I've been concerned about the high bounce rates received by this site for a long time: Does the high number mean that my articles are irrelevant to user searches, difficult to understand, or just plain boring? While I'm sure that's true to a certain extent, my return visit rate isn't that bad, but even the bounce rate for the returning visitors is still very high?!
Well, hopefully all isn't lost: Maybe I can gain some insights by having a look at exactly how a bounce is defined, analyzing the type of content my site contains, thinking about the reason people might be interested in such content, and doing some testing using Google Analytics event tracking to reduce bounce rates for people who stay on a page for a while. So, let's get started and see if I can boost my ego a little bit here….
What is a “Bounce” in Google Analytics?
Google defines the Bounce rate metric as "the percentage of visitors that see only one page during a visit to your site": meaning that the user either left your site without loading another page (either via a link or refreshing the current page), or didn't load another page before their GA session expired. This is different to what's commonly called a “Short Click”, where a visitor quickly notices that the page they've landed on doesn't have the information they want and promptly moves back to their original location.
This raises the following two questions:
- Does it matter if people leave my site after only viewing a page?
- Are people actually reading my content or just leaving my site very shortly after arriving?
The answer to the first question requires some thought about the value of the content on my site, while the second question can be tested using some modified GA tracking code.
Why Might a Visitor Want To Read the Information on My Site?
The relevance of a bounce to the value of the content on a site depends on who's perspective you take:
- A site owner will normally want the visitor to perform an action on their site that's valuable for them; however, it usually requires more interaction with a site before a visitor is engaged enough to take an action that has real value.
- However, a visitor could well see a bounce as a positive outcome: When you're browsing Google for a quick answer to a question, then finding it on the first page of a site is ideal; the visitor will often leave satisfied with their visit, and may even come back again.
While this is a very narrow analysis of the topic, it's enough for me to form a basic hypothesis about my bounce rate: Given that most of the articles on my site relate to programming topics that developers usually want to find a quick answer to, I believe that my bounce rate could be caused by visitor satisfaction, instead of visitor discontent. A small number of visitors want to find out about me and my products, so will browse further into the site and contact me, which is where my value comes into play; however, most people will find an article on a programming topic, scan through it, copy some code snippets into their IDE, then leave—possibly coming back in the future for more information.
So that's the hypothesis, but I now need to put an experiment together to see if it improves my understanding of the situation.
Is My Site Receiving Too Many Short Clicks?
Ideally, I'd be able to actually survey each visitor to find out if they found my content useful or just meaningless dribble, but in reality most people don't answer surveys unless they're really happy, aggravated, or just plain bored. This means I'm going to have to use the usual method of analyzing click streams with Google Analytics.
I think that a reasonable indicator suggestive of a user actually finding my articles useful would be based on the time they spent viewing them. While my visitors could be just staring blankly at my site, I think that if they spend more than two minutes on a page, then they could well be doing something of value; like reading a tutorial or copying some code. I just need to tell that to Google Analytics.
Changing the Bounce Rate for a Visit Using an Event
The simplest way to stop the bounce rate increasing for a visit is to register a new page view using the
__trackPageview() function; however, this approach isn't desirable as it will impact all your other metrics that relate to page views. A better solution is to use Event Tracking because, by default, sending an event to GA will cancel out the bounce. So, just add code similar to the following onto a page, and your bounce rate should go down to zero:
You just need to make sure that the last value is omitted or set to
false because it's the
opt_noninteraction parameter that stops the event counting towards the bounce rate when it's set to
Informing Google Analytics that a Visitor hasn't Bounced
This can now be combined with the
window.setTimeout() function to wait for a specified period of time, then send an event to GA to ignore the current visit in its bounce rate calculations. Here's a function, called 'bounceTimeoutFunc()', which does just that:
The first line of the function retrieves a reference to the default GA tracker using the
_gat._getTrackerByName() API call; note that this presupposes that the standard tracking code has been installed and run before this function is called.
The next line defines a variable, called
minSecondsOnPage that contains the number of seconds you want the script to wait until you consider the visitor to be engaged with your site; I'm trying out two minutes to start with, but you might want to set this to a higher or lower number depending on the type of content you have on your site.
Next we have the definition for a function, called
cancelBounceFunc(), that will be called after the specified number of seconds. This function will contain the key line of code that sends a tracking pixel off to GA to register an event:
I'm giving the event a category of
TimeOnPage and an action of
TimeThresholdReached; you can use different values if they're more suitable for additional reporting you want to undertake. I've also set the optional
label parameter to the name of the current page, the
value parameter to the number of seconds I decided to wait, and finally set the
noninteraction parameter to
false to ensure the event cancels out the bounce.
The final line of code simply registers a
timeout event to run
cancelBounceFunc() after the number of seconds defined in the
minSecondsOnPage variable (the seconds are multiplied by 1000 because the event registration function uses milliseconds, not seconds).
Inserting the Code into your Page
The only thing left to do is place the code onto your page; however, you also need to ensure it plays well with the rest of your GA code, so you need to sequence it correctly using the
_gaq: The function assumes that a default tracker is already available so you need to make sure the initial tracking code has been executed before it's run.
This is simple enough to achieve because the function of the
_gaq is to queue operations up so they're executed in the correct order; this means that we just need to place our new function onto the queue and
ga.js will deal with the rest:
Stopping the Event Firing on Subsequent Page Views
You might want to make an additional modification to the code, so it only registers events on the first page of your site; this isn't too important, but reduces network bandwidth a little and will make your event reports clearer should you wish to use them.
A new visit to a site is usually indicated by a HTTP
referer header that's not the same as the current domain, so code such as the following should identify it:
To get this working you need to enter your domain name into the
currentDomainName variable; there's more generic ways of doing this without needing to enter your domain, but this should suffice as an example.
For your convenience, here's the code again ready to be placed into a page, so you can see how it all fits together:
Results After Two Weeks
OK, so after all that, has my bounce rate improved? Well, I guess I wouldn't tell you if the outcome was terrible, but luckily I've seen a positive result! Here's my bounce rate for the past 30 days reported in GA:
You can see the difference in bounce rate by comparing the last two weeks (in blue) with the previous two weeks (in orange). Apart from the blip on the first of April, the bounce rate metric has seen a significant drop; it's gone from an average of 72% to just 28%:
I'm satisfied by my results that people are actually interested in my content: Many people leave my site after only viewing one page, but because of the type of articles I've written, I need to measure how long people spent reading them, instead of how many of them got read during a single visit. However, I should note that this type of tracking will probably work better on an informational site like mine—where success is defined by how useful the information is, not how well it converts into business actions—as sites with a stronger commercial focus need to drive a certain level of engagement, especially if they're paying for traffic.
I think that the most useful outcome of this investigation is that I've cut out some noise, so I can now get a better view of the real bounce rate and time on site metrics, and use them to home in on the content that really is or isn't working.