☰ menu

Blog

No, Google Is Right: Bounces Are People Too

A recent post by Brandt Dainow entitled “The disturbing inaccuracy behind Google Analytics” has created some healthy debate about website visitors who bounce.

tramp.jpg

Before we get to the debate, let’s define what a bounce is, and what it means.  The basic definition of a bounce is when a visitor enters a website and does not view any other pages.  In most cases, we want to keep the bounce rate as low as possible (my opinion of an acceptable bounce rate is below 30%, and a very good bounce rate is below 20%).

There are some exceptions where a high bounce rate might be acceptable, or even favorable.  First is when you have a website with the primary focus of advertising other sites or services.  So your goal is to get people to your site and have them click on an ad or some other exit link and leave your site – that’s how you make money.  Another case could be when a user finds your site and is looking for a very specific piece of information, like your hours of operation.  They search for you via a search engine, land on your “hours of operation” page, and leave.  Those visitors accomplished the task they needed to and left the site.  Blogs could be another exception, where visitors come and read your latest blog entry and leave.  These really are the exceptions to the rule, though.  In most cases, you want visitors to take some type of action when they come to your site (make a purchase, generate a lead, signup for a newsletter, etc.).

The debate in this post begins with the author’s opinion that bounces should not be treated as visits, and because Google Analytics (and many analytics tools) includes bounces as visits, the data is highly inaccurate.  He believes that it skews some very important metrics, such as visit count, time on site, and a variety of conversion rates.

Visit Count

The author states, “If you see Total Visits as the number of people who entered your site, who reacted to the sales pitch, who engaged with your content, who potentially could have bought products, then you are wrong. It is the number of people who arrived at the front door of the site, nothing more.”  I don’t understand where he is coming from here.  Every single time a visitor comes to a website, there is the opportunity to engage that visitor, whether it is with news, sales, research or whatever you have going on.  If visitors are leaving without engaging, then that is a sign that something is wrong (unless the bounces fall under the exceptions I’ve mentioned).

Time on Site

clock11.jpg

Google Analytics, according to Dainow, calculates bounced visits as 0 time spent on the site.  So when you add all of your 0 minutes from bounced visits to calculated time from non-bounced visits, your average time is site is much lower than it should be.

This is very interesting.  Omniture handles bounced visits differently.  Bounced visits are not included when calculating total time on site because the user did spend some time on the site, but because there is no time stamp from a secondary click, the system cannot determine how much time was spent.  So it does not count it as 0 time, it just does not count it at all.  This is why Omniture’s average time on site is generally higher than what Google Analytics shows for the same site.

This raises the point I was talking about in my “Dirty Data” blog post.  Clearly, these are two different ways of measuring the same thing.  I personally agree with Omniture’s way of calculating this metric.  However, that does not mean that Google’s way is wrong, nor does it mean that the data can’t be used.  This metric should to be used for trending purposes.  If Google Analytics shows your average time on site to be 3:45 last month, and 4:30 this month, can’t you easily see that users were engaged for a longer period of time this month?  Does it really matter that the exact time this month might have actually been 5:00 by someone else’s standards?

Conversion Rate

Brandt says, “The Conversion Rate tells me how successful my site is at selling. It is legitimate to calculate Conversion Rate including bounces, but my personal experience is that it is misleading to do so. I use Conversion Rate to improve my site's sales pitch. People who bounce were never exposed to it, so including them in the calculation means I cannot possibly know whether my sales pitch is working or not.”  Again, I have to respectfully disagree.

Every page on your website is a sales pitch – if it’s not, you’re not using the site to its fullest potential.  No matter where your visitors enter your site, there should be some call to action leading them further into the conversion funnel – whatever that conversion might be for your site.  If you have a high bounce rate then, yes, your sales pitch was not very good.

Where We Agree

Dainow does mention an excellent calculation that he calls “Retained Visits.”  This is calculated as:

Retained Visits = Total Visits – (Total Visits * Bounce Rate)

I love this calculation and will start implementing it as a new custom metric in my Omniture reports. This definitely gives you a picture of “engaged” visits, and can be trended over time.

Take-Away

We need to go back to basics – look at the big picture when it comes to analytics data.  What trends do you see month-to-month, year-to-year?  Are bounce rates going up or down?  Is that a good thing or a bad thing for your site?  Is the average time on your site going up or down?  Is that a good thing or a bad thing for your site?  You get the point…it’s all about using a yardstick, whether that yardstick is 36 inches or 39 inches.

 


Subscribe to our RSS feed

more articles