The project proposed in this thesis aims to bring analyses and tools to support ad
vertising marketers in their daily decision-making process for advertising campaigns
in a media mix context.
Online advertising allows advertisers to reach users on multiple channels, who of
ten get in touch with multiple advertisements. On the web, the users’ interactions
can be tracked and an attribution model can achieve the issue of measuring the
effectiveness of ads on each user assigning a percentage of the conversion value to
each ad in the customer journey. The results of an attribution model represent
meaningful information to evaluate and optimize campaigns. This thesis remarks
the lacks of the rule-based attribution model and treats chronologically some data
driven multi-touch attribution models proposed by academic researchers, who be
lieved that models should not be based on assumption but data. In according with
them, the analyses proposed in this work are based on data calculated by a data
driven multi-touch attribution model.
This project faces the budget allocation problem proposing an analysis using model
driven approach and marginal analysis to maximize the expected revenue. KPIs
for marketing campaigns and e-commerce performances have been calculated from
different sources and reported in dashboards and reports for further analysis by
marketers. The project is part of six months internship at Haensel AMS - Advanced
Mathematical Solution GmbH located in Berlin.