Tracking | Attribution Modeling

Introduction

Attribution related to marketing analytics refers to how we assign credit for a conversion. Users may be influence by multiple ads and interactions before they actually make a purchase or submit a form, for example. 

mceclip4.png

Attribution Models - Each attribution model takes a different approach to assigning credit. Some attribution models are more appropriate than others depending on marketing objectives. 

Last click attribution models - probably the most common attribution model - assigns all credit for a conversion to the last user interaction immediately before they land on a site and convert. If a user sees a series of ads then clicks on one to complete a sales flow, only the ad that linked the user to a conversion session will be attributed with credit for a conversion. 

Understanding attribution models is important when analyzing marketing data. If your attribution models are different between digital marketing platforms and Google Analytics (GA), there may be inconsistencies between reports. 

Example: a user clicks a Facebook ad, linking them to a shopping site. The user then closes the browser session and later returns via email marketing campaign link, and makes a purchase. Facebook's model may attribute credit for the purchase to the ad clicked whereas models applied through GA or ChannelMix may attribute to another step, depending on the model being used. 

GA4 Attribution Modeling

GA4 Attribution Tools can be found under the "Advertising" tab. 

mceclip1.png

Model Comparison Tool - this will allow you to compare channel attributions by model. Choose between any two models to see how the outcomes differ by channel. 

mceclip0.png

Conversion Paths Tool 

This Tool shows two views:

1. (Immediately below) Shows data for user conversion paths with two drop down pickers. The first picker allows you to select Channel, Campaign, Medium or Source. The second picker allows you to choose the model. This first view visualizes the results. 

mceclip1.png

2. The second view, immediately below the visualization shows a table with the data that was visualized above. In this example, we see the most common Conversion paths at the top. The column to the left shows the Medium of the for each step in the Conversion path. 

mceclip2.png

GA3 (Universal Analytics) Attribution Modeling

Multi-Channel Funnel (MCF) Reports

UA Multi-Channel Funnel reports allow you to analyze the path users take to arrive at a conversion. They use look back windows of up to 90 days. This means that any interactions with a cookied user will be linked together so an analyst can review customer paths by applying different models. 

To access these reports in Universal Analytics, use the left column in GA to navigate Conversions > Multi-Channel Funnels. Select the Model Comparison Tools to see the results of multiple models side-by-side.

mceclip2.png

Choose a conversion goal and models to generate the comparison report. Consider the outcomes of each model to determine which is most useful for analyzing your data. 

mceclip3.png

You may find value in the other MCF reports provided, but bear in mind that these reports use a different API and lookback window than other reports in GA. Therefore, we should not expect these reports to match other non-MCF reports one-to-one. 

ChannelMix Multi-Touch Attribution - Rules Based Attribution Analysis

ChannelMix provides its own model comparison dashboard so you can analyze the same data using each. Additional advanced models are available here. Talk to your Client Success Manager to find out how you can get this dashboard.

mceclip5.png

Conclusion

The tools above are useful for comparing the outcomes of different models. Last click attribution is great for analyzing which step leads immediately to a conversion but omits prior steps which could be just as influential. Therefore, it is important to have multiple models available and recognize the shortcomings of each. 

Was this article helpful?
0 out of 0 found this helpful
Have more questions? Submit a request

Comments

0 comments

Please sign in to leave a comment.