Some Thoughts on Exit Rate and Bounce Rate

I was thinking the other day about the relationship between exit rate and bounce rate. It’s often assumed that there’s some type of mystical relationship between these two metrics, so I thought it would be worthwhile to dig deeper into this relationship to see what’s actually going on.

First, we can define these measurements using rather simple equations.

exit rate = exits / pageviews
bounce rate = bounces / entrances

Now if we think about it, every visit to your site has an entrance. And unless you have visitors who stay active on your site 24/7, taking 15 minute power naps in lieu of actual sleep just to keep their current session alive, every visit to your site will also have an exit. Therefore, if we’re talking about the exit rate and bounce rate of your site, we can say that entrances = visits = exits and make the appropriate substitutions in the above equations:

exit rate for the site = visits / pageviews
bounce rate for the site = bounces / visits

This would seem to indicate that if the number of visits increase then exit rate will increase while bounce rate will decrease, and alternatively, if the number of visits decrease then exit rate will decrease while bounce rate will increase. Of course, this assumes that visits are independent of both pageviews and bounces, which they aren’t. So to understand this relationship, we have to think about the quality of visits that we’re getting to the site.

If the additional visits to the site were of the same quality as we were getting before, this would mean that the average pageviews for the site would be the same as the previous visits. Average pageviews, however, is the reciprocal of exit rate for the site, so exit rate would remain constant in this case. We would also expect roughly the same percentage of single pageview visits for this extra traffic, so bounce rate would also remain constant.

If, however, we obtained higher quality traffic, then average pageviews would go up, so exit rate would go down. Similarly, we’d expect a higher percentage of multiple pageview visits, so the bounce rate would also go down. Obviously, if we obtained lower quality traffic, then we’d expect exit rate and bounce rate to go up instead.

So how close is this relationship then? If we’re given a particular bounce rate or exit rate, for instance, can we reasonably predict the range of the other metric? Let’s start with the relationship between bounce rate and exit rate for a single page.

If you have a 0% bounce rate for a page, this tells you nothing about its exit rate. If the bounce rate for a page is greater than 0%, you know that your exit rate will also be greater than 0%, but other than that, anything goes.

If the exit rate for a page is 0% or 100% then the bounce rate for that page is 0% or 100% respectively. An exit rate for a page of anything other than 0% and 100%, however, tells you nothing about bounce rate. So aside from a few extreme cases, the relationship between these two metrics for a page is just about zilch.

So on the single page level, knowing one of the metrics in question gives us little to no information about the other metric, except in certain extreme cases. So let’s examine this now at the site level.

If your site has a 100% bounce rate (egads!) it will also have a 100% exit rate. If the bounce rate is less than this, however, all you know about your site’s exit rate is that it’s greater than 0% and less than or equal to 1 / (2 – Bounce Rate).

Similarly, a site with an exit rate of 100% (the horror!) will have a bounce rate of 100%. An exit rate of 0% for a site is impossible unless no one visits your site. Other than that and the bounce rate for your site must be less than 100% and greater than or equal to 2 – (1 / Exit Rate) or 0%, whichever is greater.

Looking at these relationships raises an obvious question: why would we want to look at maximum possible exit rates and minimum possible bounce rates anyway? The answer is that we wouldn’t. There are much better uses of our time, such as looking at trends.

So, it turns out that bounce rate and exit rate have less to do with each other than we might think. It might even be best to think of them as entirely separate metrics. My intention over the next few weeks is to write more about bounce rate and exit rate to clear up any confusion as to what these metrics mean and talk about how we can use them.

So please, feel free to leave any comments about this topic.