Web Analytics, Behavioral Targeting, Online Marketing -

I lecture about web statistics several times a year in Norway. The courses are mainly targeted for beginners. There’s one test that’s an easy test to start with, that I do with participants on these courses. Of course only with those that are willing to volounteer their analytics data. I call it the 5 minute test.

As the name suggests it only takes 5 minutes and for that reason is easy to integrate into a live course. The test is about identifing specific pages, with an insight into what might be done better with this page. The basis for this test is the list of the most viewed pages on the site. A list that’s available in all the relevant web analytics tools.

From this point I will use Google Analytics and numbers from http://www.astrolog.no/ in my example.

The GA Top Pages List includes the following measures (variables):

Pageviews, unique pageviews, time, bounce rate, exit rate, value.

Pageviews or unique pageviews are used as the sorting variable to rank the pages after importance. A store could use the value index as an alternate sorting variable. It’s the time, bounce and exits that have the focus in the 5 minute test.

Generally if two of the three variables are outside the average/normal range, then this is a page to look at. Scan the list and find which pages that stand out.

This test does not hold ground to statistical significance, it’s only an indicator, so all pages highlighted by the test has to be viewed and scrutinized before a conclusion are reached. Especially the bounce rate can be an error because we use the top content pages.

(Bounce rate refers to how the page works as a landing page, but that might be only a fraction on the traffic for that page. E.g. the page might have 1.000 PV, but only 10 entries as landing page. The bounce rate will be based on the actions of the 10 entries that used the page as a landing page.)

Specifically, if these relationships are present, then you have an indication, an hypothesis to validate:

Time is too long, bounce is high, exit is ok:
The page is crowded with information, the information is not living up to the users expectation.

Time is too long, bounce is ok, exit is high:
The information is meaningless, or this is a page where users get their tasks completed – so it could be a completion.

Time is ok, bounce is high, exit is high:
Really not living up to expectations.

The values of bounce and exits can be compeared to the bounce rate of the whole site. If your site has a bounce rate of 20%, then 40-50% bounce or exit on a page is too high. If your bounce rate is 50%+ then you should look for rates over 70%.

Time is a more complex measure. If it’s a page used for navigation, then the time spent on the page should be minimized. More than one minute is too much. On a relevant content page you need to allow for more time. Usually I find that 3 minutes are a good max.limit, but some (science) sites could argue that you should use 5 minutes as the maximum time for one specific page.

Too short a time period could also mean that the content is not relevant. So if you have a high exit, high bounce and only a few seconds of view time, then people are not finding this page relevant. This is the case for the page that is identified in this example. The homepage is about astrology horoscope, while this one page that no one likes… is about chinese horoscope … something the auhor of this website know nothing about.

My recommendation… Delete this page!