Why your website has a high bounce rate, and when it matters
A high bounce rate is often not a problem at all. How to tell the harmless kind from the expensive kind, and what to measure instead.
Entrovix AIAug 29, 2026 11 min read
Somebody searches for your opening hours, lands on your contact page, reads the hours, and leaves. That is a bounce. It is also a complete success, and if you go looking for ways to stop it happening you will make the page worse.
This is the central problem with bounce rate as a metric. It measures whether a session contained more than one interaction. It does not measure whether anybody got what they came for, and those two things come apart constantly.
The short answer
A high bounce rate is only worth investigating on pages where a second action was the point. On a service page, a pricing page or a paid landing page, people leaving immediately is a real signal. On a page that answers one question completely, it is the sound of the page working.
What the number actually measures now
Worth being precise, because the definition changed and a lot of advice still describes the old one. In Google Analytics 4 the headline metric is engagement rate, and bounce rate is defined as its exact inverse: the percentage of sessions that were not engaged. The two always add to 100.
A session counts as engaged if it lasts longer than ten seconds, or includes a conversion event, or includes at least two page views. A session that meets none of those is a bounce. So a visitor who reads carefully for thirty seconds and leaves is engaged, while one who opens the page and closes it in four seconds is not.
That is a considerably more useful definition than the old one, which counted any single-page session as a bounce regardless of how long it lasted. If you are comparing today's figure against a number from a few years ago, you are comparing two different measurements.
Where a high bounce rate is fine
| Page type | Is a high bounce rate a problem? | Why |
|---|---|---|
| Contact, hours, directions | No | One fact was needed and supplied |
| A blog post that fully answers its question | No | The reader got the answer and left satisfied |
| A page whose conversion is a phone call | No, and it is invisible | The call does not appear in analytics unless the tap is tracked |
| Service or product page | Yes | A second step was the whole point |
| Paid landing page | Yes, and expensively | You paid for every one of those arrivals |
| Homepage | Usually yes | It exists to route people somewhere |
Read that table before reading any benchmark. Comparing your site's overall figure against an industry average tells you very little, because the average is a blend of page types that have nothing to do with each other and your blend is different.
The version of this that genuinely matters
There is one pattern worth taking seriously: a visitor arrives from a search result, spends a few seconds, and goes straight back to the results page to try something else. That is a reader telling you the page did not answer the query, and it is a different thing from an ordinary bounce.
Analytics does not label this for you, but you can infer it. Look for pages with high impressions in Search Console, a click-through rate that is not terrible, and almost no engaged sessions once people arrive. That combination means the result looked right and the page was not.
The question is never whether people left. It is whether they left with what they came for, or left to keep looking.
The causes, in the order they are usually the cause
1. The page answers a different question from the one asked
By a wide margin the most common reason, and the one least often diagnosed, because from the inside the page looks obviously relevant. Somebody searching a how-to question who lands on a sales page has not found the wrong page by accident; the page was aimed at the wrong query. Matching content to search intent covers the audit that fixes it.
2. It is slow, particularly on a phone
People abandon slow pages, and they abandon slow commercial pages fastest, because there is always another result. Measure the field data rather than a lab score: Largest Contentful Paint at 2.5 seconds or less, at the 75th percentile of real visits. Why your website is slow works through the four underlying causes.
3. The answer is buried under an introduction
Four paragraphs of context before the thing the reader searched for is a habit from print, where the reader had already committed. Somebody arriving from a search result has not committed to anything. Put the answer in the first screen and expand underneath it.
4. Something interrupts on arrival
A newsletter popup, a cookie wall with the accept button hidden, a chat widget that opens itself. Each one asks for something before the page has given anything, and on a small screen they often cover the content entirely.
5. The layout moves while it loads
Text reflows, an image pushes everything down, and the reader taps the wrong thing. Cumulative Layout Shift should be 0.1 or less. The fix is always the same: reserve the space before the content arrives, by setting width and height on images and giving embeds a fixed box.
6. There is no obvious next step
A page that answers the question and then stops has done half its job. One relevant link onward, placed where the reader has just understood the effort involved, is the difference between a bounce and a session.
What to measure instead
Bounce rate is a symptom metric. These are the ones that tell you whether the site is working.
- Enquiries, and which page the visitor first arrived on. Not the page they converted from, which usually gets all the credit and did the least work.
- Engaged sessions on content pages, which is the same data expressed the useful way round.
- Whether people return to the search results after visiting, inferred from impressions and click-through against engagement.
- Conversions per landing page, so a page that bounces heavily but produces enquiries is left alone.
- Phone and email link taps, which are invisible unless somebody set them up as events.
A sensible order of work
- 1Split your pages into the two groups in the table above. Stop looking at the site-wide figure entirely.
- 2For the commercial pages only, check whether enquiries are actually being recorded. Submit your own form and confirm the email arrives.
- 3Search the term each page targets and read what ranks. If your format is different from all of it, that is your answer.
- 4Measure the field Core Web Vitals on the same pages and fix anything in the poor range.
- 5Open the page on your own phone, on mobile data, and try to do what a customer would. Time it.
If step two shows that traffic is arriving but nothing is converting anywhere, the problem is upstream of any single page. Traffic but no customers separates the two versions of that, and conversion optimisation covers the stage-by-stage fix.
Questions people actually ask
What is a good bounce rate?
There is no useful universal figure, and any article quoting one to two decimal places is quoting a blend of page types that does not match yours. Compare a page against its own history and against pages of the same kind on your own site.
Does bounce rate affect rankings?
Google has not published bounce rate as a ranking factor, and it has no access to your analytics unless you give it. What plausibly matters is the behaviour the metric sometimes reflects, such as people returning to the results to try another link. Fix the reason, not the number.
My bounce rate jumped overnight. What happened?
A sudden change is usually measurement rather than behaviour. Check whether the analytics tag was duplicated, removed or changed, whether a new script broke on some browsers, and whether a redesign shipped that day. Real audience behaviour rarely moves that sharply.
Should I add related links to reduce bounces?
Add them where they genuinely help the reader, which is usually one relevant link rather than a grid of twelve. Adding links purely to manufacture a second page view improves the metric and changes nothing about the business, which is the trap this whole article is about.
