Google Is Serving Rank Trackers Fake Search Results

2026-10-17 — brand building small business India

A rank tracker called Nozzle just released data showing something odd.

Not blocked. Not rejected. Fabricated.

That's the distinction that matters here. A poisoned result looks perfect. Normal formatting. Same template. Real-looking websites. Except the actual rankings have been swapped out with completely different sites. The tracker stores it. The tracker reports it. By then the damage is done.

Nozzle analyzed 27.5 million search results pages across 49 countries between September 13 and October 9. The poisoned pages started appearing around September 20, peaked at 27% a few days later, then settled into a steady 5-11% per day.

Why Google Would Do This

The obvious target is AI companies. LLMs need training data and they're scraping Google's results to ground their answers. Google serving them garbage degrades their output. Eventually the AI companies have to either stop scraping or find another source.

Rank trackers like Nozzle get caught in the cross-fire. They collect SERPs exactly the way AI scrapers do—automated, bulk, relentless. From Google's perspective, there's no meaningful difference. Both are bots taking data at scale.

There's probably a secondary benefit too. If you're trying to game Google's rankings through black-hat SEO or reverse engineering, poisoned results would make that nearly impossible. Feed someone corrupted data and they can't learn the pattern.

What The Poisoned Results Actually Look Like

On a poisoned SERP, familiar websites just vanish. Nozzle found that normally Google retains 88.5% of the same domains when you search the same keyword twice. On a poisoned result, that number drops to 20.7%. It's almost random.

Instead, you see more Reddit, Facebook, Instagram, LinkedIn, Quora. Social platforms dominate. Meanwhile MapQuest and Yelp show up less. YouTube too. The pattern seems deliberate—flood the results with user-generated noise.

Actually, that's not quite right. The pattern is probably just "whatever dilutes the ranking signal most." UGC is harder to game. Harder to predict. So when Google wants to mess with a scraper's data, fill the page with it.

That makes sense. Generating an AI Overview costs more compute than rendering ten blue links. If you're already poisoning the page, why waste the inference cost on the AI feature too.

The Damage Is Real But Temporary

For the websites being tracked, the consequences are confusing. A site that normally ranks in the top ten drops out completely. Then three days later, it's back. To someone glancing at their rank tracker, this looks like a catastrophic ranking collapse and a miraculous recovery. In reality, nothing happened to their site. The ranking drop only existed in the poisoned data.

Eighty percent of the time, the poisoning lasted for just a single result fetch. Fifteen percent lasted for two consecutive fetches. Only 5% persisted across three or more requests for the same keyword. So Google isn't trying to permanently corrupt the data—just disrupt it enough that it's unreliable.

Most consultants get this wrong, including us sometimes. The natural reaction is to panic. New algorithm update? Penalty? But the real question is simpler: did my tracker get a poisoned page, or is this real?

Why This Matters For Anyone Tracking Organic Performance

If you rely on rank tracker data to make decisions, you need to cross-reference it against Google Search Console now. Search Console still shows the same clicks and impressions? Nothing happened. You got poisoned data.

This also raises a harder question for the SEO industry. If 8% of the data coming back is unreliable and you don't know which 8%, how much trust should you place in tracking at all? Some agencies are already seeing the unreliability. It's going to get worse.

Google has not commented on any of this. Nozzle published the research. DemandSphere published similar findings. Google stays silent. You can interpret that silence however you want.

Source: "Google has been answering a share of rank-tracker requests with complete, normal-looking results pages whose listings have been swapped out." — Relevant Audience