Improper Measuring Techniques

We have all seen many presentations and dashboards that featured averages and percentages. What we don’t always notice is how poorly they are commonly used to convey what is salient. The methods in which these measures are employed are flawed if they do not help you understand current performance against objectives.

What exactly is an average?

I know most assume that the average being used is the mean, which is the basic arithmetic average of all samples from a population. Unfortunately in business presentations or dashboards the author might be deceptively using the average to represent the median (The middle ranked value in a series). The average being used could also refer to the mode (the value with the highest frequency in a series). In optimal conditions the distribution of the sample or it’s population is normally distributed:

In this situation the mean, median, and mode would be approximately equal. The data you might be examining in your industry however might have a multitude of variables affecting  instances of operational performance scores, employee salary, or quarterly budgets. These are all data that can easily skew from normal distributions.

So be careful as the data you are using to support important business decisions could be manipulated in an attempt to persuade or be more compelling.

 Properly Leveraged


Take a look at this Google Analytics dashboard screenshot:

Neat…we know the average visit duration is 3 minutes and 15 seconds, but what do these rows in the overall dashboard tell us? We didn’t gain any insights from this and yet many dashboards feature averages of various metrics. So how do we fix this dashboard view? I am a fan of actionable data, sure there are some statistics that are simply “interesting” to know, but I feel that dashboards should always offer decision-making support.

A method to remedy this is to branch or slice the data relative to the overall average. In other words let’s see what this overall average is comprised of:

This isn’t the best possible display of the data but it will offer context to decision makers. In order to produce questions make sure your presentations and dashboards that use performance indicators with averages are shown segmented.