Keep customers before they leave.
ChurnRadar combines online and offline retail data in a hybrid deep learning model to flag customers at risk of leaving, so marketing teams can act in time.
The problem
Losing customers is expensive, and usually spotted too late.
In omni-channel retail, customers interact through stores, websites and apps. Data from all these channels makes churn hard to predict with traditional methods.
churn prediction accuracy of our hybrid CNN-LSTM model.
ROC-AUC, showing strong separation of churners and loyal customers.
F1-score, balancing precision (89.7%) and recall (88.5%).
Source: the founding research paper (ICBATS, 2025).
How it works
How it works
A clear path from data to decision.
Unify
Online and offline customer data are combined.
Engineer
Behavioural features are extracted from every channel.
Predict
A CNN-LSTM model scores each customer's churn risk.
Retain
Teams launch targeted offers for high-risk customers.
Product
What makes it different.
Every feature is grounded in the founding research.
Hybrid CNN-LSTM
Captures patterns and sequences in customer behaviour.
Omni-channel
Built for stores, e-commerce and apps together.
Proven accuracy
91.2% accuracy and 0.93 ROC-AUC.
Scalable
Built with TensorFlow and Keras for real-world retail.
Who it's for
Who it's for.
Retail chains
Reduce churn across stores and online.
E-commerce brands
Target retention spend where it counts.
Loyalty programmes
Spot disengaging members early.
The research
Built on peer-reviewed research.
- Hybrid deep learning framework blending CNN and LSTM networks.
- Extensive data processing and feature engineering on online and offline sources.
- Accuracy 91.2%, precision 89.7%, recall 88.5%, F1 89.1%, ROC-AUC 0.93.
- Outperforms baseline models and scales to real-world retail.
Ahmed, T. A., et al. (2025). Deep Learning Models for Predicting Customer Churn in Omni-Channel Retail Environments. ICBATS 2025, IEEE.
Team
The team behind it.
Founded in 2025 by a student team from two faculties.
Faculty of Administrative & Financial Sciences
Marketing analytics, retail strategy and customer insight.
Faculty of Science (Computer Science)
Deep learning models and data engineering.
FAQ
Questions we hear often.
What data do we need?
Transaction, visit and engagement data from your channels.
Can it integrate with our CRM?
Yes. Risk scores can be exported to CRM and marketing tools.
Is this a finished commercial product?
It is a research-based student startup moving from validated prototype toward pilots with real users.
Can we partner or run a pilot?
Yes. We are looking for organisations willing to test the solution in real conditions and share feedback.
Bring ChurnRadar to your organisation.
We are onboarding pilot partners now. Read the research and get in touch.