3 min read

“SERP poisoning”: how Google scrambles its results for SEO tools, and how Semji is adapting
For several weeks now, Google has been deliberately returning false results to the tools that collect its search results pages. Every player in the market is affected. Here is what is happening, what it may have changed in your data, and what we have put in place.
What is SERP poisoning?
To analyze your competitors or track your rankings, SEO tools query Google automatically, thousands of times a day. This is called scraping.
Until now, Google sometimes blocked these requests. It is now more subtle: it responds normally, but with a “poisoned” results page. It looks like a real SERP in every way, but its content differs from what a real user sees. Some results are off-topic, others are simply out of order.
One example cited by Semrush: a Brazilian search for a stroller model returned the page of a dental equipment supplier, because both use the Portuguese word “carrinho”.
Google has also changed the way it displays the links in its results. They are now encrypted, and only become readable after a delay. The goal: to prevent its results from being reused in real time.
Why is Google doing this?
Google has made no official statement. The analysis shared across the industry is this: Google is trying to protect itself from massive scraping, especially by AI players who feed their models and answer engines with its results. SEO tools are collateral damage.
The whole market is affected
Every tool that relies on Google data faces the same phenomenon:
- Ahrefs describes an “industry-wide” problem, with 5% to 20% of results affected depending on the data provider.
- Semrush has published a dedicated note on “poisoned SERPs” and adjusted its data sources.
- SERP data providers themselves have documented successive waves of degraded results since mid-October.
The phenomenon is also unstable: Google regularly changes when and how it applies these altered results. What works one day may stop working the next.
What this may have changed in Semji
Two features may have been affected at times:
- SERP analysis in the editor: the competitors analyzed for a keyword, and therefore the Content Score, may have reflected less relevant pages than usual.
- Rank tracking: some ranking changes may have been unexpected, especially on keywords unrelated to your pages.
Data from your Google Search Console (clicks, impressions, average position) is not affected: it comes directly from Google, not from scraping.
What we have put in place
Our teams have been following the issue daily since the first signals. In practice:
- An adapted scraper: we have upgraded our own collection tool to handle these new constraints, including the encrypted links.
- Several data providers, tested every day: we compare the main providers on the market daily against specific criteria: relevance of results, presence of AI Overviews and ads, behavior across several languages and markets.
- Fast switching: we can enable or disable a provider automatically, within minutes, depending on the quality of its data.
The result: a workaround is in place and the data is back to normal. As Google will keep evolving its protections, we remain on active watch and ready to roll out new adjustments whenever needed.
What should you do if you spot an anomaly?
- Compare with your Google Search Console data, which remains the reference.
- Contact your CSM or our support team, specifying the keyword and the page concerned. This feedback helps us detect new variations faster.
The reliability of your data is a priority. We will keep you informed if the situation changes.
See it on your own brand
Find out where AI engines mention you
Semji tracks which prompts surface your brand across ChatGPT, Google AI Overviews, Perplexity, Gemini, Claude and DeepSeek — then shows you the content to write to close the gaps.
ChatGPTGoogle AI OverviewsPerplexityGeminiClaudeDeepSeek
By
Olivier Balais
CTO


