When I was in stats class we had to learn z-test by hand even though you could toss the numbers into a calculator and get a value in about 10 seconds. The reason for the long hand is to understand where numbers come from and how you got to your final answer. If you blindly follow the numbers a calculator spits out, you are limiting your view of the problem and setting yourself up for mistakes.
Google Analytics is no different. It is an incredibly powerful calculator but you need to know where the numbers come from in order to make accurate decisions. One such example is when looking at the acquisition report you will see a revenue figure associated with each channel. Great! You can see what channels are driving the most sales, which ones are not, and can readjust budget accordingly. You are rocking on all cylinders using data informed decision making.
Not so fast.
Google Analytics default definition for revenue is last non-direct click attribution. This means 100% of the credit goes to the last non-direct channel before a conversion. This is equivalent to saying that you got married only because what happened on your wedding day, forgetting all of the other events leading up to your wedding like your first date and even your engagement. Remember when she/you said yes? Yeah, that matters too! So let’s count it. Last non-direct click doesn’t sound like a great number anymore does it?
Introducing Assisted Conversions & Attribution
Sticking with the marriage analogy, assisted conversions are the equivalent of each date. On average customers visit your site 4.3 times the days leading up to a purchase before they actually convert (Kissmetrics). Each of these visits is considered an Assisted Conversion.
Diving into the Assisted Conversion report you can see how different channels are involved in the customer journey. Does social attract a new audience? Does email marketing encourage your audience to purchase? What is the channel that pushes your audience to conversion? Assisted conversions will help you with these answers.
The downside of assisted conversions is that equal weight is given to each visit. What if we were able to weigh some assisted conversions more heavily than others? What if we could place importance on the first date, engagement, and other important events leading up to the wedding day.
Welcome to Attribution Modeling.
This report divides & allocates the conversion & conversion value across all assisted conversions according to the specific model. Google Analytics comes with a set of default models and you can, of course, create your own. Before you dive into creating your own model, look at the default ones. Click through all of them and see how your conversion attribution changes.
Compare Last Non-Direct to other models to see how the values change compared to others. Remember that Last Non-Direct is the model used across all of Google Analytics. Can you see how making decisions only based off of Last Non-Direct clicks is just a bad idea? I’ll go into the difference between the models another time.
You now understand the math behind the numbers and can complete z-tests by hand. Your new found deeper understanding of Google Analytics revenue figures will empower you to make smarter decisions.
Have questions about attribution? Assisted Conversions? How to use this information for your next meeting or marketing budget report? Leave a comment or get in touch with me here.