How do Gmail's spam filters work (generally)?

Gmail.com uses layered reputation engagement based content based and machine learning models to evaluate each message. Its filtering leadership began years ago when it pioneered large scale Bayesian filtering learning continuously from user actions such as the Report Spam and Not Spam buttons.

They rely heavily on domain reputation subscriber level behavior and global sender patterns. Gmail learns from billions of interactions. Users rescuing your messages from spam boosts trust. Users deleting without opening reduces it.

It remains one of the most adaptive filtering systems ever built adjusting to new threats and sender behavior rapidly.