why the bounce rate metric in web analytics can be very misleading…

Bounce Rate is often considered as a good indicator of the quality of your webpages and website in general. As a rule of thumb, it is considered bad to have a high bounce rate as it indicates that the user is not finding what he or she is looking for and is bouncing away to another site or back to the search results page.  While this logic holds true in many cases, it does not work very well in several other scenarios.

Also, the technology and mechanism that is used to track a bounce is not totally reliable or fool-proof, which means that it is totally possible that some of the data is captured or calculated incorrectly. I will focus on Google Analytics mostly on this post as it is one of the most widely used web analytics software, but it should hold equally true for most other web analytics software.

Google Analytics uses cookies to track the various interactions of the users visiting the website. There are certain scenarios where the Google analytics tracking code cannot trigger an event, in which case the visit will be marked as a bounce.  This includes actions (among others) like

  • The user closes the browser
  • The user closes the tab
  • The user directly enters a URL into the browser and exits the website.
  • Session timeout, usually about 30 mins of inactivity

While these actions do fall within the definition of a bounce, it is quite possible that the visitor may have spent a substantial amount of time on the page reading the content before exiting. If the time spent could be reliably measured then an exit could be calculated differently and seperately for different times rather than dumping all of these exits into the same Bounce Rate basket.

Another reason that the bounce rate could be calculated incorrectly is tabbed browsing. It is dependent on how the user uses tabs to navigate the site. Different users use tabs in slightly different ways. I mostly open all the relevant links or webpages in separate tabs in a single  pass through the webpage. These tabs are then visited at leisure usually much later, and more often than not often closed without interaction. These could remain open for minutes or hours without an interaction.

So, when using tabbed browsing over multiple webpages from multiple websites at the same time, it is challenging to correctly measure the times or order of navigation.  Also, there maybe variations on the implementation of session management logic among the browsers while using tabbed browsing.

Another reason could be that you have a very highly optimized landing page, which satisfies all the reasons the visitor landed on this page. That means, that the user found what they came for and didnot have to click on another page. Again this satisfied visit is dumped along with the other genuine bounces into the same bucket.

It is quite possible that depending on the webpage or blog post, you do happen to have a relevant external link in the page. The visitor could exit the page by clicking on this external link, which in this case could be considered as a successful conversion. But unfortunately it again end up in the same bounce rate bucket with other ones.


Screenshot of Bounce Rate by Source in Google Analytics


Due to all the above mentioned reasons, it is quite important that you segment and categorize the bounce rate statistics before venturing into fixing it. It is quite possible that it just one rouge referrer, browser, webpage or something similar which is skewing the rate of the entire website. For example, you have a webpage with a high bounce rate of 75% which let’s say is bad. On checking the stats deeper you find that the visitors who come from Facebook bounce at the rate of 98% while all other sources record a much less or good rate of 25% percent. In this case, you can safely conclude that it is the quality of visitors from Facebook that is to blame rather than your webpage content itself.

There are several ways you can segment your data to get a better understanding of why your bounce rate is high or low. Try segmenting the data based on any or all of the following properties

  • Sources (or referrers)
  • Geographic region
  • Type of Browser
  • Language of the user
  • Traffic type (paid vs organic search)
  • New vs Returning visits

These are only some of the visitor specific properties that you can segment on. Depending on the web analytics software you use, there should see plenty more options.