Universal Analytics counted a bounce as one thing: a session with exactly one hit sent to Google, no second pageview and no event after it. GA4 replaced that with an entirely different test, and the two numbers are not comparable even when they are measuring the identical visitor.
The exact formula, GA4
In GA4, bounce rate is the inverse of engagement rate: Bounce rate = 1 minus (Engaged sessions / Total sessions). A session counts as engaged if at least one of three things is true: it lasted 10 seconds or longer, it included a conversion event, or it had 2 or more pageviews or screenviews. A session that misses all three counts as a bounce in GA4. Hitting even one of them stops that, regardless of what the visitor actually did on the page. Both formulas operate on sessions, not on visitors, so one person triggering two separate sessions in a day can bounce on one and not the other.
The old formula, Universal Analytics
Universal Analytics defined a bounce far more strictly: a session that sent only a single hit to Google Analytics and then ended, with no second interaction of any kind recorded, no matter how long the visitor stayed on that one page reading it. A visitor who spent four minutes reading an article and then left without clicking anything else counted as a bounce in Universal Analytics. The identical visitor counts as engaged in GA4, purely because they crossed the 10 second threshold. This is the single biggest reason bounce rate dropped, often sharply, for sites that migrated from Universal Analytics to GA4 with no actual change in visitor behavior.
| Universal Analytics (legacy) | GA4 | |
|---|---|---|
| What counts as a bounce | Exactly one hit sent to GA, session ends there | Session under 10 seconds, no conversion, only 1 pageview |
| Does long dwell time on one page prevent a bounce | No, time on page was not counted at all | Yes, 10 seconds or more is enough on its own |
| Related built-in metric | None | Engagement rate, the direct inverse |
| Adjustable | No | The 10 second threshold is fixed, but the underlying session timeout it depends on is adjustable |
Bounce rate vs engagement rate
In GA4 these are the same measurement stated two ways, not two independent metrics measuring different things. Engagement rate is engaged sessions divided by total sessions, and bounce rate is just 1 minus that number. A site with a 65% engagement rate has a 35% bounce rate by definition. Reporting both in the same dashboard is redundant, not complementary. If a report shows engagement rate improving and bounce rate not moving to match, something is wrong with the report, not the traffic.
What is a good bounce rate
There is no fixed target here. Any number thrown around without naming the page type is a guess dressed up as a benchmark. A single-topic blog post that fully answers the search query in one scroll is supposed to have a high bounce rate under the old definition, because the visitor got what they came for and left, which was never a failure. A checkout flow or a multi-step signup should have a low bounce rate, because bouncing there usually means someone abandoned before finishing. Judge bounce rate against the job the page is doing, and against its own trend over time, not against a number pulled from a different site built for a different purpose.
Take two pages on the same site. A pricing page with a 40% bounce rate and a blog post with a 70% bounce rate are not evidence the blog post is failing and the pricing page is succeeding. The blog post likely answered its query in one scroll and the reader left satisfied, while a pricing page bouncing at 40% means most visitors left after ten seconds without checking a plan, viewing FAQs, or clicking through to signup, which is a genuinely worse sign given what that page exists to do. The number only means something once it is read against the page's job.
Where bounce rate quietly breaks
Single page applications and heavy scroll-based layouts can under-report bounces in the old sense and over-report engagement in the new one, because a visitor who never triggers a second pageview event can still be counted as engaged in GA4 simply by staying past 10 seconds, and a site that fires extra events on scroll or video play can push a genuinely uninterested visitor into the engaged bucket the moment a script fires, whether they were paying attention or not. This is worth knowing before trusting engagement rate as a stand-in for actual interest.
Bounce rate on a link in bio page
A link in bio page is not really the kind of page bounce rate was built to describe. It has one job, get someone to tap the right link, and most visits are meant to be short by design. Raydar records every tap on that page, referrer and UTM data included, which answers a more direct question than bounce rate ever could: did people find the right link and take it. Raydar cannot follow that visitor once they leave for the destination site, and no link tool honestly can. Measuring what happens after that click, including anything resembling a bounce rate on the destination page itself, needs website visitor tracking installed on that destination, a genuinely different category of tool built to watch a whole site rather than a single page of links. See what a link in bio is for where that boundary sits.
If a report is comparing a bounce rate from one tool against a bounce rate from another and the numbers will not line up, the mismatch usually traces back to the same root cause covered in why analytics numbers never match: different session definitions, different engagement thresholds, or a tool that was never measuring the same event in the first place.