Metrics Spotlight: Why You Should Ignore Time Spent on Site

Apr 24, 2012

Metrics Spotlight is our new series that will help you understand the important numbers behind successful digital measurement strategies.

Hi there, I was just thinking about another interesting web metric that I want to talk to you about – "time spent on page."  Do you have a moment? Not yet?  That's ok, Just leave this page open and I'll wait for you.  

I know, it sounds silly, but web pages wait for you all the time.  They "wait" while you get a cup of coffee, go to the bathroom, chat with someone who just walked in to the room.  All of this time could be counted as "time spent on page," but that isn't really what we're thinking of when we start bragging that our average visit is 5.5 minutes long, is it?  Before we can start using "Time Spent" metrics it's important to understand how they are calculated and why they can be deceiving.  

How do we measure it?   Let's start with a simple visit.  I land on the homepage of WYXZFM.org and I start browsing the headlines.  When I open the page, Google sends a tracking call and records that my visit started at 08:29.  When I finish reading the headlines and decide that I want to read a story about miniature horses, Google sends another tracking call that says my visit included a page view for the mini horses story at 8:31.  Google can then calculate the time between when I opened the first page and when I went to the second page.   This difference between timestamps is the time that I spent on the homepage of WYXZFM.org.  OK, that makes sense -  my Time Spent on Page is 2 minutes.

Now I'm looking at these tiny horses and in addition to being really cute, they are also really interesting. I'm a very engaged visitor; I scroll to look at pictures.  I read the story.  Maybe I even copy and paste part of the  article into an email so I can send it to my mom.  Then I realize that it's 8:37 and I was supposed to call my veterinarian's office today to check on a prescription for my dog.  Ack!  I leave the computer up and running, but I turn away to make the call.  By the time I'm done with the call, it's now 8:45.  It has now been 16 minutes since I started my visit to WYXZFM.org.  I realize that I should really get back to work; So I close the browser.  How much time did I spend on the website?

In this scenario there are actually a few things that make the numbers tricky:

Google calculates time spent on site by comparing the last time stamp to the first.  Exits are when someone closes the browser or navigates to another site – these don't send a Google tracking call so the last timestamp that Google has is 8:31.  That means 8:31 – 8:29 = 2 minutes.  

Whoa, that's way off.

Time Spent on Site never includes the last page of the visit.  Even though I had the Mini Horses story open, I read it and I scrolled around, Google has no data about how long I was on that page.  This might be less of an issue for a commerce site where that last page is merely a confirmation page that people aren't reading anyway, but on station sites people look for information until they have found  the right thing.  If I spend 10 minutes reading a really great article and then close the browser, Google never records any time spent on that page.  Even worse, if I entered that page from a source like Facebook where I'm very likely to bounce as well, because  Google records no time for that entire visit.  Traffic from social media bounces at rates as high as 92% or more.  Would you throw away 92% of votes before you started counting them?!

But what about that phone call that I made? Let's say that I clicked back to the homepage at 8:45 instead of leaving the site.  Then Google sends another tracking call at 8:45.  Now Google thinks my session looked like this:

Again, Google compares the last timestamp to the first timestamp, so now Total Time spent on Site is now 16 minutes.  Wow!  That's incredible, but I wasn’t actually even looking at my computer.  Google doesn't know the difference between active productive time spent reading a really interesting story and the times when I get distracted or even wander away without closing my browser.  

We recently launched an enhancement to Core Publisher that allowed stations to include NPR Music Live Events on their local sites.  Since this launch people have been wondering if MLE has had an impact on Time Spent on Site, but here's an example of why I don't have much faith in those numbers.  If a loyal listener and site visitor is following you on Facebook, they might see the link you posted to a Live Event.  Great! They click the link and land directly on your MLE page where they watch the entire one hour performance only to close the window without clicking on any other content on your site – Google does not report the time spent on that page.  It simple doesn't have the data it needs to tell you how long you were there.

There are ways around this problem: some folks have tried using event tracking that "pings" your site at regular intervals or even implementing a "close" button that sends a tracking call before closing the window.  Each of these has drawbacks and limitations; pings can max out your tracking call limit – leaving you with incomplete data.  Using a close button doesn't always make sense to the user since they are accustomed to using the browser functions to close the window.  

Time Spent on Site numbers can be interesting over time but I recommend that you try not to become too invested simply because the accuracy is so low that even large swings in the numbers can have very little substance behind them.  For a site with an 75% bounce rate, Google is excluding 75% of the sessions before it even starts calculating the numbers.  For me, I've found that my time spent on the Google Analytics site isn't well spent looking at "Time spent on Site."