Marketing attribution is surfacing in marketing discussions more and more. Marketing attribution arrives with the promise of solving an old problem. How to know which half of the marketing budget is wasted. It was famously John Wanamaker who coined the phrase first.
However, marketing attribution is complicated. Attribution is not complicated because of the algorithms themselves. There are two good and well known algorithms for multitouch attribution. They are the Shapley value and the Markov model. More on them here: https://www.windsor.ai/shapley-value-vs-markov-model-in-marketing-attribution/
Advanced multitouch attribution has recently become possible for two reasons. There is more data being collected but also the processing capabilities have increased. Here is a bit more background https://en.wikipedia.org/wiki/Attribution_(marketing) Next I will go through some steps on how to benefit from all of this as a marketer.
Understand marketing attribution and the shortcomings on a high level
Be aware that Facebook and Google prefer to attribute to themselves. They also use very different attribution settings by default. How the data has been collected also plays an important role in the reliability of the attribution modelling. Acknowledging these things are important to trust the modelling going forward.
Use multitouch attribution aka. data driven attribution models
Using multitouch attribution abstracts away the need to understand the customer journeys in detail. A multitouch attribution model gives credit to the touch-points during the whole customer journey so there is no need to analyse the different journeys. It is always good though to double-check and verify.
There might be special reasons to using eg. weighted attribution models focusing for example more on customer acquisition if that is the company number one priority.
Connect the attributed conversions to the spend
This is important to make it actionable. ATTRIBUTION MODELING IN ITSELF IS NOT ACTIONABLE. Attribution modelling only gives you the contributing value of a touchpoint. To make it more actionable it has to be compared how much was paid for that value. That provides a data driven ROAS or CPA which starts to be actionable.
Trust the numbers and act on them
Doing attribution can be expensive and it can be worthwhile to the large advertisers who want to squeeze the most out of their marketing spend.
Once there starts to be numbers, ideally in the form of recommendations on what to increase and decrease its important to start acting on. This is the only way to realise the value. Maybe not necessarily making big changes immediately but starting to experiment and seeing the effect or the lack of it:)
This then paves the way for more data driven decisions and experiments that can lead to growth.
