In the effort to expand my horizon beyond finance and accounting and to learn how others think and analyze, I’ve started reading a book about web analytics: Avinash Kaushik’s Web Analytics 2.0. His name and this book keep coming up as THE source on web analytics so I decided to try reading it. It is dense, meaning the book contains lots of good material on every page, but he is also an extremely good writer and manages to keep it easy reading. But it is not full of fluff.
In order to retain some of the materials, I thought I would write down my “notes” here as a way of learning the materials. Today I am going to go over some of the basic analytical metrics that are currently in use and to point out some surprises in some of the metrics.
First off, the metrics I’m discussing today only tell you the “clickstream” activities; they don’t really provide much insight into why people act the way they do or the thoughts that led to the clicking activities. So with that in mind, here are some of the metrics you might find in your analytical tools:
1. Visits (or visitors) and unique visitors (or absolute unique visitors) – the first surprise starts here. A visit is a session that starts when the person first enters the site and progresses through the pages until he exits. This session gets a unique id. The same person can exit a site and then come back for a new session which would then be counted as another visit. So a single person could have 2 visits. To get at the number of new visitors or unique visitors, cookies come into play, provided that they don’t get cleared out every time you close your machine. But let’s ignore that problem.
The surprise comes in when you look at metrics such as daily unique visitors, weekly unique visitors, or monthly unique visitors. The numbers given may not be quite what you’d expect if you were to count unique visitors manually. The time span you choose is critical in order for these metrics to work: if you want to know the daily unique visitors, the time span should be a single day, not a week or a month. This is kind of a “duh” situation here: if you want to know the daily, then a single day works. If you run a week, then the algorithm calculates each day’s unique visits (that daily figure) and then adds them all up to give you the daily unique visitors which would be incorrect. Same methodology applies to the weekly unique visitors and monthly unique visitors. The only metric that appears to work, whatever time span you use, is the absolute unique visitors.
3. Bounce Rate – This metric basically tells you when your site or page fails to hold viewer. A bounce rate is calculated when a viewer visits a site and then jumps off without ever going to another page. Now, there may be instances where the bounce rate should be high because your site is designed to hold viewer temporarily, for example, blogs. In blogs, people come in to read just the daily blog and then move onto another website. This metric does seem to provide a lot of useful information at a glance and Avinash Kaushik sings paean to this metric.
4. Conversion rate – This metric is basically outcome divided by visits or unique visits. For most companies, I would imagine this would be the holy grail as this metric should lead to revenues or should bring you closer to it. Outcomes could be purchases or sign ups to an email list or whatever it is you want people to do. The choice between visits or unique visits is going to depend on the kind of business model you have. If you have people who come in to browse and do multiple visits before making up their mind, then you are going to want to use unique visits. Visit metrics imply “buy now” business model where you want the person to buy the very first time they visit the site. Before you can use the metric, you should dig into your analytics tool to see how the conversion rate is calculated. And finally, as a benchmark, Avinash Kaushik says most sites have a 2% conversion rate.
There are 2 other metrics – exit rate and engagement – which I’m not going into detail because a) I want to wrap up this post and b) they probably won’t be much use. Exit rate sounds meaningless because everybody has to leave the site at some point and engagement is really difficult to measure. Engagement gets into how a person feels about the site and clicks aren’t going to give you that answer.
Okay, that’s it for today.