The question of conversion attribution arises when trying to identify the advertising campaign that led to a conversion on an e-commerce or commercial website. This issue is of paramount importance because it allows for the evaluation of the effectiveness of various digital marketing channels, and sometimes even offline channels. Conversion attribution becomes complex when several advertising or marketing interactions overlap for the same individual, or when the purchase process extends over several successive visits.

In the field of web analytics, conversion is often attributed by default to the last click on an advertisement, even if that click wasn’t necessarily the deciding factor. Consider the example of sponsored links: a prospect might initially arrive on a site via a generic keyword, discover the offer, and then return later by typing a query for the brand they remembered. In this case, the conversion risks being mistakenly attributed to organic search or brand keywords, when the crucial interaction likely occurred during the initial visit via the generic keyword.

An interesting approach to overcome the limitations of last-click attribution is to consider partial attribution, distributing the conversion across different marketing campaigns. Models are then developed for this purpose. Attribution can apply to all marketing and advertising channels, but it is primarily measured for digital channels, as their interactions and visits can be tracked. For influential players who pay traffic sources based on performance, developing an attribution model can change the terms of communication.

Conversion attribution models offer the possibility of conducting a more in-depth analysis than simple last-click attribution, often requiring the implementation of a specific tracking system within web analytics tools. To delve deeper into this topic, it is