Where Did Bounce Rate Go in GA4? Engagement Rate Explained

For fifteen years, bounce rate was the metric everyone knew: the share of visitors who looked at one page and left. Then GA4 arrived without it, marketers protested, and Google added it back with a different definition wearing the old name. Understanding what actually changed saves you from comparing two numbers that do not measure the same thing.

What Universal Analytics measured

In Universal Analytics, a bounce was a session with exactly one interaction hit, usually a single pageview. Time did not matter. Someone who found your article, read every word for six minutes, and left satisfied was a bounce. Someone who clicked through to a second page after two seconds was not.

That definition made bounce rate easy to compute and easy to misread. Content sites with excellent one-page visits looked terrible; sites that split content across pages looked engaged. Everyone knew the caveats and quoted the metric anyway.

What GA4 measures instead

GA4 starts from the opposite end with engaged sessions. A session counts as engaged if any one of three things happens: it lasts 10 seconds or longer, it includes a key event (GA4's current name for conversions), or it includes at least two pageviews or screen views.

Engagement rate is engaged sessions divided by total sessions. Bounce rate in GA4 is simply everything else: the share of sessions that were not engaged. The two always sum to 100 percent; they are one measurement with two labels.

So the six-minute single-page read that UA called a bounce is an engaged session in GA4, because it clears the 10-second bar. The only sessions GA4 calls bounces are the genuinely empty ones: under 10 seconds, one page, no key event. Which is closer to what people always assumed bounce rate meant.

The 10-second threshold is configurable. In the property's tag settings the engagement timer can be raised as high as 60 seconds, which is worth considering for content sites where a 10-second glance should not count as engagement. Just remember that changing it mid-year puts a step in your trend.

Why the numbers are not comparable

Because the definitions share nothing but a name, GA4 bounce rates run dramatically lower than UA bounce rates for the same site. A blog with a 75 percent bounce rate in UA can show 30 percent in GA4 without a single visitor behaving differently. Any report, benchmark, or executive dashboard that lines up UA history against GA4 numbers under one "bounce rate" column is comparing apples to a different fruit entirely.

If stakeholders track the metric, the honest move is a one-time reset: explain the definition change, restate targets in terms of engagement rate, and stop referencing UA-era baselines. The old numbers are not wrong; they are unconvertible. There is no reliable formula mapping one to the other because the conversion depends on how long your one-page visitors stayed, which UA never recorded.

Finding it in the interface

GA4 does not show bounce rate by default, which is a statement of opinion as much as a design choice. To see it, customize a standard report and add bounce rate to the metrics, or pick it in an exploration. Engagement rate, engaged sessions, and average engagement time per session are more prominent, and they are usually the better vocabulary: "engagement rate" invites the right question, while an inherited "bounce rate" figure drags fifteen years of the wrong definition behind it.

One practical caveat: engagement measurement depends on events arriving after the first pageview, so aggressive consent blocking or a tag that fires late can make sessions look shorter and less engaged than they were. If engagement rate moved sharply when your consent setup changed, suspect the measurement before the audience.

How Analyse reports it

Analyse keeps the classic definition and states it plainly: a bounce is a visit that viewed a single page, and bounce rate is shown per page and per source in Analyse's website analytics, alongside visit duration so you can tell a satisfied one-page read from a two-second exit yourself. There is no renamed metric to translate and no hidden default to customize back in. If you moved from Universal Analytics, the number means what it meant there, which makes historical instincts and old benchmarks usable again. For how the two products differ more broadly, see the GA4 comparison.