Marketing attribution: beware of the last-click model

To make your media mix more effective, you are tempted to cut the tools with high CPA in order to focus on those that earn you more. Not so fast! Often, costs per acquisition only reflect conversion on the last click. You may be missing a big part of the story and making decisions a bit too hastily. In order to better understand a prospect's journey up to acquisition, we need to take a look at marketing attribution.   

What does 'marketing attribution' mean?

Attribution makes it possible to link acquisition with the right tool or combination of tools. By looking at the contribution of each tool, you will see a very different picture than that drawn by last-click attribution.  

In practical terms, a last-click sale will be attributed to SEA, which is a particularly effective tool. But that's not the whole story! Before clicking on your Google ad, your future customer may have clicked on a display banner first, and then a Facebook ad. Finally, it was only once the customer had decided to buy that they went to Google, entered the name of your brand and clicked on your ad. As the saying goes: It's not about the destination; it's about the journey. This is the very essence of attribution!   

Why adopt this approach? 

Understanding the contribution of each tool to the final acquisition simply enables you to make decisions based on the whole picture and, above all, to avoid making bad decisions due to a lack of information.   As an example, the Facebook page of one of our clients in the travel industry had a very high CPA, brining only one or two sales per sponsored post. However, when we looked at the sales generated indirectly by this tool, the results were very good. While the last-click model made this tool seem unprofitable, the page was actually an effective initial touchpoint. Cutting out this tool would have risked damaging the performance of other tools!   

Which model to choose?

Attribution is a complex area and there are various models available: Think carefully about the model to adopt in order to understand the result.  

• Initial touchpoint: in contrast to the last-click model, this helps you understand how the customer found you in the first place. However, it is insufficient if you want to track your user's entire journey. 
• Linear model: it attributes the same value to each visit, making it possible to follow the user's journey, but not to accurately calculate the importance of each tool. 
• U model: while the previous model flattens the importance of attributions, this model lends significant importance to the first and last clicks, and less importance to the tools in between. Remember that it is up to you to configure the importance attributed to the first and last clicks. 
• Time decay model: somewhat similar to the last-click model, it attributes greater importance to touchpoints closer in time to conversion. Actions performed the same day are therefore given more credit than those on the previous day. Touchpoints further away in time lose value. 

• Algorithmic attribution: more complex, this approach frees itself of the above models and is designed to measure the actual contribution of tools. While it is the most accurate, it is also very demanding in terms of the quality of the data to be processed. If you are unable to provide the data for all tools, including offline tools (quality, online visitor engagement, time stamping of touchpoints, etc.), this model may not function correctly.  
• It is also possible to create personalised models in Google Analytics, where you can choose how much importance to attribute to each interaction. 

How to proceed?

The complexity of this area means you need to use a reliable tracking tool, which will give you the information you need. There are many solutions available on the market. It is often difficult to choose the best partner as the solutions are so similar. When choosing, we recommend that you take two factors into account: look carefully at how the algorithms offered work and make sure that the solution does not take into account fraudulent clicks and impressions.   In order to set up your attribution model, you can use the Google Analytics tool, including Google Acquisition.

You can also create your own model using SQL. However, if you have many and complex data sources, it becomes difficult to manage the attribution model without using a dedicated tool. Several companies have specialised in a data-driven approach, such as Quantmetry, Funnel, Daitaku and Mazeberry, which was recently renamed Easyence. To find the right partner, make sure you specify the various data to be analysed and the attribution model that seems most relevant to you.  

The last-click model is now outdated and, above all, it is too blunt a tool to help you make relevant decisions. However, this remains a complex area to grasp. You need to take the time to choose the model that suits you best and the tool that will then enable you to obtain useful and intelligible data, in order to guide your acquisition strategy.