What Is Bounce Rate in SEO?
One of the most misread numbers in this industry. A high bounce rate is frequently a sign that a page answered the question. Treating it as a failure has produced years of unnecessary redesign work. This page exists to say so.
What It Actually Measures
A bounce is a visit where somebody arrived and left without a further recorded interaction. The rate is the proportion of visits that ended that way. That is the whole definition. It says nothing whatsoever about whether the person was satisfied.
The word doing the work. Recorded.
The measurement knows what it captured. Somebody who read every word, wrote down your telephone number and closed the tab has left no trace, so they count exactly the same as somebody who left immediately in disgust.
What it therefore cannot distinguish. Two opposite outcomes.
A page that answered somebody perfectly and a page that repelled them produce an identical result. Any interpretation that ignores this is guessing.
Why the name misleads. It sounds like rejection.
Bounce implies somebody arrived, disliked what they saw and left. Nothing in the calculation supports that reading. It is a name that has done real damage.
What it is genuinely evidence of. A single page visit.
Nothing more. Whether that is good or bad depends entirely on what the page was for, which is the argument this whole page rests on.
Where you will see it. Analytics.
Your site measurement platform reports it. Search reporting does not, since it has no view of what happened after somebody clicked.
The Definition Has Changed
Measurement platforms have redefined this over time, which means a great deal of the advice in circulation is describing a different calculation entirely. Anybody comparing a figure today against one from a few years ago is comparing two different things.
The older approach. One page, one visit.
A visit counted as a bounce if the person viewed a single page and nothing else was recorded. Simple, plus blind to somebody spending a long time reading.
The newer approach. Derived from engagement.
Current platforms work out whether a visit counted as engaged, using conditions around how long somebody stayed and whether they did anything. The bounce figure is then the opposite of that.
Why that matters practically. Reading behaviour now counts.
Under the newer calculation, somebody who stays and reads may qualify as engaged without clicking anything. That single change makes many older figures look alarming by comparison.
What we will not publish. The thresholds.
The specific conditions are configurable and they change, so quoting one would date immediately. Our Google Analytics guide covers the platform itself.
The practical consequence. Check before panicking.
A figure that moved sharply with no corresponding change in enquiries is usually a calculation difference rather than a behaviour one.
The wider lesson. Definitions are not permanent.
Any metric can be redefined by whoever publishes it. That is a reason to understand what a number counts rather than to memorise what a good one looks like.
A High Rate Is Often A Good Sign
Somebody searched for your opening hours, landed on the page, found them and left. They have been served perfectly. The measurement records a bounce. Treating that as a failure is how businesses end up rebuilding pages that were working exactly as intended.
The pattern behind it. One question, one answer.
Any page that resolves a single question completely will show a high rate permanently. That is the page succeeding. No amount of design work will change the figure without damaging it.
Which pages this affects. More than you think.
Opening hours, contact details, prices, a single specific answer, directions, whether you cover a particular area. All of them are supposed to end the visit.
The trade version. The telephone.
Somebody finds your number and rings it. On a mobile they may not even leave the page first. That is the best possible outcome and it frequently records as somebody leaving without doing anything.
Why the misreading persists. One number for every page.
A site total mixes pages meant to end a visit with pages meant to lead somewhere. Averaging those produces a figure that describes nothing.
What it has cost businesses. Years of the wrong work.
Redesigns, added sections and rearranged content aimed at a number that was never a problem. We have seen good pages made considerably worse this way.
The rule to apply. Ask what the page was for.
If it was meant to answer something and end the visit, a high rate is the page doing its job.
When It Genuinely Indicates A Problem
There are cases where a high rate is telling you something real. All of them share one feature, which is that the page was meant to lead somewhere and did not. Knowing the difference is the whole skill here.
The first case. A page meant to go somewhere.
A service page whose purpose is to produce an enquiry, where almost nobody proceeds. That is a genuine finding and it deserves attention.
The second case. A promise that was not kept.
The search suggested one thing and the page delivered another. People arrive, realise immediately and leave. That is a mismatch between what you appear for and what you actually offer.
The third case. Something is broken.
A page that loads slowly, displays badly on a telephone or has a fault. People leave because they cannot use it, which is a technical problem rather than a content one.
How to tell these apart from success. The purpose test.
Write down what the page was supposed to achieve. If the answer is that somebody should have gone on to do something, a high rate matters. If not, it does not.
The confirming check. Compare like pages.
Set one service page against another rather than against a site average. Comparison within a site is far more useful than any absolute figure.
The strongest signal of all. Enquiries.
If a page meant to produce enquiries is not producing them, you did not need this figure to tell you. It only helps you work out where to look.
It Is Not A Ranking Factor
Bounce rate is widely assumed to affect where a site appears in search results. That position is not established. Our approach across this site is to say what is published rather than what is asserted. The assumption drives an enormous amount of wasted effort.
Why the belief spread. It sounds plausible.
A search engine wanting to show useful results might reasonably notice people leaving quickly. Plausible is not the same as demonstrated. The industry has treated the two as interchangeable.
The practical problem with the theory. The data lives with you.
Your bounce rate is calculated by your own analytics, configured by you, using conditions you can change. A figure you control is a strange foundation for anybody else's assessment.
What we say instead. Nothing beyond the evidence.
We do not claim it is used and we do not claim it is ignored. We decline to assert either, which is the same standard applied everywhere on this site.
What follows for you. Do not chase it.
Work aimed at improving this figure in the hope of improving rankings is work aimed at an unestablished connection.
The consistent position. Across the site.
Our jargon buster takes the same line on this term. The engagement measures are treated identically.
What does matter. Whether the page works.
Not because a search engine is watching. Because a page that fails its purpose fails the business.
What To Look At Instead
Ask whether the page did its job. That is a better question and a harder one, because it requires knowing what the job was. Most sites have never written that down for any page.
Define the purpose first. One line per page.
What should happen as a result of somebody reading this. Produce an enquiry, send them to a service page, answer a question completely. One sentence, written before anything is measured.
Then measure against that. Not against an average.
A page meant to end a visit is judged on whether people found the answer. A page meant to lead somewhere is judged on whether they went.
What this reveals immediately. Pages with no purpose.
Most sites contain pages nobody can state a purpose for. That is a more useful finding than any engagement figure and it explains a lot.
The measures worth using alongside. Carefully.
Time on a page and pages per visit have the same limitations as this one. Our engagement metrics guide covers all three together.
What settles it in the end. Enquiries.
Whether people who arrived on that page went on to contact the business. That is the only measure that answers the question directly.
Where these figures earn a place. As prompts.
They suggest where to look rather than telling you what is wrong. Treated that way they are useful. Treated as results they mislead.
Never Optimise For The Number
A bounce rate can be improved by making a page worse. That is not a theoretical risk. It is a real and common outcome of treating this figure as a target. That is why the number should never appear as an objective in anybody's plan.
The commonest way it happens. Hiding the answer.
Move the information further down so people have to scroll. Engagement rises, the figure improves and every visitor now has to work harder to get what they came for.
The second way. Splitting a page.
Breaking one complete answer across several pages so people click between them. The figures improve. The reader has a worse experience and frequently gives up.
The third way. Adding interruptions.
Anything requiring a click before somebody can read. That registers as interaction and it annoys people who wanted the answer.
Why this keeps happening. The target is measurable.
A figure that can be moved is easier to work towards than an outcome that cannot. This is the general failure of proxy measures and this metric is its clearest example.
The test before any change. Would a customer prefer this.
If the change would irritate somebody who arrived with a genuine question, it is not an improvement whatever the figure does.
Where this argument continues. Vanity measures.
Which measures matter covers the wider version. Everything sits on the measuring performance guide.
They found your
opening hours
and left. That
is the page
working.
Any page resolving a single question completely will show a high rate permanently, because it is supposed to end the visit. Somebody who read every word, wrote down your telephone number and closed the tab leaves no trace at all, so they are counted identically to somebody who arrived and recoiled. The measurement cannot tell those two apart.
How we treat it:
We will not tell you what a good rate is and you should be wary of anybody who does. It depends entirely on what the page was built to achieve.
Every guide.
One question.
What each number actually means, how to read the data without being misled, what belongs in a report and what does not, then how to connect any of it to the enquiries the business is actually paying for.
