e-Commerce websites will be the Mecca for most holiday shoppers this year. Forrester predicts a 13% growth over last year’s holiday season. The general influx of holiday traffic to the site brings about the flurry of data from online activities. Extrapolating and interpreting this data for valuable insights could be a very tricky business if you do not know how to navigate around the curve balls that outliers throw at you.

Who are the outliers

The dictionary defines Outliers as:

Outliers [out-lahy-er]
noun
are defined as

1. Statistics
a. A person whose abilities, achievements, etc., lie outside the range of statistical probability.

We define Outliers as: A transaction that is 3 standard deviations above the average value. We only focus on the high end of the spectrum to find outliers; they tend to be the top 1% of orders. outliers

 

The standard deviation is a numerical value used to indicate how widely individuals vary within a group. If individual observations vary greatly from the group mean, the standard deviation is big; and vice versa.

DogSD

Source: MathisFun, Standard Deviation and Variance

Influx of holiday traffic = increase in the number of outliers

So what happens when your website gets more outliers? How does this affect you? For digital marketers, outliers are great – they make the numbers look good. For analysts, outliers present the nightmare of skewing data making them unreliable sources for optimizing variables. For e-Commerce, outliers are by definition not representative of the 99% of users you are optimizing for. This means that, if you were to bank on this skewed data, your visitors wouldn’t have the optimal digital experience you were hoping to provide them on your site.

How different is the digital behavior of Outliers

In order to fully understand the repercussions of not excluding outliers from your holiday traffic dataset; we need to look at how different is their consumer behavior. To do this, we created a segment for outliers, buyers, and all sessions to compare performance on a client’s website. This is what we uncovered:

Engagement:

Outliers are big spenders and are far more engaged than the normal user – they visit more pages and spend more time on site.

 Return Buyers:

The outlier segment is 75% return visitor, which is much higher than the normal buyer.

UX or Optimization efforts do not affect them:

Typically bigger spenders won’t get dissuaded from minor UX adjustments or optimization efforts.

GAoutliers

How does this affect your optimizing efforts?

Optimizing for Average order value (AOV):

Outliers account for AOV. AOV for outliers is 4X the normal AOV. As one would imagine, it is unlikely that a user would turn away from their $480 purchase on account of the variables in the lower checkout funnel.

Optimizing for Conversion rate (CR):

Outliers do not necessarily affect the Conversion Rate for e-Commerce transactions. However, in the long run they will have an impact especially when you take into account how much people have spent aka revenue per visit (RPV). By factoring outliers in the decision process you could be leaving valid RPV lifting variables on the table.

Optimizing for RPV:

When you plot the revenue over time for each visitor, you will see the havoc outliers create in misinterpreting your data. It is best practice to optimize for RPV as opposed to CR as RPV is a more holistic measure.

Conclusion:

It is important to exclude outliers when planning your optimization efforts. Their exclusion eliminates false positives and also helps you account for false negatives in your campaigns. Especially in the coming few days and weeks when you will be receiving a spike in traffic to your website, you need to be aware of the increase in the number of outliers and be educated about how they will affect your data.