Bounce rate and engagement rate are very easy to misinterpret and are very specific to the analytics platform you’re using. If you were using Universal Analytics (the version prior to GA4) when it was around you might want to check out this post we wrote about the difference in engagement and bounce rate between GA4 and Universal Analytics. Otherwise, read on!

Measuring engagement and the user_engagement event

GA4 measures the actual active time that someone spends on your website/app. So if you have 100 tabs open and are navigating between them, it doesn’t start the timer until the user is actually on your website and starts to interact with it.

GA4 keeps a counter of all the active time a user has spent on a tab. It tries to send it back to you alongside standard events that happen on the page like scroll, click etc. However, your page might only have the basic page_view event fire and that’s at the very start of the user being on that page so the timer hasn’t started yet. To overcome this, GA4 tries to send a special event called user_engagement just before the user closes the tab or navigates to another page. When we say tries, sometimes this doesn’t actually fire and it depends on the browser, but the majority of the time it works fine.

This means the “engagement time” metric you see in GA4 is fairly accurate. If it’s lower than what you’d expect, it’s probably because it’s only measuring active time.

Engagement rate and bounce rate

A session is considered engaged if (more info here) one of the following happens:

• The user spends at least 10 seconds of active time on your website OR
• The user views at least 2 pages on your website OR
• The user completes a key event (formerly known as conversion, see the next post in the sequence for more)

If a session doesn’t match any of this that’s counted as a bounce, meaning the user left within 10 seconds by closing the tab or just changing tabs. Because sessions are either engaged or a bounce, your engagement rate and bounce rate will add up to 100%. For example, if 30% of your sessions are bounces, you have a bounce rate of 30% meaning you have a 70% engagement rate.

Below is a graphical summary of 4 users, only 1 of whom has a session that’s a bounce:

While these metrics are more accurate and useful than in Universal Analytics, we don’t recommend getting too hung up on them especially on industry benchmarks. After all, if you define more key events you will reduce your bounce rate (and if you define the page_view event as a key event, your bounce rate will be 0%). It’s probably more useful to monitor this over time on your own website, meaning that your own historical data is the best benchmark.

A high bounce rate is not necessarily bad, sometimes you want the user to get the information quickly and leave, for example single-message sites like this one or pages on your website that might answer just one question (eg. contact details).