is why they characterize these kinds of search results as poisoned:
“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.”
Their opinion then is that SERP poisoning is happening in order to thwart AI companies that rely on Google’s Search Results from LLMs that use it to ground their answers and to confound rank trackers.
Although Nozzle did not mention it, in my opinion it may also be a way to confound spammers who are trying to distill Google’s search results answers and reverse engineer how pages are ranked.
Poisoned SERPs Are Widespread
The analysis appears to show that the misleading search results are widespread but varies across countries, industries, and the types of searches.
But what the poisoned SERPs are said to have in common are:
- They introduce artificial SERP volatility
- Show more user generated content (UGC)
- Show fewer AI Overviews
- Advertising and shopping results appear to be stable
Artificial SERP Volatility
Normal search results retain 88.5% of the same websites for repeated searches with the same keyword phrases. Poisoned results retain only 20.7% of the websites from successive searches for the same keywords.
That means that Google may be diluting the rank tracking reports so that it shows a greater variety of websites ranking temporarily for the same keyword phrases. To an outsider looking at the statistics that will appear as extreme ranking volatility.
They also shared these statistics:
- 80%: The poisoning lasted for one search result for a set of tracked keyword phrases.
- 15%: The poisoning lasted for two successive search results for the same tracked keywords.
- 5%: The poisoning lasted for three or more successive search results for the same keyword phrases.
Fewer Websites/More UGC
Google appears to be poisoning the SERPs by showing more user generated content from social media and forum platforms. Facebook, Instagram, LinkedIn, Quora and Reddit appear more frequently in poisoned search results.
At the same time, MapQuest, Yelp, and YouTube appear less frequently in poisoned SERPs. There is no explanation as to why that may be.
MapQuest and Yelp are both local search entities but SERP poisoning was found to happen in local searches as much as non-local searches when everything else was equalized.
The analysis explains:
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.
Fewer AI Overviews
Poisoned SERPs that are shown to rank trackers are less likely to show AI Overviews (AIO). This makes sense because the LLM inference necessary to generate AIO costs more than it does to display a regular ten blue links search result. Presumably, this reason also holds for other search features like People Also Search.
Here’s what Nozzle’s data shows:
- AI Overviews appears 34.9% of the time in poisoned SERPs versus 54.2% in the baseline.
- The “people also search for” feature appears 52.7% of the time in poisoned SERPs versus 87.2% in baseline.
- Image packs appear nearly twice as often in poisoned SERPs: 44.2% of the time versus 23.0% in the baseline.
- Forum blocks shows more often in poisoned SERPs: 27.0% versus 20.5% in the baseline.
We can only speculate as to why Related Searches feature appear less frequently. Could it be shown less often because Google is minimizing actionable data or do Related Searches appear less often because that feature costs more to generate?
Why do image packs show up more often? Is it because SEOs and rank trackers view text data as actionable and image data is perceived as less useful?
Different Search Features
Two other notable changes that the Nozzle analysis documented are that Google displays fewer pages of search results on some desktop searches and that advertising and shopping results remain fairly stable.
The analysis also indicates that the disruption is widespread but varies significantly across countries, industries, search types and request routes.
Takeaway
Nozzle’s analysis suggests Google may be swapping out search results with misleading data in order to disrupt AI-based scraping and rank trackers. The artificial ranking volatility introduces challenges for rank tracking services, individuals who are tracking the search results themselves and possibly for spammers who are trying to reverse engineer certain of the search results, including Google’s AI Overviews.
Read more about the Nozzle Analysis here.
Hat tip to Cyrus Shepard’s LinkedIn post for bringing this to attention!
Featured Image by Shutterstock/Toey Andante