Traffic
Bounce rate
Also called bounce percent.
Bounce rate is the percentage of sessions that met the bounce rule in a given period. Two sites with the same figure can have very different visits behind it.
How it is measured
Divide bounced sessions by total sessions for the same filter and date range. The denominator is sessions, not pageviews and not people. A segment filter changes both parts, so keep the filter on when you compare two numbers.
Tarsier counts the rate from cookieless visits, so someone who comes back on another day adds a fresh visit to the denominator. Tools that keep a long-lived cookie merge those returns and will print a different figure for the same traffic.
Worked example
A SaaS docs site has a 71 percent bounce rate on /docs/install and 28 percent on /docs/api. The install page gives one command that people copy and run in a terminal. The API page links to a dozen siblings.
The team tries to push the install figure down by adding a sidebar of links, and the rate drops to 55 percent. Support tickets about install do not change, which tells you the old number was the answer, not the problem.
How it differs
Bounce rate is a percentage of sessions. A bounce is the single event it counts. Exit rate divides exits by pageviews of one URL and can be high on pages that sit at the end of a good journey. Neither tells you anything unless you know what the page is meant to do.
Common errors
Comparing a landing page to a blog post. Chasing a site-wide target. Reading a rate computed from very few sessions. Looking at it without segmenting by device and channel. Comparing different bounce definitions between tools.
In practice
List your five top landing pages, write down what a good visit looks like on each, and judge the rate against that. Keep the sample size next to the percentage, and ignore any page with fewer than a few dozen sessions.