As part of the work I like to do with clients (Rob talking!) I enjoy perusing weekly trading reports. It gives a much better insight into the way the ecommerce business is run within an organisation, what the focus is, and general performance. It all helps us drive better recommendations for where to take the site next in terms of optimisation, functionality and so on.

One of the things I often see that I like to be different is how average order value is reported. The very definition of this metric is the average, and averages can always be dangerous. In general we want a higher AOV, this is true. Often though, there can be several different AOVs to pay attention to, hidden in that one average.

For an average to be accurate we need to assume that the distribution of that average is ‘normal’. It looks a bit like this:

nd.png

The issue is that very few ecommerce sites have their orders spread beautifully and evenly around the AOV average. In fact this is quite rare. As a result, we should make sure that we know what this graph actually looks like for your site.

To do this, we need an order extract that just has the value of the order (preferably excluding shipping as this will also skew the results). Then, we want to pump that info into Excel to get the type of graph we have above. I imagine other tools like Tableau could be used to do a very similar thing. In effect you want to create a banded bar chart but with very small band sizes like £0 - £5, £5 - £10 and so on to really see the frequency across the range.

This simple video helps explain what I mean:

Once done, you can see whether your AOV is representative of how most of your customers order, or whether you’ll actually find you may have two very different and important types of order value to concentrate on (probably around your free delivery thresholds or promotional thresholds!)

Happy AOVing.

Comment