A Simple Introduction to Multi-Touch Attribution

When people talk about attribution, the concrete ‘what’ and ‘how’ is sometimes unclear. Although multi-touch attribution has been around for years and has gained general acceptance as the model that gets you closest to reality, we still observe companies basing decisions on the results of last-click models in systems such as Google Analytics. We provide a quick overview of what multi-touch attribution is, how it is measured and why you should really, really stop relying on last-click attribution.

What is multi-touch attribution?

Let’s start with what multi-touch attribution is. From the name itself, we can already extract some information:

Multi: multiple channels

Touch: places where you leave a footprint

Attribution: giving a value to something

So MTA is, in essence, the process of assigning value to where a customer has been across multiple channels.

What is wrong with the last-click model?

Taking a step back to look at last-click models, the most commonly used manner to attribute value to online channels, we can see the added value of multi-touch attribution. Last click attribution assigns the total value of a conversion to the last touchpoint in a customer journey.

For example, imagine that a customer has the following customer journey:

Using the last click attribution model, you will assign the total value of the conversion to the affiliate touchpoint.

However, we do know that the customer followed a journey containing multiple touchpoints (SEA, organic and direct) before finally clicking on the affiliate link and actually buying something on your website. Last click completely ignores the other touchpoints in the journey.

Using multi-touch attribution, you can capture the value of all the other touchpoints in the journey, and learn how much each touchpoint eventually contributed to the conversion.

In the table, you can see the difference between last click and multi-touch attribution:

How do you measure the value of each touchpoint using multi-touch attribution?

This leaves us with the question: how is the value of each touchpoint in the customer journey measured? We collect every single occurring customer journey and calculate the conversion rate per path. For example, one path consists of the following touchpoints:

multi-touch attribution journey oneThe conversion rate of 12% means, that if this path with these particular touchpoints in this particular order came across our dataset one hundred times, twelve of those paths led to a conversion and 88 didn’t.

Since we have a very rich dataset consisting of many paths, we can also find the paths which differ from our first path by just one touchpoint:

multi-touch attribution journey twoThis path without the banner view has a conversion rate of 10%. In this example, we can see that a banner view increases the conversion rate by 2%.

All the paths in our rich dataset are compared and analysed in this manner, and we are then able to measure the impact of every online channel. Imagine the huge amounts of information hidden in these paths that last click attribution ignores.

multi-touch attribution

Eventually, you can attribute the value of each conversion to the specific campaign on that day the click took place.

The value of multi-touch attribution

Multi-touch attribution is the only method that assigns to each touchpoint the value it actually contributed to the conversion. This allows you to analyse the impact of your channels and allocate your media budget effectively.

Geertje Keulers

Geertje Keulers

Data Scientist