Understanding Analytics Bounce and Exit Rates

You may have noticed the "Bounce Rate" column in your Analytics reports. It's in all your content reports and your Adsense earnings reports.

It is expressed as a percentage, and could be anywhere from zero to 100%. There is another column right beside it, which is the "Exit %". What do those figures mean?

If you didn't already know what those percentages indicate, you might have gone to Analytics Help for clarification, and you would have found some terse explanations:

Bounce rate is the percentage of single-page visits or visits in which the person left your site from the entrance (landing) page. Use this metric to measure visit quality - a high bounce rate generally indicates that site entrance pages aren't relevant to your visitors. The more compelling your landing pages, the more visitors will stay on your site and convert. You can minimize bounce rates by tailoring landing pages to each keyword and ad that you run. Landing pages should provide the information and services that were promised in the ad copy.

We'll see later that this unfortunately is not really a complete explanation at all, but at least it would give you the rough idea of what a "bounce" is in Analytics reports.

What about that "Exit %"?

This metric identifies the number of exits from your site, and, as with entrances, it will always be equal to the number of visits when applied over your entire website. Use this metric in combination with particular content pages in order to determine the number of times that particular page was the last one viewed by visitors.

We'll see that this is a little incomplete and possibly misleading too, but for now, let's just examine what you should be thinking about those two figures.

Exits

Sooner or later, everybody leaves, right? So why would you even care about this number?

For many pages, we don't care. Where you would care is if the page is a funnel page that exists to try to send traffic to something more important. This might be a page that describes your wonderful e-book and you want your visitor to move from that to another page where they can buy it or download it. You certainly would not want to see a 100% exit rate on that page, would you?

Many e-commerce sites have a definite path that they want people to follow to complete the sale. High bounce rates mean customers who left without even browsing, and that certainly is not the desired outcome.

I will not forget thee

What happens when a user just forgets about you?

That is, they opened the page, and then the phone rang, or the dog needed to go out NOW, or they went to do something in another browser tab and never came back to you. What does Analytics say about that?

To find out, I created some pages in a special directory and visited them using another browser. I first visited the page called "safariopen" in the image at right, and after waiting a short period, clicked from their to "safariopen2". I then just went about my business, leaving that browser open.

As you can see, Analytics shows an exit from that page. I spend a little time on the "safariopen" page, and that shows, but where's the time on "safariopen2"? I was there a long time - long enough for the browser or Analytics to give up and mourn my apparent demise, but where does that show here? It doesn't.

That's refreshing

What happens if you refresh a page and then exit?  I loaded up "refresh.html", refeshed it several times, and then exited.  Notice that Analytics shows 5 visits and an exit rate of 20%.

If you have dynamic content on a page and visitors might refresh to see that, your Analytic exit stats are all off.

Now normally these lone visits would be mixed in with others so it might not stand out in the crowd, but we'll soon see that there are conditions where you might see some odd things in your reports because of this behavior.

Bounces

Google Aanlytics specialist Avinash Kaushik has stated:

"It is really hard to get a bounce rate under 20%, anything over 35% is cause for concern, 50% (above) is worrying."

That opinion came from his "Standard Metrics Revisited: #3: Bounce Rate" and not everyone agreed. But before we look at some of the carping, let's see what it looks like in our reports.

For all the tests that follow, I did a number of things to ensure correct results. First, I waited at least half an hour between tests. I cleared browser cache and exited the browser and then reset my Verizon router to obtain a new IP address. This is all necessary to be certain that Analytics would see these visits as unrelated to previous activity.

Enter and Forget

For this test, I simply loaded the page and then went off to do other things.

Note that Analytics records this as both a bounce and an exit and does not record any time on the page.

A timeout is a bounce as far as Google knows. The person might have read your page over and over, but Google records zero time on the page. Think about how common that is with tabbed browsers!

I regularly and frequently leave tabs open for hours. Every time I do that, that's a 'bounce'. Even if I read every word on the page, it's a bounce, and (accordingly to Google), bounces indicate less than stellar content. Uggh!

In a forum post, Peter Hoggan correctly noted that Google often knows more than they tell us in Analytics. Furthermore, bounce rate is only one of many factors Google uses to judge your page.

A perfect bounce

To show this, I created a "perfectbounce.html" page, entered it and shortly clicked on a link in it that took me to another page. That's a bounce. Analytics records it as an exit and a bounce. Note that it shows no time. That is surprising as I did wait at least 30 seconds before clicking to the off-page link.  I tried it again, waiting more than 90 seconds; same 0 time recorded.

Bounce and Return

For this test, I entered the page, clicked to an off-site link, and then returned to the page with a back arrow. I then exited the browser.

Notice that Analytics shows two pageviews and no bounce, but a 50% exit rate.

You get the same results if you refresh the page before exiting by way of an off-site link.

What's a bad bounce?

I would sure hate to see a high bounce rate on index pages. I selected a bunch of those from recent visits to my site and although the average is a bit high (59%), many are very low, so I'm not upset by this. The one that has the highest percentage is understandable, because it is an index to off-site content - the Feedburner RSS feeds I provide.

However, the main site index also has a fairly high bounce rate. What's up with that? Other index pages seem to do well, why should the main site index be less successful at drawing in visitors?

We can get some idea by drilling into the keywords that people used to find the page. That's one of the things suggested at that "Standard Metrics" article referenced above, by the way.

When I looked at some of the very specific searches with high bounce rates, I immediately saw a problem. The information these folks wanted is available at my site, but because they were sent to the index page rather than a page that was actually a far better match, they left disappointed.

Searches that are more generic and less focused tend to lower bounce rates - those people seem to realize that the site probably has what they want and go off exploring.

I realized that many of those who left were seeking information about SCO Unix. I used to support a lot of those systems, and there is a great wealth of information to be found at my site about that subject, but I had nothing on the main index page that would tell these visitors that.

I have added a specific link and hope to see better results in upcoming months. I will update this page at a later time, but this should show you why it is important to look carefully at your bounce rates.

Update: I checked my main index page a year after writing this and found the bounce rate had increased 6 points, so adding that link wasn't the problem.

I then realized that I had only looked at monthly figures, not the full year as I had done here, so I adjusted that and found that it has gone down, but only 3 points, so I'm not sure I have learned anything useful.

I've kept checking it - it's now down to 48%, so perhaps I need to make that and other subjects more prominent.

Is this bad or just too geeky?

Here is a series of articles written by "BillV". The bounce rate is high on all of them except the index page. There is one (part4) that is a bit lower than the others, but none of it is very good after the index.

The first thing we'd look at is why that one page has a lower bounce. The answer is simply that it happens to be about QIC tape formats and that gets more search interest from Google than what the other pages talk about.

In looking at the searches that brought people to these pages, my best guess is that the whole article set is far more geeky than most people want. Some people are interested, yes, but most would not be quite as geekish as Bill.

Test Pages

I have a series of (very out of date) question and answers pages where people can read a question and click on a link to see the answer and then proceed on to the next question. I would hope that these would have a pretty good rate, and indeed they do. The over-all rate is 34% and many pages show rates under 15%.

However, there are some pages with high rates.  Why?

The answer seems to be that these pages have links on other sites. I didn't bother to go track them down, but the purpose is likely to illustrate some specific point rather than to send someone to the test set, so exits would be reasonable.   I note that the time on page seems to support that theory.

What else could cause bounces?

Consider someone researching some subject. I do this all the time: do a search, visit the page, copy the URL and perhaps some text to remind me and then back to the search to look at the next page.

I might never go back - the info I copied is what I need for my research and I have the URL if I am going to reference it. 100% bounce but the pages are very meaningful (at least the ones I copied the url from are). If I do return, it might be hours or days later, long after Analytics has decided that this was a bounce.

The Perfect Answer to a Question

What happens when you have some serious problem and you search Google for help? If you get sent to a page that doesn't answer your question at all, you probably click Back very quickly. But if the page does answer your problem, you probably will spend a little time reading the page to understand it and then run off to implement the solution you just found.

Do you exit the page? Do you close your browser? Shrug ... who can say? All we can really guess is that you probably stayed on the page more than a second or two.

When I look at a sampling of my pages with high bounce rates, I do see that the time on page was not insignificant and I also know from examining the search strings that these pages are often very good answers to specific problems. Not always - some search strings are so off the wall that you have to wonder why Google ever sent them to the page at all, but those are a smaller number.

But can you trust time?

Apparently not.  Avinash explains that elapsed time is a very slippery measurement and that tabbed browsing really screws things up!

For search visitors, it's probably reasonable to think that not too many have your pages open in multiple tabs, but you never know, do you?

What else didn't people like about the 50% statement?

You can go read the comments at that page. I agree with those who think that Avinash had certain types of sites in mind and that his advice really doesn't apply to others. Indeed, he himself noted that one exception is the typical blogging site. Many of the visitors will be regular readers who have no reason to read anything but the latest page - they have already read the others on another day.

That suggests that you might want to use the custom segment method (as I demonstrated for Adsense) to break down repeat visitors bounce rates - that can change the look of your reports entirely.

One comment pointed out that people interested in Adsense income via clicks are usually happy with a high bounce rate if the bounce was to an ad.

Several comments pointed out that content based sites (as opposed to sites trying to sell product) often expect high bounce rates for the reasons I mentioned above.

This "Half the Story" link notes that a bad search result (I gave an example above) can cause a lower bounce rate, but that definitely is not what you wanted yo happen!

Another response objected to the "I came, I puked, I left" sentence, saying that for many sites, a high bounce rate actually suggests "I came, I found what I needed, Thank You, Good Bye."

What's the takeaway?

If you think about your pages, bounce rate may not be anything you need to worry about. That doesn't mean you shouldn't investigate the causes - there might be simple changes that you could and should make (such as my adding a link for SCO Unix pages as explained above).

You definitely need to take Avinash's 50% figure with a healthy dose of salt. It may not apply to you at all.

Do you know someone who should be reading this?  Click the Share button below to send it to them easily or to post it to Facebook or Twitter.