Interpreting Website Statistics: Embracing Perspective

Analytics can be tricky to understand. Commonly, people want a clear decision rule. We want to know what is good and what is bad. This is, for better of worse, not possible without a perspective. In this post, I am going to explain how to put web analytics into the perspective of your whole site and gain better insight as a result of it.

Statistics Are Relative

Any metrics that we measure either on or offline will always be relative to something. This is a key concept of statistics and is important to understand. We are given a number which by itself is just a lone number. To be able to say something meaningful based upon it, we need to put it in relation to another number to get a feel for its size.

For example, say your e-commerce site has five purchases in a given day. Is this good or bad? We can easily put this in relation to say, the normal number of orders that a common day has. This says something about whether we are below or above an average. Additionally, putting this number into the context of how many visitors the site has, gives us an addition aspect of whether this metric means that we can be proud, or should re-think something.

Statistics Need Context

The latter example shows the importance of a context. Without setting a metric in relation to other measured metrics, we don’t get a full picture. For example, let’s say we notice that our site has a bounce rate of 60%, which is not uncommon. Again we ask ourselves: Is this good or bad? Without knowing the context, it is impossible to do any form of analysis.

Here we need to look at our site to find possible factors of explanation. What type of content do we have? Are users entering our site on pages where they are likely to be satisfied right away? Could it be that users are having a hard time finding something interesting to move on to? Are we writing the content that users want to read? The questions that are raised by this metric alone are several and interesting. However, just looking at this metric by itself gives us indications but no clear answers.

Statistics Need To Be Viewed Together

As is the case with the bounce rate above, it raises several interesting questions. Many of these questions can’t be sufficiently answered by looking at one metric alone, but could be answered with the full perspective of several statistics. Following the example, we can build ourselves this perspective by looking at the bounce rate in relation to how much time users spend on the site and the average amount of pages per visit.

By doing this, we can better answer whether we are writing content that people want to read. A high bounce rate, but a high amount of time spent on the site at least gives us an indication that people are reading, whereas a high bounce rate and low time on site gives us the indications that people are not finding our site useful, and cannot find something else useful to move on to within the site.

As you can see, by relating several metrics, we are able to color in the picture to answer the questions that one metric alone might raise.

Setting Your Own Goals

Finally, you need to set your own goals for your site. It is impossible to make any meaningful statistical improvement unless you have goals. To see a metric drop, or increase means little unless you put this in relation to how you wish things to perform. Again, we need the relation for data to be meaningful. A 1% increase in sales might be amazing for one type or store, while it may not for another.

Set your own goals, make yourself a complete picture of the visitors on your site and their behavior by combining metrics, analyze the site in context of what you are offering and relate statistics to other numbers and you will be on your way to using web analytics in a useful fashion that can help you improve your site.