Attribution: What is it and why is it important?

On CXL

Before I begin this week review, I would like to take a moment and comment on the Growth Marketing minidegree from CXL institute for which I am writing these reviews. I am at the mid point this week, 6 out of 12 weeks done. I have put over 15 hours each week into this course for the past six weeks. I have to say that after only this brief period I feel far more confident with regards to digital marketing, understanding it is key elements and the course I should take to begin marketing my products. I would recommend the CXL institute to anyone, it goes beyond the usual quality of material in online courses, and I guarantee you will not be disappointed.

Attribution

This term – attribution can have different meanings but in its broadest term with regards to marketing it can be understood as: Understanding the context behind the data. Also, it is important to know what attribution is not.

Framing attribution within your tactics and strategies is necessary for understanding what actions return value and have a good ROI and which ones are not. Understanding that attribution is more than distribution of value is crucial step in understanding.

A quick recap of attribution.

First the four models and custom models:

  • Last non-direct click: So, let us assume your cookie tracking is set to 30 days and the user had a total of 4 sessions through this period. With the last, the fourth, being direct i.e., typing in your web address directly into the browser. During this session, the user converted, whatever that is in your case, the attribution for the conversion would go the third session whatever it was. In this case the last non direct click would get 100% of attribution.
  • Last interaction: This is the last actual interaction, the session which caused the conversion. In this situation the last true interaction would get 100% attribution.
  • Linear attribution model: As the name states in this model, you attribute the same amount of value to each step – from the first interaction then through all the steps in between and finally to the last interaction which had the conversion. In this model each session gets an equal percentage amount to each step.
  • Position based model: This model was popularized by Google Analytics as the most accurate. This model favors the first and the last interaction the most while giving everything in between a divided fixed percentage. E.g., the first interaction gets 40%, the last gets 40% and the ten interactions in between would get 20/10 = 2% each.

Custom models: This is the future of attribution – creating your custom model which considers all interactions your brand may have. For example, if you have a physical store and a user sees something through digital, goes and checks it out in the physical store. You would not have this information with the first four models. However, this is not to say you should collect user personal data (although there are benefits to this) but collect all the offline data, as best as you can, so that you can draw conclusions from the aggregate form of this data. This is important to understand so you can attribute value properly to all your channels. You may also market on the radio, TV, direct mail, or something else. Finding ways to incorporate this attribution into your attribution model will go a long way to increase your ROI. Finally, if you have a big company with multiple marketing teams it is paramount that everyone is on the same page with regards to attribution models. In other words, it is not good if two teams use different attribution models.

A brief overview of tactics:

  • CRO – Conversion rate optimization: Usually CRO focuses on a single step in your conversion, or better to say on a single interaction. However, if you have a proper, good attribution model you will understand that conversion if a broader process and it will give you a better overview of all the steps leading to conversion and what CRO represents.
  • PPC – Pay per click advertisement: Deploying your attribution model properly will reduce the chance of under or over valuing your PPC campaigns and will therefore increase the analysis of ROI of PPC marketing.
  • SEO: Good SEO is needed and necessary for you whatever you do to drive organic. This channel is always undervalued, and recent research shows it is undervalued with regards to attribution as much as 16%.
  • Display ad marketing: The follow through from display ads is abysmal however they do increase awareness and it’s important to know and attribute value to them as well. A good attribution model will position their value as best as you can.
  • Affiliates: If you have affiliates depending on their type they can be overvalued or undervalued and therefore you should strive to give them the correct attribution. For instance, content creators are usually the first step in a customer journey and therefore these affiliates get the least attribution usually. On the other end affiliates with vouchers are usually the last step in customer journey and they get the most credit under most attribution models. Understand the user journey to properly value your affiliates.
  • Email marketing: Well, we all know that click through rates from emails are small. You might send thousands of emails and get some or none click backs. But most people saw the email. With a proper model you will be able to understand the value of an impression and the value of a click.
  • TV and Direct mail: There are many options with attribution with these types of marketing. With recent advancements in technology you will be able to gather data at the household level with new TVs which will reduce the difference between the web and TV attribution models. However again value is in the aggregate not individual level, better to say you can gain many insights from the aggregate form if you don’t have access to the individual level. Finally direct mail. You can have a specialized QR codes in each letter to track opened direct mail or utilize some other method to understand the attribution of direct mail.

 

It is paramount, for the long run, for you to develop a proper, unified attribution model which you will use in your company.