Almost every ecom manager or marketer knows what their average order value is. It’s one of those things that get brought up all the time.

I showed a colleague a frequency distribution of product sales and pointed out that most of the sales were between $50 and $150. She didn’t believe me because she knew the average order value was $175. She didn’t understand that average is misleading in certain situations and wasn’t the best indicator of the population as a whole.  MATHS TIME!

Remember the bell curve from high school? It represents a normal distribution. What a normal distribution means is that the balance of low and high scores is symmetrical – there are about as many scores that are way below the average as there are scores way above the average.

But not all populations are like this. In finance  and commerce especially, there are far fewer scores below the mean, since $0 is the lowest possible number, but there is no particular upper value. Income, or the cost of shoes or the cost of a house could theoretically be incredibly high, whereas $0 will always be the minimum. So therefore if you create a frequency distribution of your transaction values they will look like this:

Have you ever noticed that income and house prices are presented as medians? That’s because the mean is not a good measure for those populations. It isn’t as useful or representative as it is for normal distributions.

So here’s the point – if your ecom sites order values are a skewed distribution, which they usually are, you should use median as the measure of central tendency. And here is how you find out what kind of distribution your audiences transactions follow.

1. In GA, go to the transactions report. Export as many transactions into excel as you can, 3 months worth would be a minimum. Remember you could export your transactions or your purchases, both are worth doing for this exercise. Your spreadsheet only needs 1 column – the order value. Now we’re going to turn this puppy into a frequency distribution.

2. Decide on the order value intervals that are appropriate to your business. If your maximum order value is ~$500 then use $50 intervals. If the max order value is $5,000, then use $100 intervals.  It would be ideal for each group or bucket to have about 200 individual orders in each one. Create a column of these values to the maximum you’ll require. This is called the bins array.

3.  In an empty cell, type =frequency(

The formula which will appear before you will be

=frequency([data array],[bins array])

4. The data array is your column of order values – just click and drag to select them.

5. And your bins array is the intervals you typed out before. Click and drag to select.

6. When you press enter one cell will have a figure in it. Select all cells alongside the bin array.

7. Now we create the array. Click into the formula bar at the top, then press and hold CTRL, Shift and Enter.

All done, now turn that data table into a bar graph and see what kind of distribution you have.  If it’s a normal distribution then the mean is a good measure of the population. If it’s skewed then you may be better served using the median. Or use some basic maths to figure out your mean and median and plot them on the distribution and see how representative they are of the centre of the population.