"The bounce rate is too high" is one of the most repeated sentences in website meetings. It is often one of the least well founded. Bounce rate and time on page look like intuitive measures of satisfaction, but the way they are calculated makes them fragile. Understanding the mechanism helps you use them for what they are worth.
How bounce rate is calculated
In the classic definition, a session is a bounce when it contains a single page view and no other recorded interaction. The bounce rate is the percentage of such sessions.
The problem is obvious: someone looking for opening hours who finds them at the top of the page and leaves satisfied is a bounce. Someone who reads a ten-minute article from beginning to end and then leaves is a bounce. For a contact page or an article that answers a precise question, a high bounce rate can be a sign that the page works.
How time on page is calculated
Even more fragile is time on page in its traditional form. The tool only knows the moments when it receives a signal. Time on a page is calculated as the difference between the load of that page and the load of the next one.
Two consequences follow:
- for the last page of a session there is no subsequent signal, so the time is unknown and often recorded as zero;
- for a single-page session the time is always zero, unless other events are sent.
The average time on site of a blog, where many visits are single-page, can therefore come out at a few seconds even when readers read for minutes.
Engagement metrics
To get past these limits many tools have introduced engagement-based metrics. In Google Analytics 4, for example, a session counts as "engaged" if it lasts longer than ten seconds, contains at least two page views or includes a conversion event. Bounce rate is redefined as the share of sessions without engagement.
This approach is more reasonable, but it is still a convention. Ten seconds on a page can mean careful reading or a forgotten open tab. Engagement time is measured only while the page is in the foreground, which reduces the noise without eliminating it.
More reliable signals
To find out whether content works, it pays to supplement or replace these metrics with signals tied to the page's purpose:
| Page type | Useful signal |
|---|---|
| Long article | Share of readers reaching the final section |
| Service page | Clicks towards the contact form or pricing page |
| Documentation | Copied code blocks, follow-up searches |
| Contact page | Clicks on phone, email or map |
| Product page | Add-to-basket actions, opening images |
These events answer the real question: did the person do what the page was meant to let them do?
When bounce rate is still useful
Bounce rate should not be thrown away but put in context. It is most useful for comparisons between similar items:
- the same page before and after a change;
- the same landing page for different campaigns;
- pages of the same type, for example all service pages.
A much higher bounce rate for one campaign than for the others, on the same page, suggests that campaign brings people who expected something else. That is concrete information.
A sudden rise in bounce rate across the whole site, on the other hand, should first be treated as a suspected technical problem: a script that no longer sends events, slow loading that makes people leave before any interaction, a spike of automated traffic.
A practical example
In my work I came across a documentation site with a bounce rate of around eighty per cent, which everyone considered a terrible sign. After adding an event to the "copy" button on code blocks, it turned out that a substantial share of those single-page visits copied exactly the command they were looking for. The site was not failing: it was answering the question as quickly as possible.
In short
- A bounce is a session with one page and no interaction, not necessarily a failed visit.
- Traditional time on page cannot measure the last page of a session.
- Engagement metrics are more sensible but remain conventions with arbitrary thresholds.
- Use events tied to the page's purpose to tell whether content works.
- Bounce rate is mainly useful for comparing similar pages or campaigns.
