What Is Bounce Rate?
Bounce rate in GA4 is the percentage of non-engaged sessions, not single-page visits like the old UA definition. What changed, and a real example.
Bounce rate in GA4 is the percentage of sessions that were not engaged, the mathematical inverse of engagement rate. A session counts as engaged if it lasts 10+ seconds, includes 2+ page views, or triggers a key event; anything else counts as a bounce.
Why bounce rate matters
Here’s the correction most glossaries and blog posts still get wrong: bounce rate does not mean “single-page visit” anymore. That was the old Universal Analytics definition, retired in 2023. GA4 redefined it as the inverse of engagement rate, based on time, interaction, and page count, not just whether someone viewed a second page. If your team is still benchmarking current GA4 numbers against old UA-era bounce rate stats, stop, they’re not measuring the same thing and the comparison is meaningless. A high bounce rate isn’t automatically bad either way: someone who reads a full article for four minutes and leaves satisfied now counts as an engaged session in GA4, not a bounce, which is actually a more useful signal than UA ever gave you.
How bounce rate is calculated in GA4
- A session starts when a user lands on the site and GA4 fires
session_start. - GA4 checks whether the session becomes “engaged” against three criteria, met if any one applies: lasts 10+ seconds (this threshold is configurable up to 60 seconds), includes 2 or more page/screen views, or triggers at least one configured key event (like a form submission or purchase).
- Engagement rate is calculated as engaged sessions divided by total sessions.
- Bounce rate is the inverse: 100% minus engagement rate. The two always sum to exactly 100%.
How to actually improve it
- Match the page to search intent. Visitors who land on the wrong type of page for what they wanted leave fast, no content fix solves a mismatched intent.
- Speed up the page. A slow Largest Contentful Paint pushes visitors to leave before the 10-second engagement threshold is even reached.
- Add a clear next step. An internal link, a related-content section, or a simple CTA gives an otherwise-satisfied reader a reason to view a second page.
- Configure key events properly. If your important actions (newsletter signup, scroll depth, video watch) aren’t set up as GA4 key events, engaged sessions that should count aren’t being counted.
- Don’t chase an arbitrary benchmark. Compare a page’s bounce rate against its own historical trend and its intent type, not a generic industry number pulled from an outdated UA benchmark list.
UA bounce rate vs. GA4 bounce rate
| Factor | Universal Analytics (retired) | GA4 (current) |
|---|---|---|
| Definition | Single-page session, any duration | Non-engaged session (under 10 sec, 1 page view, no key event) |
| A 4-minute single-page read | Counted as a bounce | Counted as engaged, not a bounce |
| Relationship to engagement rate | No fixed relationship | Always exactly 100% minus engagement rate |
| Historical comparison | — | Not directly comparable to UA-era numbers |
Real example
For example, if 100 people land on a blog post and 70 of them read for over 10 seconds before leaving without viewing a second page, GA4 counts those 70 sessions as engaged, not bounced, since they cleared the time threshold. Under the old UA definition, all 70 would have counted as bounces simply for staying on one page.
Bounce rate and AI search
Zero-click searches and AI Overviews are shrinking the pool of visitors who ever reach your site at all, but for the ones who do click through, engagement-based bounce rate is arguably more useful now than ever: it tells you whether the click was worth the AI Overview sending it. A visitor who arrived skeptical after an AI summary, then read for two minutes, registers as a real engagement signal, not noise.
FAQ
Is a high bounce rate always bad?
No. On a single-page reference article or a page designed to answer one question fully, a moderate bounce rate is expected and not a problem, especially since GA4’s definition already excludes visitors who genuinely engaged for 10+ seconds.
Can I change the 10-second engagement threshold?
Yes. It’s adjustable up to 60 seconds per web data stream in GA4’s Admin settings, useful if your typical content requires longer to genuinely engage with.
Why did my bounce rate change dramatically after migrating to GA4?
Because it’s measuring something fundamentally different from Universal Analytics. A direct before-and-after comparison across the migration is not valid; treat GA4 bounce rate as a new baseline, not a continuation of your old numbers.
What’s a good bounce rate in GA4?
There’s no single reliable industry benchmark yet for GA4’s engagement-based definition, old UA-era benchmarks don’t apply. Compare a page against its own trend and similar pages on your own site instead.
Related terms
If your bounce rate benchmarks are from before 2023, throw them out. GA4 changed what the number means, not just how it’s displayed.