Marketing Attribution: Assigning Conversions to Individual Sources and Mediums

21. 3. 2023

Introduction

Marketing attribution is a key element in measuring the success of marketing campaigns and optimizing budgets. It helps identify which sources and mediums contribute the most to achieving conversions. In this article, we will focus on attribution models used in Google Analytics 4 (GA4) and compare some common models, such as the last click non-direct and linear attribution models. Finally, we will briefly discuss data-driven attribution, such as the Shapley model and first- and second-order Markov models.

Attribution Models in GA4

The default attribution model in GA4 is the “last click non-direct” model, which assigns the conversion to the last visited source or medium if it is not direct traffic. This model enables simple measurement of the performance of different marketing channels and campaigns.

Differences between Last Click Non-Direct and Linear Attribution Models

1. Last click non-direct model: This model assigns all the conversion value to the last visited source or medium, provided it is not direct traffic. While this model provides simple measurement of campaign performance, it can be misleading as it overlooks the significance of other touchpoints that contributed to the user’s journey to conversion.
2. Linear attribution model: In contrast to the last click non-direct model, the linear model evenly distributes the conversion value among all touchpoints on the user’s journey to conversion. This model provides a holistic view of campaign performance and takes into account all sources and mediums that contributed to achieving conversions.

Data-Driven Attribution: Shapley Model and Markov Models

1. Shapley model: The Shapley model is a data-driven approach to attribution based on cooperative game theory. It distributes the conversion value among individual sources and mediums based on their contribution to overall success. This model provides a fair distribution of conversion value among different marketing channels, thereby helping to optimize the marketing budget.
2. Markov models: Markov models (first and second order) are also data-driven approaches to attribution that use probability theory to analyze customer journeys and assign conversion value.
   • First order: The first-order Markov model considers only the previous touchpoint when analyzing the user’s journey. In this model, the conversion value is assigned based on the probability of transition between individual touchpoints.
   • Second order: The second-order Markov model takes into account the last two touchpoints, allowing for better analysis of more complex customer journeys. This model provides a more accurate and detailed assignment of conversion value among individual sources and mediums.

Conclusion

Marketing attribution plays a crucial role in measuring the performance of marketing campaigns and identifying the most effective sources and mediums. GA4 defaults to using the last click non-direct attribution model, which, while simple, can be misleading if other touchpoints are not considered. The linear attribution model offers a holistic view of campaign performance by evenly distributing the conversion value among all touchpoints. Data-driven attribution models, such as the Shapley model and Markov models, bring more sophisticated methods for assigning conversion value and enable the optimization of marketing budgets.