Developing a data-driven predictive lead scoring solution for the B2B on-trade market of the beer industry
Keywords
Loading...
Authors
Issue Date
2025-04-24
Language
en
Document type
Journal Title
Journal ISSN
Volume Title
Publisher
Title
ISSN
Volume
Issue
Startpage
Endpage
DOI
Abstract
This thesis explores, implements and evaluates a data-driven approach to predictive lead
scoring in the B2B on-trade market of the beer industry. The aim of the research is to
leverage machine learning models to enhance traditional sales approaches, enabling more
efficient and effective customer acquisition strategies. The research was conducted using
data from a diverse set of internal and external sources, provided by a Dutch beer brew
ery. The proposed approach involves two separate machine learning models: a prospect
scoring model to estimate the probability of lead conversion, and a volume projection
model to predict the expected sales volume of an outlet. Separate models are devel
oped for the contracted and non-contracted sales divisions, which have very different
market dynamics. The prospect scoring model demonstrates strong predictive perfor
mance. with Random Forest emerging as best performing model. The volume projection
model shows poor performance, likely due to limitations in data quality and availabil
ity. Qualitative feedback from sales representatives acknowledges the high potential of
data-driven lead lists, but also reveals the importance of contextual factors, which are
often not captured in the data. The findings suggest that predictive lead scoring can
effectively support customer acquisition in the non-contracted market, where the sales
process is less complex. However, the contracted market likely requires a more adaptable
approach that incorporates regional sentiment and other contextual factors that have a
strong influence on lead relevance.
Description
Citation
Supervisor
Faculty
Faculteit der Sociale Wetenschappen