Definition
Bounce rate is the percentage of website sessions in which a visitor did not engage meaningfully before leaving. In Google Analytics 4 it is defined as the opposite of engagement rate: the share of sessions that were not 'engaged'.
The definition changed with GA4. In Universal Analytics, which stopped processing data in July 2023, a bounce was any single-page session with no further hit, so a visitor who read an article for five minutes and left still counted as a bounce. According to Google's GA4 documentation, a session now counts as engaged if it lasts longer than 10 seconds, includes a key event, or has two or more page or screen views. Bounce rate is the percentage of sessions that meet none of those criteria. Bounce rate and engagement rate are not shown in most GA4 reports by default; you add them to detail reports such as Pages and screens.
A high bounce rate is a symptom, not a verdict. It often points to a mismatch between what a visitor expected and what the page delivered, a slow or unstable page on mobile (see Core Web Vitals), intrusive interstitials, or a landing page with no obvious next step. But some pages are supposed to bounce: a store-hours page or a definition that answers the question in eight seconds has done its job. Compare bounce rate across similar page types and traffic sources, not against a universal benchmark.
Bounce rate is not a Google ranking factor. Google has said it does not use Google Analytics data in ranking, and many sites do not run GA at all. Search engines and AI systems can only infer satisfaction from their own signals, which is a separate topic covered under dwell time. Treat bounce rate as a diagnostic for your own content and conversion work.
It has become more useful for reading AI search traffic. Visitors who click through from a ChatGPT, Perplexity or Copilot citation, or from an AI Overview, usually arrive after the AI has already answered the basic question, so they tend to want depth, proof or a next action. Segmenting GA4 sessions by AI referrer (chatgpt.com, perplexity.ai, copilot.microsoft.com and others) and comparing their engagement rate, key events and conversion rate with organic search shows whether the pages AI engines cite actually serve those visitors. A cited page with an unusually high bounce rate from AI referrals usually means the citation promised something the page does not deliver, or the answer is buried below the fold. Track this alongside AI-referred traffic volume, since a small, highly engaged AI segment can matter more than its raw session count suggests.
To lower bounce rate where it matters, match the page to the search intent of the queries and prompts that send traffic, put the answer or offer near the top, fix load and layout issues, and give each page a clear next step such as a related guide, a comparison or a trial.
Examples of Bounce Rate
- A SaaS pricing page shows a 71% bounce rate from organic search but only 38% from ChatGPT referrals, because AI visitors arrive already comparing plans and click straight into the plan details.
- A how-to guide moves its step-by-step instructions above a 400-word introduction and sees GA4 engagement rate rise from 44% to 63% within a month.
- A store-locator page with a 90% bounce rate is left alone after the team confirms most sessions end with a click on the phone number, which they then configure as a key event.
- A B2B site finds Perplexity visitors bounce heavily from an old statistics page that Perplexity keeps citing, and updates the page with current figures and a link to the full report.
