Aaltodoc publication archive (Aalto University institutional repository)
School of Business | Department of Information and Service Economy | MSc program in Information and Service Management | 2016
Thesis number: 14422
Sales response modeling in paid, owned and earned media ecosystem
|Title:||Sales response modeling in paid, owned and earned media ecosystem|
|Year:||2016 Language: eng|
|Department:||Department of Information and Service Economy|
|Academic subject:||MSc program in Information and Service Management|
|Index terms:||tietotalous; knowledge economy; markkinointi; marketing; mallit; models; myynti; sales|
|Key terms:||marketing mix; marketing mix modeling; sales response modeling; regression; structural equation modeling; SEM; two-stage least squares; paid, owned and earned media; POE; POE ecosystem|
BACKGROUND AND OBJECTIVES: The focus of this thesis is on sales response modeling methods. Sales response modeling refers to the econometric modeling of sales response to the marketing actions. The goal is to find out the factors that drive sales, separate the amount of sales brought in by media from the short-term baseline sales, and calculate the return on marketing investment (ROMI).
With social and digital media, new types of media channels have emerged. Paid media refers to the traditional way of marketing in channels that the marketer has to pay for. Owned media refers to the content in the channels that the marketer controls itself, such as the company website. Earned media is content generated about the offering of the marketer, but by others. The challenge of modeling is to capture the impact of each media type on sales.
The thesis is a real world application conducted in close collaboration with a company that wants to explore ways to test and extend their current approach of modeling, called nested regression. One issue is how to best model the effects of paid, owned and earned media landscape. A theoretical problem related to endogeneity was found when analyzing the model. Endogeneity problem results in simultaneous equations bias in nested regression.
Possible solutions are to use two-stage least squares and structural equation modeling methods. The purpose of this thesis is to study how changing the modeling method affects the modeling outcomes: variable coefficients and ROMI.
DATA AND METHODS: The data are customer marketing data from three actual cases. The data are modeled using the two-stage least squares and structural equation modeling methods, keeping the variables the same as in the original analysis, and the results are compared.
RESULTS: It is found that the choice of method does not have a practically relevant impact on modeling outcomes. Furthermore, the bias resulting from the endogeneity problem of the nested regression approach is small. Structural equation modeling has potential to be used for modeling POE effects, but it requires further research.
Master's theses are stored at Learning Centre in Otaniemi.