Straight answer: bounce rate is the percentage of visits where someone leaves your site without any meaningful engagement, but what counts as “meaningful” changed completely when Google moved from Universal Analytics to GA4. In old Analytics, a bounce meant one page view and nothing else. In GA4, a session only counts as a bounce if it lasts under 10 seconds, has no conversion event, and no second page view. That difference matters more than most explanations of this metric let on.
What bounce rate really measures
Bounce rate answers one narrow question: did the person do anything after landing on your page? Not “did they read it,” not “did they like it,” just did they trigger a second action within the same session. That’s it. It says nothing about whether they got what they came for, whether they’ll come back, or whether they bought something later after leaving.
People treat it like a report card on content quality. It isn’t. It’s a behaviour flag, and behaviour has dozens of innocent explanations. Someone reads your entire 2,000-word article, gets the answer, and closes the tab. That’s a bounce. It’s also a success. Nobody tells you that when they’re explaining the metric, because “lower is always better” is a much easier sentence to write.
Old Google Analytics vs GA4: the definition changed
If you learned Analytics before 2023, unlearn some of it now. In Universal Analytics, bounce rate was simple: single page view sessions divided by total sessions. One click, no second page, no event fired, it’s a bounce.
GA4 flipped the whole thing round. It doesn’t report bounce rate as its headline metric anymore, it reports “engagement rate,” and bounce rate is just the inverse of that. A session counts as engaged (so, not a bounce) if any of these happen:
- The session lasts 10 seconds or longer
- The user views 2 or more pages or screens
- The user triggers at least one conversion event
So a visitor can land, scroll for 8 seconds, and leave, and that’s a bounce. Another visitor can land, sit on the page reading for 40 seconds without clicking a single thing, and that’s not a bounce, because time on page alone clears the 10-second bar. This is why bounce rates dropped for almost everyone the moment they switched to GA4. It’s not that behaviour improved overnight. The ruler changed length.
A real example: the “85% bounce rate” panic that wasn’t a problem
I had a client a couple of years back, a home improvement business in the north of England, who rang me in a bit of a state because her blog’s bounce rate was sitting at 85%. She’d read somewhere that anything over 70% meant her content was rubbish and people were fleeing in disgust.
We pulled the report together. Average session duration on those same pages was just under 4 minutes. People weren’t leaving because the content was bad, they were reading a full how-to guide on choosing a boiler, getting their answer, and closing the tab satisfied. Zero clicks, zero second page view, one long stay. Classic single-session read.
What mattered wasn’t the bounce rate at all, it was that her newsletter signups from that same traffic were converting at a decent rate weeks later, through email, not through more clicks on the same visit. If she’d chased the bounce rate down by stuffing internal links into the article to force a second page view, she’d have made the article worse and probably lost some of that trust that led to the signup in the first place. This is one reason I keep telling people bounce rate is a signal to investigate, never a verdict to act on blindly. If you want the fuller picture on how Google Analytics fits into your wider marketing setup, this is exactly the kind of nuance that gets lost when people only look at one number in isolation.
What counts as a “good” bounce rate
There’s no single healthy number, because different page types behave completely differently. Rough benchmarks I’ve seen hold up across dozens of client accounts and industry studies:
- Blog posts and long-form content: 65% to 90% is normal, and often fine
- Ecommerce category or listing pages: 20% to 40% because people naturally click into products
- Ecommerce product pages: 30% to 50%
- Landing pages built for a single paid ad campaign: 60% to 90%, because there’s often nowhere else on the page to go
- Contact or “thank you” pages: high bounce rate here is expected and completely irrelevant
- Homepages: 30% to 55% is a reasonable range for most small business sites
If your blog is sitting at 55% and someone tells you that’s a red flag, they’re using the wrong benchmark. If your paid landing page is at 40%, that’s worth investigating, because it’s unusually low for that page type and might mean your tracking is misfiring rather than your page performing brilliantly.
How to see bounce rate in GA4 (step by step)
Here’s the annoying bit: GA4 doesn’t show you bounce rate by default anywhere in the standard reports. Google buried it in favour of engagement rate. You have to go and add it back yourself.
- Open your GA4 property and go to Reports, then Engagement, then Pages and screens
- Click the pencil icon in the top right corner to customise the report
- Under Metrics, click Add metric, then search for and select Bounce rate
- Save the customised report so it stays there next time
- For a deeper cut, use Explore, then Free form, drag in Session medium and Bounce rate so you can see it by traffic source, not just by page
That last step matters more than people realise. A page’s overall bounce rate might look fine at 55%, but if you split it by source you might find organic search visitors bounce at 40% while a specific paid campaign is dragging in traffic that bounces at 90%. Same page, wildly different story depending on who’s landing on it and why.
When bounce rate is telling you something is wrong
I don’t want to swing this too far the other way and say bounce rate never matters, because sometimes it’s exactly the right alarm bell. Look closer when you see:
- A page load time over 3 seconds paired with a bounce rate way above the norm for that page type, because slow pages push people away within seconds
- A big jump in bounce rate right after a redesign or a new page template went live, which usually points to a broken layout or a missing call to action
- A landing page bounce rate above 90% where the ad copy promises one thing and the page delivers something else, which is a mismatch problem, not a content problem
- Mobile bounce rate sitting 20 or more points above desktop on the same page, which almost always means a mobile display or speed issue
The pattern in every one of those is that bounce rate on its own tells you nothing, but bounce rate compared against something else (load time, device, source, page type) tells you plenty. It’s a diagnostic tool, not a diagnosis.
Why I stopped treating bounce rate as a top-line KPI
Here’s the uncomfortable bit that most explanations of this metric skip past: bounce rate doesn’t measure whether your marketing is working, and it never did, even in the old Universal Analytics days. It measures a very specific, very narrow bit of on-page behaviour that can mean success or failure depending entirely on context you won’t find in the metric itself.
I’d rather look at what happened after someone left. Did they come back later through email or a branded search? Did they convert on a different visit a week on, something you can only see if you’re tracking events and conversions rather than just page views. A blog post with an 88% bounce rate that quietly drives newsletter signups and repeat visitors is doing its job. A checkout page with a 15% bounce rate that never converts anyone is failing, and bounce rate alone would never have told you that. If you’re serious about knowing whether traffic is turning into anything, pair bounce rate with the actual conversion rate formula for that page, because one without the other is half a picture.
I also think the obsession with bounce rate quietly discourages good writing. I’ve watched businesses chop up useful long articles into three separate pages purely to force a second page view and drag the number down, and the content got worse for it, not better. If you’re weighing whether long-form content is even worth the effort in the first place, that’s a fair question, and I’ve written separately about whether blogging still makes money for small business owners, because the honest answer depends far more on what happens after someone reads than on whether they clicked a second link.
The one thing worth checking before you touch anything else
Before you panic about a bounce rate number, check that your tracking is even set up consistently across your site. I’ve lost count of the number of accounts I’ve audited where the Analytics code wasn’t on every page, which quietly inflates bounce rate because visitors moving between pages get counted as leaving entirely when the second page simply wasn’t tracked. Fix that first. Every other number downstream depends on it.
Frequently asked questions
Is a high bounce rate always bad?
No. For blog posts, guides, and reference pages, a bounce rate of 70% to 90% is often completely normal and can even signal that someone got their answer quickly. Judge it against the page type and what happens after the visit, not against a single universal target.
What is a good bounce rate in GA4?
There’s no single good number, but as a rough guide, homepages around 30% to 55%, product pages 30% to 50%, and blog or content pages 65% to 90% are all within normal range for most small business sites. Look at engagement rate alongside it since GA4 defines bounce as the inverse of engagement.
Why did my bounce rate drop when I moved to GA4?
GA4 counts a session as engaged, not a bounce, if it lasts 10 seconds or more, includes 2 or more page views, or triggers a conversion event. Universal Analytics only avoided a bounce if the visitor clicked to a second page, so many sites saw bounce rates fall simply because the definition became easier to avoid, not because behaviour improved.
How do I lower my bounce rate without ruining my content?
Focus on page speed, clear internal links to related content, and matching what your ads or search snippets promise to what the page delivers. Avoid artificially splitting good articles into multiple pages just to force a second page view, since that usually hurts the reading experience without improving actual results.
