Institutional Repository of Vilnius University
Marketingo kanalų efektyvumo modeliavimas /
Abstract
dc:descriptionTo optimize marketing spend and achieve maximum profit, companies need to evaluate marketing effectiveness. Often, costs are allocated based on intuition or short-term indicators, creating a need to estimate return on investment using historical data. Different methods can be used to model revenue generated by various channels and to forecast revenue: linear regression, XGBoost, Random Forest machine learning algorithms, time series models, or the more advanced marketing mix model. In this study, all of these models are tested, taking into account both revenue growth trends and weekly seasonality. It was observed that linear regression is unsuitable neither for forecasting, nor for interpreting channel effectiveness. Machine learning algorithms and time series models can forecast revenue but provide misleading estimates of channel significance when compared with business insights. In contrast, the marketing mix model with a delayed carryover effect function performs best for interpretation and is therefore the most appropriate model for this task. According to this model, email campaigns and Bing advertising have the highest return on investment, even though the least amount of budget is allocated to these channels. Influencer marketing, while third in terms of revenue generated, has the lowest return on investment. The best forecasting accuracy was achieved with the Random Forest algorithm, but its mean absolute percentage error is only 0,11 % less than the error of the marketing mix model selected as the best overall.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2025
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Trasykytė, Augustė Ona,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD239029102&prefLang=en_US