analytics.

Web Analytics that Lie

With the recent proliferation of free web analytics tools, and most importantly Google Analytics, it has become a standard practice to install analytics on websites. This is an important move for any business, as the data collected can provide invaluable insights. It also, however, presents a danger – if you don’t understand the data and haven’t been trained to see it in context, not only will your web analytics reports not give you any useful insights, but they may even lie to you and lead you to make incorrect business decisions.

When Analytics Lie

Here we will examine a specific instance where the web analytics lied. At an initial glance, Google Analytics shows us a bounce rate of 18% for the site.

Sounds good, right? 18% a pretty good number; it must be indicative that the people who visit the site find it relevant and useful. Except it’s not. It’s lying. After deeper analysis, we made some changes to the site to correct for the problems in the data capture, and the true bounce rate came through.

The true bounce rate is up around 55%, hardly the 18% that we had thought. Had we stopped our analysis at a first glance, without actually looking at the data on a deeper level, we would have been making decisions under the faulty assumption that our site was already well optimized. The truth is, it needs some work.

Deeper Analysis to get the Truth

Let’s take a look at some of the factors that tipped us off that something was wrong. One, a bounce rate of 18% is really, really good – almost too good. That made us suspicious. So we took a look at bounce rates for individual pages – and we found that all of them, with the exception of two, had a bounce rate of about 50%. So why would only two pages have such a low bounce rate, while all others did not?

We took a closer look at the two pages. One was the home page, and one was an iFrame that was embedded on the homepage. What was happening is that when visitors entered on the home page, the javascript from the Google Analytics would fire not only on the homepage, but also on the iFrame. As a result, all of these visitors were almost instantly counted as visiting two pages, turning their bounce rate to almost 0. This was skewing the entire average down to 18%.

A simple change so that the javascript did not fire in the iFrame fixes the problem, so that our data shows us accurate results.

The point is, don’t just settle for a quick glance at the analytics – then you give them the chance to give you wrong assumptions. Dig around in the data and make sure you are getting the truth from your numbers. If you don’t ensure that your analytics are giving you an accurate picture, they may be leading somewhere you don’t want to end up.

Author:.

Curtis Seare is the Chief Analytics Officer of Vault Analytics, a firm that uses marketing analytics to help companies develop and implement measurable marketing programs.

Go Deeper | Website

Have a question for Curtis?

* Required information
Name:
Email Address:
(never displayed)

Your question or comment:
Human? Enter the word hand backwards.
 
Enter answer:
 
Tell me when Curtis responds to me.
 
Remember my form inputs on this computer.
 
 
 
New Graphic
Subscriber Counter