Predicting Google rankings with machine learning

Years of tracked rankings are a training set waiting to happen. A new machine-learning model predicts a page's likely search position from around twenty signals — domain and page authority, referring domains, trust and citation flow, page-speed and Core Web Vitals, and content length.

It turns the platform's historical data into something forward-looking: not just where a page ranks today, but where it's likely to land. For an agency, that's the difference between reporting the past and advising on the future.