Redirect

A redirect is when the server, instead of answering directly, sends the browser to another address; each redirect adds an extra round trip to the loading chain.

A redirect is sometimes necessary — moving HTTP to HTTPS, unifying the www and non-www address, or pointing an old URL somewhere new — but its cost is hidden: each hop takes a full network round trip before the real page load even begins, and a chain of several redirects in a row can add seconds. The common trap is the redirect chain: one address goes to a second, the second to a third, and the user waits for all of them. There is a data nuance too: a zero for redirect means “there was no redirect,” not “the redirect was fast,” so any redirect metric should be read only over visits that actually took at least one hop. Cutting unnecessary redirects is one of the cheapest speed improvements.