To alleviate this problem and try to reflect this complexity in a more reliable way, multi-channel models emerge. The idea is to stop giving 100% of the conversion to a single impact and start distributing it throughout the entire conversion flow.
We have, for example, models such as 40/20/40. 40% of the conversion is assigned to the first and last impact, because it is understood that it is important to give credit to both the channel that is able to impact the user for the first time and the one that closes the sale. And 20% is assigned to all the channels and impacts that maintain the user's interest between that first impact and the last.
A step forward from the last click model, but still very lbosnia and herzegovina business email database imited. Although the complexity of considering multiple impacts is incorporated, the distribution of these is closed and decided a priori. The distribution to be used is determined more by intuition or business objectives than by real data.
Data driven models
Data-driven models arise from the need to work from real data. Instead of using a distribution already decided a priori, an algorithm is applied and it is this algorithm that assigns the value of each conversion. The most popular data-driven models are based on Shapley or Markov, both based on game theory.