Aaltodoc publication archive (Aalto University institutional repository)
School of Business | Department of Information and Service Economy | MSc program in Information and Service Management | 2014
Thesis number: 13986
Household consumption of Sanoma's media product combinations: Using association analysis to drive sales
|Title:||Household consumption of Sanoma's media product combinations: Using association analysis to drive sales|
|Year:||2014 Language: eng|
|Department:||Department of Information and Service Economy|
|Academic subject:||MSc program in Information and Service Management|
|Index terms:||palvelut; service; viestintä; communication; kustannustoimi; publishing; media; media; tuotteet; products; myynti; sales; kulutus; consumption; kotitaloudet; households|
|Key terms:||association analysis; media industry; recommender system; targeted marketing; cross-selling|
OBJECTIVES OF THE STUDY:
The main objective of this study was to use household subscription data in order to increase sales by cross-selling to existing customers. Research questions related to this objective was first to determine which product combinations were over- and under-subscribed, and then to determine how to use association analysis results in order to increase cross-selling.
A secondary objective was methodological in nature, and was to determine the necessity of segmenting household data according to key demographics before analysis. Theoretically the adverse effect of not segmenting is well-understood; however it was of interest the extent to which analysis of unsegmented data would give false confidence to poor decision making. Research questions related to this objective were identifying which household segments were similar and what the effects were of combining similar segments.
ACADEMIC BACKGROUND AND METHODOLOGY:
Association analysis was performed on household data, which had been segmented according to Sanoma's geographical areas and age-groups of interest. Student's paired t-test was performed on each pair of segments' association analysis lift results in order to identify similar segments. The most similar segments were then combined and tested again.
Ranking association analysis results by lift quickly identified over- and under-represented product combinations. Other scoring measures were combined with two cross-selling strategies (recommender system and targeted marketing) in order to develop analysis models that could select optimal cross-selling opportunities.
FINDINGS AND CONCLUSIONS:
Statistical testing on combinations of similar segments agreed with theory that segmentation of data is important preceding analysis. Empirical data provided numerous examples of suboptimal decision-making that arose by using unsegmented or combination analysis data.
The recommender system model selected optimal cross-selling opportunities that could be acted upon immediately. By contrast the targeted marketing model needs further refinement by consideration of non-recurring and recurring marketing campaign costs, but succeeded in showing that such an approach was possible with this data.
Master's theses are stored at Learning Centre in Otaniemi.