It is critical to track user acquisition source to identify the best performing advertising campaigns. In this tutorial, we are going to explore Google Analytics's source attribution process. In other words, what piece of information is recorded when.
Attribution is about specifying a referral source of a particular activity. Those activities are typically "macro-conversions" or "micro-conversions". Macro being things like purchases, micro being things like registration, email sign-up, blog comment, etc.
Each time one of those things happens, you have to record some referral source information. So what is the source?
The reality is that users often come from many sources before they hit/commit a micro or macro conversion (i.e they may come to the site via organic, then leave, then come via paid search, then leave, then come directly to the site itself.) This source tracking information is often "provided" to the site via UTM parameters, but there are more sophisticated systems out there too. For our purposes, we will focus on UTM.
When the UTM parameters are specified on the URL, these get parsed out and placed into a Google Analytics (GA) cookie. If a website doesn't have GA, there is no point in having UTMs. GA has rules for how it deals with a user who hits multiple URLs with UTMs in them in the course of their lifetime (more on that later). Therefore, when a micro or macro conversion happens, whatever is in the GA cookie would get stored into the DB if they are capturing UTM info into their DB.
Whatever is in the GA cookie before a conversion, which most of the time linked to the last source of the user's visit to the website, is what is recorded in the database. This refers to Last click attribution. Note that the GA cookie only overwrites the previous UTM parameters if the user clicks on a new URL that contains a new set of UTM parameters.
For example, if I visit a website via paid search, then I visit again in via organic search, and then I visit the website directly or via an email link without UTM parameters, then my GA cookie says the source is organic. It neglected the direct source. Now if i visited the website from an email link with UTM, then the GA cookie would say that the source is "email". Therefore, if there are existing UTM parameters in the cookie, and i come in via direct, the GA cookie will always show the UTM parameters rather than "direct". (Note: A specific user's GA cookie parameters will be erased when the cookie expires, or when a user clears his or her cookies in the browser.)
Some paid attribution tools will allow you to capture "the pancake stack" of sources in a user's path. In that situation, in our above example, first click attribution would tell us paid search. Alternatively, a minority of websites implement their own cookies that capture a pancake stack and store the first source into their database.
GA has some more robust functionality in their web interface that lets you perform four different attribution models: first click, last click, linear (divide revenue equally across all the sources in the path), and weighted (a custom thing).
Now that you understand what is the attribution model for each micro or macro-conversion, the question becomes what do you do with the totality of a user's conversions? For example, let's say we just look at the UTMs recorded based on the GA last click logic:
Here's where you ask: How much revenue did I get from paid search? From email? From organic? You could say the answers are 5, 50 and 10 (i.e. whatever the Last source was), or you could also say that you attribute all revenue to the first source (i.e. all 65 goes to organic). You could also apply some weighted analysis or apply the linear model (i.e. roughly 22 each).