Why Attribution model definition matters? | Publications

As marketeers, one of the most important task consists of optimizing your advertising investments. And that is even more true when it comes to digital advertising. Indeed, you always want to know which marketing channel brought you the most revenue, helped you improve your retention or contributed the most to your brand awareness. As digital behaviour changes, so does digital advertising and the need for more accurate measuring or attribution model.

A more complex digital world

Your customers spend more time online, interacting across an expanding universe of multiple channels, touchpoints, and data sources. Their online universe becomes more complex; so do your marketing actions. To make accurate budget allocation decisions, you need as marketers to take into account these multi-channel and multi-touch customer cycles. And that is where technology can help you; but you need to be data-driven and care about attribution.

Attribution modelling

Attribution model is a hot topic in our industry but is often misunderstood or confused with “big data”. By Attribution, we generally mean realizing deep analyses of past campaigns over time to consider how to better allocate investments for future media planning purposes.

So why should you consider attribution?

1. To better evaluate past actions

2. To better predict results

3. To better decide which actions to take

In the past, attribution models have been based on the “last click model” or “last touch” attribution. In other words, it means that a conversion was granted to the last marketing channel with which your customer interacted, regardless of how many other channels played a role before. With this type of model, we were confronted to the following sort of question: “if most of my conversions came from branded SEO and direct traffic, should I keep investing in other marketing channels?” “Should I stop my emailing acquisition campaign?” or “Are my bannering campaign really useful?”


With the rise of big data, the volume and the complexity of the data, you need more advanced analytics skills and solutions. The multiplication of new data sources requires new analytics models that go beyond the classic “last click” attribution. Marketeers need to understand the impact that their entire flow of media has on transactions or generated leads. That is why we now talk about “Attribution model”. But how should you proceed to define an attribution model?

1. Define clear business objectives (e.g. → Raise of % in Sales)

2. Choose the right KPI to measure results ---> Categorize your metrics based on objectives

3. Be ROI-oriented (e.g.→ How much can you afford per channel to attract new customers?)

4. Try not just to integrate digital data but also other data (like CRM data) that willl help you to have a complete view on your customers

5. Facilitate automation process to focus on strategy and concrete results → Rely on technology and tools available

Leverage technology

As the amount of data to analyze might quickly become gigantic, you need to rely on technology. Indeed, the complexity of those multiple touchpoints is too challenging to be analyzed and managed manually. It is why analytics tools now integrate multi-channel features and starded to integrate attribution tools. For instance Google Analytics provides its own “Attribution Modeling Tool”. Google Analytics engineers have invested a lot of time to polish the tool in 2013 and with the arrival of Universal Analytics, you might be able to include not only channels but also devices focused on users.

That being said, you should always keep in mind that no allocation model will answer perfectly to all your questions. Indeed, there is not a single model, but several models that will help you to better understand the interaction between your different marketing channels. The final goal is to provide you with enough business insights to make smarter decisions.

Specialists are there to help you

As you can see, defining attribution model might not look so easy. It requires a certain level of skills. It is why you need the right people and where specialists might help you. Once these are defined, you can concentrate on implementing these across your entire digital advertising measurement and assessment in order to gain valuable actionable data from it.

This practical approach will be explained in a later article where we will explain the different models available and where, with a concrete case, we will demonstrate the power of attribution modelling.

written by Stéphane Juricic


Get in touch

Semetis | 44 Rue des Palais, 1030 Brussels - Belgium

+32 (0)2 211 34 50


Connect with us