Analytics Basics: Metrics to mind out for (Part 1)
This article, written by Neil Mason, was originally published on Clickz.com on 21/05/10 and is republished here with permission.
In courses and workshops that I run on web analytics these days I often find myself talking about “data governance”. As the amount of data we get exposed to explodes from web analytics systems, voice of the customer programmes, social media listening tracking and so on, we need to know where this data comes from and how it’s created. By understanding how our data is created we are in a much better place when it comes to interpreting it. So I always think its worth going back to “analytical basics” every now and then and reminding ourselves how some of these metrics are created and what that means in terms of the way we use them.
Unique Visitors
Unique Visitors (UVs) is one of the fundamental metrics of web analytics. These days it’s still used to measure overall level of traffic to the site and is particularly important for those sites that are dependent on advertising revenues as a major source of income. The actual definition of a UV is: “The number of inferred individual people (filtered for spiders and robots), within a designated reporting timeframe, with activity consisting of one or more visits to a site. Each individual is counted only once in the unique visitor measure for the reporting period.”
These days the vast majority of web analytics systems use “cookies” to identify “people”. Also the majority of them use first party cookies. However cookies are not the same as people for all sorts of reasons and so this metric must always be treated with caution and never taken at face value. The main reasons why counting cookies is not the same as counting people are that tracking cookies can be blocked, cookies can be deleted and people may use more than one device or multiple browsers to access a website. In turn each of these reasons can cause an underestimate or an overestimate of the Unique Visitor count. Generally the most common reasons above will cause the Unique Visitor count to be inflated. In one project we did recently for a client we were able to look at the number of cookies we could associate with a unique account number. We found in this particular case that about 10% of accounts had more than one cookie attached to them, but those 10% of accounts accounted for about 30% of all the cookies. So, for Unique Visitors we wary of using the absolute values but as an old boss used to say to me “a trend is a friend” and so assuming that there isn’t a massive shift in the number of people deleting their cookies or using multiple devices or browsers, then it’s probably safe to assume that any change in the numbers over time is reflecting a genuine trend.
The other key thing about Unique Visitors is that it is a “non-addable” metric. The count of Unique Visitors relates to a specific period of time, such as a day or week or month. What I still see happening today is people taking the Unique Visitors for the seven days of a week, adding them up and calling that the weekly Unique Visitor count. This is simply wrong for this reason. If I visit on Monday then I get included in the UV count for Monday and if I visit again on Thursday then I get counted again in the UV count for Thursday. If I then add together all the UC counts for each day or the week, I will be counted twice instead of only being counted once as the UV count as I should be. So adding up the daily numbers will tend to inflate the weekly numbers. How some web analytics system are better than others at producing correct UV counts for different time periods so it’s worth finding out how yours works.
New vs Returning Visitors
As a direct result of the cookie issues describe above, the new versus repeat visitor metrics need to be treated with caution as well. A visitor to a web site is identified as being new if the device does not already have one of the web analytic systems tracking cookies. If the device does not have a cookie then (assuming that the device does not block cookies) then it is given one. The fact that the device has a cookie means that visit gets treated as a return visit.
The problem comes along when I visit the website a second time and either my device doesn’t accept cookies, or I have deleted the cookie or I’m using a new device or browser. The reality may be that I have been to the website before but as far as the web analytics system is concerned I don’t have a cookie on my device and so it will treat me as a new visitor. As a result the proportion of visitors who are considered to be new is generally an overestimate. As before though “a trend is a friend” and so is there has been a change over time, it is probably reflecting some real underling behaviour.
This may seem like basic stuff but with more and more business people being exposed to web analytics data in reports and dashboards, it worth reminding ourselves where this data comes from. Next time I’m going to take a look at the use and abuse of ratios and averages. Till then…