Proactive-Engagement-Analytics-Predicting-Customer-Retention-Interaction-Trends

Predicting customer engagement for SaaS businesses to reduce churn.

SaaS Customer Engagement Forecasting Project

Overview:

This project focuses on predicting periods of low engagement in a SaaS business by forecasting logins per week and session duration per week. The goal is to help customer success teams proactively engage customers and reduce churn.

Problem Statement:

The SaaS industry heavily relies on recurring subscriptions, but customer churn poses a significant threat. By predicting low engagement periods, the customer success team can intervene early and retain customers.

Dataset:

We used a dataset of 520,000 observations spanning 12 months, with the following features:

• LoginsPerWeek
• SessionDurationPerWeek
• FeatureInteractionsPerWeek
• TimeSinceLastLogin
• LifecycleStage

Tools and Techniques:

• Programming Language: R
• Model: ARIMA for time series forecasting
• Evaluation Metrics: MAE, RMSE, Shapiro-Wilk, Ljung-Box
• Visualization: Trends, forecasts, and comparisons between predicted and actual values

Key Findings:

The ARIMA model provided accurate forecasts, with low MAE and RMSE values, indicating the model captured engagement trends effectively. This provides actionable insights for the customer success team.

Files:

• plots/: Contains data visualisations, Forecasted results.
• scripts/: Includes all R scripts for data preparation, modeling, and evaluation.

Conclusion:

This project demonstrates my ability to work with time series data, build predictive models, and generate actionable insights to improve customer retention in a SaaS business.