BBE-LSWCM: A Bootstrapped Ensemble of Long and Short Window Clickstream Models

TL;DR

This paper introduces BBE-LSWCM, an ensemble clickstream modeling framework combining long-term and short-term user data for real-time customer event prediction, demonstrating superior performance and live deployment results.

Contribution

The paper presents a novel low-latency ensemble architecture that effectively integrates long and short window user behavior data for improved real-time predictions.

Findings

Superior performance over baseline models in subscription cancellation prediction

Effective detection of user intended tasks in real-time

Successful deployment and positive online experiment results

Abstract

We consider the problem of developing a clickstream modeling framework for real-time customer event prediction problems in SaaS products like QBO. We develop a low-latency, cost-effective, and robust ensemble architecture (BBE-LSWCM), which combines both aggregated user behavior data from a longer historical window (e.g., over the last few weeks) as well as user activities over a short window in recent-past (e.g., in the current session). As compared to other baseline approaches, we demonstrate the superior performance of the proposed method for two important real-time event prediction problems: subscription cancellation and intended task detection for QBO subscribers. Finally, we present details of the live deployment and results from online experiments in QBO.