I’ll keep it pretty short and go into it more later. But it is a good one to finish out the year. SPC – have you heard of it? Probably not. Statistical Process Control (SPC)

I learned about if from an incredibly interesting webinar featuring Nick Desbarats explaining why most dashboards fail. and how SPC can help save them. I won’t go into detail of its origin, but you should read these great posts for background. Four Thresholds.- Automatically Flag MetricsMore Excerpts from the book, “Practical Dashboards.”

To keep it obscenely short and summarized, think of setting Custom Alerts and Annotations in your Google Analytics when metrics hit one of four conditions:

Crisis – Stop everything and fix! Hopefully this rarely, if ever, happens.
Actionably Bad – Pause what you are doing and figure it out
Actionably Good – Someone should get an ‘atta-boy’
Extraordinary – Someone should get a promotion, well, at least a raise. Sadly, this will probably not happen much, if ever.

But how do you know how to set them? It’s not statistically or mathematically hard. Using the GA Store Demo account as an ecomm example, I looked at the Ecommerce Conversion Rate. Because it is an ecomm example, I set two versions of conditions, January – October and November – December.

But for this, already too long, post, we can just look at January – October.

Just by eyeballing the peaks and dips, I can easily spot the ‘Crisis’ and ‘Extraordinary’ (which becomes even more useful if you’ve annotated when there were spikes for unique situations – so you can ignore those). And getting the value is easy by mousing over that day.

To get the ‘Actionable’ conditions, I can use the eyeball/mouseover technique or export the data to Google Sheets and use conditional formatting.

So using a combo of eyeballing and conditional formatting in Google Sheets, I came up with the following thresholds:

Crisis – 0.00
Actionably Bad – 0.04%
Actionably Good – 0.40%
Extraordinary – 1.12%

Notice how different the values are compared with what they would be if they were, say 20%, above or below the average (0.13%).

So there! Long post. But pretty useful, don’t you think?

Leave a Comment

Your email address will not be published. Required fields are marked *


Enjoy this blog? Please spread the word :)