Attribution models today: from classic to data-driven… and beyond
Posted: Sun Jan 19, 2025 3:20 am
By Alex Masip
ANDThe topic of attribution models has always been in the news and is still as hot as ever. In this article, we will briefly review traditional models, multi-channel models and focus on data-driven models and the role of machine learning in attribution.
Traditional models
The attribution model par excellence in the digital world is the last click model. It may sound old-fashioned to some of you, but the vast majority of you are probably still working with it or have done so until very recently. It is still, by far, the most widely used model even today.
It's not difficult to explain: the total sale is attributed to bosnia and herzegovina business email database the channel/campaign that brought the last visit that ended in conversion. Its limitations are not difficult to explain either: the number of touch points that a user makes before buying has long been very complex, and it is becoming more so. We only have to think about our usual process when we are interested in a product or service: we go to a search engine, we look for a price, we read reviews, we go to a potential supplier or store, then to another, we return to the first one, etc. etc. During this entire process we are impacted by advertising in one way or another, a process that a model like the last click ignores completely.
ANDThe topic of attribution models has always been in the news and is still as hot as ever. In this article, we will briefly review traditional models, multi-channel models and focus on data-driven models and the role of machine learning in attribution.
Traditional models
The attribution model par excellence in the digital world is the last click model. It may sound old-fashioned to some of you, but the vast majority of you are probably still working with it or have done so until very recently. It is still, by far, the most widely used model even today.
It's not difficult to explain: the total sale is attributed to bosnia and herzegovina business email database the channel/campaign that brought the last visit that ended in conversion. Its limitations are not difficult to explain either: the number of touch points that a user makes before buying has long been very complex, and it is becoming more so. We only have to think about our usual process when we are interested in a product or service: we go to a search engine, we look for a price, we read reviews, we go to a potential supplier or store, then to another, we return to the first one, etc. etc. During this entire process we are impacted by advertising in one way or another, a process that a model like the last click ignores completely.