Since late September, Google has been responding differently to a substantial part of the automated requests it receives. Instead of blocking a request or showing a CAPTCHA challenge, it returns a complete, normal-looking SERP with many of the usual organic results replaced by irrelevant ones. Nothing on the page directly signals that anything is off, which is why these pages have become known as poisoned SERPs.
This phenomenon isn’t specific to any single tool. All major SERP data providers and rank trackers report the same problem, and our Google SERP data has been affected as well. Our team is working on a reliable solution non-stop, and we’re confident the poisoned SERPs issue can be solved.
Below, we explain what poisoned SERPs are and why they appeared, what we see on our side, and what we’re doing about it. We also share advice on how to handle this issue in client communication and how to read your ranking data until the issue is resolved.
What poisoned SERPs are and why they appeared
Simply put, a poisoned SERP is a complete results page that retains its normal structure and set of extra features, but swaps out most of the organic results for unrelated ones. We consider a SERP to be poisoned if fewer than half of the websites that usually appear for its keyword appear anywhere on the page. That means if your website consistently ranks first for the keyword “best marketing automation software”, the poisoned SERP may not show your website in the top 10 results at all. Basically, the data in that SERP does not represent what real users see, making it unreliable.
Here are other key features that distinguish poisoned SERPs from normal ones:
- They appear only in response to automated queries. No verified mass occurrences of poisoned SERPs for normal users were reported.
- They can occur with any combination of input parameters. Automated requests with any combination of keywords, location, language, or other parameters can return a poisoned SERP.
- The organic results are most affected. Other SERP features, like AI Overviews, People Also Ask, and Local Packs, can also return incorrect results, but organic results are most likely to be swapped.
- They don’t appear consistently. Poisoned SERPs constitute only a specific share of scraped results. Not all automatically scraped SERPs are poisoned. However, the poisoned SERPs share can vary significantly.
However, the main threat of poisoned SERPs is that scrapers still treat them as valid pages and get them stored as real data. That’s why it is hard to instantly spot poisoned SERPs and sort them out from the actual results.
The poisoned SERPs issue is industry-wide and is not localized to a specific group of data providers or rank trackers. It affects SERP data collection globally, and as far as we know, no SERP data provider has fully solved it to date. On our side, the problem began on September 26.
Why did poisoned SERPs even appear? The exact reason is yet unknown. The incident occurred nearly at the same time Google began the September spam update, but poisoned SERPs persisted even after the update had fully rolled out on October 8. To date, Google has published no official statements regarding this.
We can stipulate that the reason is likely related to Google’s efforts to make automated search and scraping harder, particularly data scraping for LLMs. These efforts can be traced back to January 2025, when Google introduced SearchGuard – a JavaScript challenge intended to stop scrapers from crawling SERPs. Since then, Google has issued various updates that have made scraping harder, such as removing the num=100 parameter and deploying the “goto” redirect mechanism.
All these actions made SERP scraping more expensive, but we overcame them nonetheless. The poisoned SERPs, though, are an attempt to make it unreliable.
Where we stand
Like other data providers, we were affected by this issue and have been seeing poisoned results during scraping. According to our CTO, Oleksandr Nabatchikov, poisoned SERPs accounted for approximately 20%-30% of all results in the scraped data since September 26. They show up across different locations, languages, and keyword sets, and most of the damage falls on organic results, as described above. Here’s how we identify poisoned SERPs and what we are doing about this problem.
How we see the poisoned SERPs and why there is no instant fix
Recognizing a poisoned SERP is the hardest part of the issue because a poisoned page doesn’t appear broken at first glance. Its results are real, often relevant to the query, and laid out like a normal SERP. They just don’t match what a real user sees for the same search. Take the “iphone” query for the United States and English. A real user sees Apple’s website listing first, followed by Wikipedia and other familiar pages. A poisoned SERP for the same request may lead with YouTube videos and other pages that are still about iPhones, so nothing on the page looks wrong on its own. Subsequently, the only way to recognize a poisoned SERP is to compare it with what Google shows a real person.
You can run the same check yourself. Our Google Organic SERP API responses include a check_url field. It contains a link to the Google results page that matches your request parameters. Our documentation recommends opening it in incognito mode to confirm that the returned data is relevant. If you open it in a regular browser session, the same search almost never returns a poisoned page. However, this approach obviously is not scalable and cannot be used when processing millions of SERPs a day.
Since nothing on a poisoned SERP marks it as such, the only reliable signal is the keyword’s own history. Each SERP must be compared with the sites that typically rank for that keyword and re-requested if most of them are missing. Every repeated request lowers the chance of keeping a poisoned page.
The hard part is doing this reliably at our volume. On the one hand, if validation rules are too strict, the system can reject genuine ranking changes. On the other hand, if rules are too loose, the risk of fetching poisoned pages is higher. Besides, every extra check adds time and workload to processing the request.
This is exactly what makes the issue so complex. According to our CTO, even two retries with historical validation still leave 2-3% of results pages poisoned, with every extra step slowing down the processing time for millions of requests. That’s why there is no solution yet that filters out poisoned SERPs with 100% reliability.
What we’re doing about it
The poisoned SERPs issue is serious, but it is not unsolvable. We are confident it can be solved for a simple reason: real results are still within reach. A regular browser session retrieves them almost every time, which means the data hasn’t disappeared. The problem lies in how the data is collected.
Our team is working around the clock, testing new approaches every day to get clean results. None is reliable enough to roll out yet, but we’ll let our customers know as soon as we find an effective solution. We can’t give an exact date, but we’re doing our best to resolve the issue as quickly as possible.
There’s some positive news, too. Since October 10, Google has been returning noticeably fewer poisoned SERPs, and you can track the trend on our SERP Volatility Index.
What we recommend in the meantime
Until the issue is fully resolved, we suggest a few practices to help you keep your reports accurate and avoid reacting to ranking drops that never occurred.
Reporting and client communication
- Annotate the affected period. Add a note to your dashboards and reports starting September 20, and explain to clients that some ranking drops during this period are an artifact of data collection rather than real changes in position.
- Treat one-day drops as unconfirmed. Mark a sudden drop as unconfirmed until a second reading or Google Search Console confirms it. For volatile keywords, report the position that was held for three days.
- Adjust your alerts. Set ranking alerts to trigger only when two consecutive readings show the same change.
- Read aggregated metrics as trends. A handful of poisoned SERPs can pull down visibility or share of voice for a day, so evaluate the trend rather than a single data point.
- Export Search Console data for this period. Search Console retains data for the last 16 months, so save it now as a clean reference for future comparisons.
- Separate real volatility from noise. The September spam update rolled out during the same period, so not every change in late September and early October is a data issue.
Before you react to a drop
When a keyword suddenly loses positions, a few quick checks can tell you whether it’s worth acting on:
- Did the position come back the next day? If it did, the drop was most likely caused by a poisoned SERP.
- Did your competitors disappear as well? If most of the usual top 10 disappeared along with your page, the problem lies in the data. In a real drop, your position changes while competitors generally stay in place.
- What took their place? Does the page match what you see in a regular browser? When collecting data through our SERP API, you can open the
check_urlto see what a real user sees. If the usual sites are missing even though the results look plausible, it’s likely a poisoned SERP.
What to expect next?
Poisoned SERPs have made Google data harder to collect and easier to misread, but the clean results haven’t gone anywhere. Our team acknowledges that problem completely and is working around the clock on a reliable solution. We’ll notify you directly as soon as it rolls out. In the meantime, if a specific project looks off, you can always reach out to our 24/7 support team for assistance.