2999. Trial-to-Paid Conversion Prediction Based on a Hybrid and Explainable Tree-Based Sequential Usage Pattern Mining Approach
PENDINGInvited abstract in session Removed abstracts, stream Removed, cancelled and rejected abstracts.
Authors (first author is the speaker)
| 1. | Koen W. De Bock
|
| Department of Marketing, Audencia Business School |
Abstract
Predictive approaches to lead scoring can help enhance the effectiveness and profitability of customer acquisition efforts. This study introduces a novel hybrid and explainable methodology that integrates sequential usage pattern mining with binary classification to identify frequent conversion paths and predict trial user behavior conditional on path traversal. At the core of the proposed approach is a hybrid decision tree-based structure that effectively identifies service usage sequences and their impact on conversion likelihood. We validate our method using a real-world dataset from a leading B2B SaaS platform provider, benchmarking its performance against conventional approaches for sequential pattern mining, combined with various classifiers. The results demonstrate competitive performance and strong interpretability which not only stems from insights in the features that drive conversion, but also how different usage patterns determine conversion. Our findings highlight the value of hybridizing sequential usage pattern mining and classification in an explainable model.
Keywords
- Analytics and Data Science
- Marketing
- Decision Support Systems
Status: accepted
Back to the list of papers