Bounce rate is the share of visits where someone viewed one page and left without another interaction. Google does not use it as a ranking factor, and it is one of the most misused numbers in SEO reporting.
Why it isn’t a ranking factor
Google does not have your bounce rate. It sits in your analytics, which Google Search does not read for ranking, and Google representatives have said so repeatedly over more than a decade.
There’s a measurement problem underneath the ranking one. A bounce is not failure. Someone searching for your opening hours, finding them, and leaving satisfied is recorded identically to someone landing on the wrong page and hitting back in disgust. The metric cannot distinguish the best possible outcome from the worst.
What GA4 changed
Universal Analytics defined a bounce as a single-page session with no interaction event. GA4 inverted the framing: it measures engaged sessions, defined as lasting over 10 seconds, firing a conversion, or including 2 or more pageviews. Bounce rate in GA4 is simply the inverse of engagement rate.
That’s a better metric, because a 30-second read now counts as engagement rather than a bounce. It still isn’t a ranking signal, and the change means historical comparisons across the migration are meaningless.
What is true about user behaviour
The instinct behind the bounce rate obsession is not silly, which is why it persists. Google does use click behaviour, just not yours.
Navboost, confirmed under oath during the US antitrust trial, re-ranks results using aggregated click data over roughly a 13-month window. That is behaviour on the search results page, measured across many users, and it is a different thing from what your analytics records after the click.
| Your bounce rate | Navboost click signals | |
|---|---|---|
| Measured by | Your analytics | |
| Measured where | On your site, after the click | On the results page, across users |
| Google can see it? | No | It is Google’s own data |
| Ranking factor? | No | Confirmed under oath |
| You can change it by | Editing the page | Being the result people choose and stay with |
What to use it for
Diagnostics, within a page type, over time. A sudden jump in bounce rate on one template after a deployment is a genuinely useful signal that something broke: a form failing, a layout collapsing on mobile, a script timing out.
Comparing bounce rate between a blog post and a checkout page tells you nothing. Comparing your bounce rate to an industry benchmark tells you less, because the benchmarks circulating online are aggregated across incompatible measurement setups.
Common mistakes
- Reporting it as an SEO KPI. It measures your analytics configuration nearly as much as your content.
- Chasing it down with tricks. Auto-playing video, forced interactions and scroll events fired on load all improve the number and change nothing real.
- Comparing across page types or industries. Close to meaningless.
- Confusing it with dwell time. Related confusions, neither visible to Google from your analytics.
- Assuming a high bounce rate is bad. On a page whose job is to answer one question, it may be the mark of success.
Related
Dwell time is the same confusion in different clothes. Google’s ranking systems covers what Navboost actually is.