I paraphrase a colleague when I say “everything before a conversion is a tease”, however not all conversions are equal. In this post I’m going to go over how each model attributes conversions, as well as why we need each one. The slightly illusive but ever so useful Model Comparison Tool is located under the conversion tab in Google Analytics.
By far the most commonly used attribution model, this model does what it says on the tin; all the conversion’s weight is put on the last interaction. The final point of contact before the conversion is given all the credit.
This attribution model shows what people convert on; it gives you a sense of what their final point of contact is.
It works well for conversions that require less consideration and repeat visits.
Similar to the Last Interaction model, this one puts full emphasis on the last click assuming it is not a Direct Click. If the last click is a direct click, it will not be counted and the next most recent click will be given the conversion.
Non-Direct filters out clients that already know about your company, and it shows you a more representative view of what lead to the Direct Click conversion.
The last paid click is given full value in the Last AdWords Click.
The Last AdWords Click gives a good representation of what AdWords keywords/ads are converting or leading to conversions. It also identifies what keywords/ads aren’t having any impact on the conversion funnel.
Quite simply, the first interaction that has lead to a click is the one that is attributed the conversion.
The First Interaction lets you figure out what created the initial awareness, it is effective for branding goals and to see what attracts customers (such as ad copy, keywords, image ads, etc.)
This is where things start to get a little trickier
The linear model takes each point of contact into an equal split, meaning that a single conversion will be split between all interactions.
It gives a broad picture of what types of interactions lead up to the conversions. Since the emphasis is equally divided, it is good for when many points of contact are required for purchases or conversions.
The more time that has passed since an interaction, the less weight is put into the interaction. Therefore the touch points nearest to the time of conversion gets accredited the most value. The half-life of each interaction’s credit is 7 days, meaning each 7 days the credit is reduced by half.
Similar to linear that it shows the points of contact, but it also shows what people are more likely to actually convert on since the closer the clicks are to the conversion the more they are stressed. This will also give a directional look at the process in which people convert, from when they start to the final conversion.
This is also useful for time sensitive campaigns so that we know which campaigns converted or lead to eventual conversions.
It is essentially a mix of the First Interaction with the Last interaction. The slight difference is that the credit is distributed so that the intermediary interactions get credit, just less.
Being a blend of the first and last interaction, this allows us to see what introduced the product and what brought in the conversion.
Written by Jeffrey Chang