If your rank tracking has looked wrong since mid September, it is not your imagination and it is not an algorithm update.

Google is serving automated requests result pages that are complete, normal looking, but intentionally wrong. Every tool that depends on scraped results is reporting numbers it cannot stand behind, and Google has said nothing and it was entirely predictable.

A quick note on how these tools work, because the context matters. SEO platforms do not get ranking data from Google. They get it by sending millions of automated searches, pretending to be users, and copying what comes back. Google has never liked this. For the last year it has been making it harder, and in September it changed tactics.

The latest chapter in a long saga

Google removed the parameter that let tools pull 100 results per request in September 2025, making tracking roughly ten times more expensive overnight. It started routing result clicks through redirects in July, which pushed tools into 500 to 1,000 requests for a single query. It has been ramping “are you human” prompts all year. Then, from 13 September, requests it did not trust started failing in volume, and from 20 September the ones that got through started coming back different. This is Google fighting crawling, and it is escalating.

Enjoying Mumbrella? Sign up for our free daily newsletter.

Rank tracker Nozzle analysed 27.5m result pages across 49 countries and published on 10 October. Against a baseline where 0.17% of pages looked abnormal, 8% of pages collected after 13 September had most of the........

© Mumbrella