Attribution models 101

Some time ago (before joining the Good Rebels team) we answered the most frequently asked questions about attribution in Google Analytics on the Aukera website, and

told you some great news: Google has a new tool called Attribution that simplifies and improves how we consult data related to attribution. 

Whilst this tool can be of great help to us to know which of our channels is the most profitable, and therefore the best, we must first understand what attribution is, what the benefits of developing a detailed analysis thanks to Google’s tool are, and what attribution model we should establish based on our business needs. This is precisely what we will see in today’s post. 

Get ready to become experts in all that Google’s Attribution tool has to offer!

First things first: what is attribution?

In order to control the touchpoints and interactions between users and our brand, we need to have control over attribution, as it will allow us to know in detail the performance of each channel. This information will allow us to make informed decisions when it comes to where we should invest our time and money, as well as to establish priorities, and detect opportunities for improvement. 

Up until now, one could determine the contribution of each channel thanks to Google Analytics, but it was a complex process and required a great deal of time and effort. 

However, there is good news! Google has launched a new attribution tool called (surprise) Attribution! 

Born as a replacement for the reports in the Attribution section (forgive  the redundancy), the Attribution tool determines the contribution of each touchpoint to the conversion. In other words, it assigns a value to the different channels based on the attribution model selected and how much they have helped the user convert.

Very interesting, but what good is it to me as a brand?

Attribution is part of the main criteria when deciding how we divide our budget, time, and effort among all our marketing actions. Attribution can also be of great help when it comes to making decisions at key moments in the development of our digital strategies:

• Need for growth: when a brand has already achieved a certain recognition and has consolidated its position in the market, it is time to go a step further. This is precisely where attribution comes into play, as it will allow us to study which channels present growth opportunities.

• Allocation of scarce resources: if the opposite is the case, and instead of being in a growth phase we are at a time when we do not have too many resources, we can use attribution to find out which channel offers the best return on investment.

• Detecting problems: our work as digital analysts doesn’t end once we make the investment decision, as there is a constant monitoring process. The objective is to identify any potential problem in any given channel, such as when the user has to contact the brand several times before converting, or when many days have passed since the user first interacted with our brand until he finally buys our product or hires our service. Knowing that something is happening is the first step to solving it, so once the bottleneck is detected, we can take the appropriate actions.
• Process optimisation: on the other hand, knowing whether the different channels are meeting the objectives we have set will allow us to take action in time to correct possible deviations and to improve the overall performance of each channel.

For all this, it is important to have a defined attribution model, meaning a set of rules that define how and how much value is assigned to each of the channels on conversions.

So, is there more than one attribution model?

In total there are 6 attribution models that Google offers by default.

First click

As the name suggests, this model grants 100% of the conversion value to the first channel with which the user has interacted.

It is a model entirely designed for communication actions that are not aimed at purchasing the product or contracting the service, but are in the initial stages of the process and seek to make the brand known to users.

Last click

The last click model is completely opposite to the previous one, as here 100% of the conversion value goes to the last channel that the user came into contact with.

This is the default model that Google Analytics uses. It is better understood if we take into account that most campaigns, both paid and organic, are orientated to the final purchase, which is precisely what this model is used for. 

Position-based

This would be a mix of the two previous models. 

As you can see in the image, 40% of the conversion value is given to the user’s first interaction with the brand, 40% to the last interaction, and the remaining 20% is distributed equally among the contact points involved.

This conversion model is used when it is estimated that the first and last channels have been decisive for the conversion, meaning that the aim is to prioritise brand awareness and the final conversion itself.

Time decay

This is a somewhat peculiar model in which the value increases as we approach the moment of conversion, generating this distribution of percentages:

• The highest percentage goes to those channels with which the user has interacted the same day they converted. 
• Touchpoints from the 7 days prior to conversion take half the value of the day of conversion.
• Channels that the user interacted with within 14 days prior to conversion would take a quarter of the value assigned to the day of conversion.

These dynamics would continue to play out for up to 30 days. Ideally, this conversion model should be used when a very short sales cycle is expected.

Linear

If we want to consider each of the channels as equal, then we should opt for the linear model. 

This model, which assigns the same value to all touchpoints, is used when the conversion process is long and with frequent interaction between the user and the brand.

Data-driven

This model is gaining popularity lately, as it is based on using machine learning with our real conversion data to assign value to each of the channels. 

Although in theory it is an incredible idea, several factors make its use complicated. First, the model is solely available in the paid version of Google Analytics (GA 360). Second, it has high-performance requirements: you need to receive a minimum of 15,000 search clicks and at least 600 conversions over the last 30 days. Lastly, you need to maintain interaction levels over time for the model to work. 

When it comes to the question “Which attribution model do I use?”, the thruth is that there is no exact answer. We will need to study our product or service in detail, and take into consideration our sales cycle, as well as the performance of our campaigns and channels up until now.. However, it is clear that attribution is a fundamental tool to understand the performance of your digital actions and the investment you make in them. 

Here at Good Rebels, we not only develop all kinds of digital analytics projects but we are also experts in Attribution, so do not hesitate to contact us if you want to measure and analyse the performance of your channels!