Re-activation of 17% churned bank clients with AI use in banking

Innowise has employed AI and ML algorithms to predict customer churn and develop targeted retention strategies for a retail bank.

Customer

Industry

Banking

Region

MENA

Client since

2021

Our client, a prominent retail bank, holds a strong position within the MENA (Middle East and North Africa) region. With a significant presence and influence in the local market, this bank has established itself as a trusted financial institution catering to individuals.

Detailed information about the client cannot be disclosed under the provisions of the NDA.

Challenge : Decreasing customer churn rates through Artificial Intelligence in banking industry

Our client was undergoing a global digital transformation. Traditional customer retention methods proved ineffective, prompting the bank to seek a personalized approach. One of the strategies the bank adopted as part of their digitization efforts was the implementation of targeted advertising campaigns within automated marketing aimed at specific user groups, with the objective of retaining customers using AI and predictive analytics.

However, the bank lacked a unified system capable of gathering user data, identifying behavioral patterns indicative of potential customer churn, and analyzing it comprehensively. Innowise was tasked with developing such a system, leveraging ML-models to detect customer attrition based on behavior patterns.

Solution: Analyzing and predicting customer behavior with AI-driven predictive banking software

Innowise has developed an AI-driven predictive banking software solution to analyze individual churn rates to help our client implement highly targeted retention strategies. This solution optimizes resources by enabling focused efforts on high-risk customers, ensuring maximum impact in retaining valuable clientele.

Enhanced customer data analytics

The analytical system operates on the back-end, seamlessly integrating with the bank’s data warehouse to collect customer data. We used the Spark engine to develop an efficient system that provides ML pipelines, data preprocessing, model training & evaluating, anomaly detection, and data scaling. The system uses a multi-faceted approach to analyze various aspects of customer information, including transaction history, customer complaints, demographics, etc.

By analyzing customer data through natural language processing (NLP), the system captures the sentiment and customer feedback. This functionality empowers the bank to proactively address customer issues and concerns before they escalate, thereby reinforcing customer loyalty.

One of the primary challenges faced was an imbalanced dataset, where only a small fraction of customers had churned. Therefore, it was crucial to ensure that the selected model accurately predicted this minority class with higher precision. The presence of such an imbalance could potentially lead to biased model performance. To address this issue, we conducted extensive research into existing solutions specifically designed for handling imbalanced data samples to mitigate any potential bias and improve the overall performance and accuracy of the model.

To evaluate the models’ precision, recall, and F-measure, we helped our client identify custom model metrics and acceptance criteria for each specific customer case in accordance with the business value. However, we have focused on F1-score as it illustrates a balance between precision and recall.

Our final solution encompassed a diverse range of machine learning algorithms, incorporating both classical boosting models and modern self-supervised techniques. By leveraging boosting models, we effectively addressed the original churn problem with a high degree of accuracy, ensuring precise predictions for customer churn.

Churn risk evaluation

The system’s AI algorithm provides ongoing analysis of user metrics and determines their churn classification group. This information is then incorporated into the bank’s marketing system, allowing analysts to present it in a clustered view. This facilitates efficient filtering and segmentation based on specific user categories.

The implementation of AI predictive analytics and intelligent segmentation empowers the bank to develop targeted campaigns and highly personalized offers. By tailoring individual cash back options, exclusive bank promotions, and personalized discounts, the bank can effectively cater to the unique requirements and needs of each customer. The system also displays churn risk percentage for each customer on CMS cards, enabling bank staff to gain valuable insights during their interactions and implement retention strategies to retain customers.

Technologies

Front-end

React, Redux, Redux-Thunk, React-hook-form, SASS, Axios, Storybook, Jest, Cypress

Back-end

Java, Spring (Boot, Data, MVC, Security), REST, SOAP, Liquibase, Maven, JUnit, WSDL, Mockito, Hibernate

Data engineering

Apache Hive, Apache Spark, PySpark, Apache Airflow, Redis Feasts

Cloud

Oracle

DevOps

Kubernetes (k8s), Docker, Docker Compose, Jenkins

Machine learning

Apache Spark MLLib, Scikit-learn, LightGBM, XGBoost, Hyperopt, Numpy, Pandas, SciPy

MLOps

DVC, MLFlow, Comet

Process

Innowise offers a comprehensive suite of AI solutions for banks. These solutions encompass multiple essential phases, ensuring a robust implementation and seamless integration.

Problem framework

Through extensive collaboration and requirements gathering sessions with our customers, we established a clear problem framework. This involved engaging key stakeholders and bank experts to identify the specific challenges associated with bank digitalization.

Data acquisition and exploratory data analysis

After defining the problem framework, we focused on dealing with a large amount of customer data. Our first step was to perform exploratory data analysis. This helped us validate statistical hypotheses and laid the foundation for feature engineering. For instance, we observed that the churn rate among female customers was higher than that of male customers, and neither the product nor the salary significantly affected churn likelihood. Feature engineering played a crucial role in updating and refining the features at this stage. We evaluated various machine learning algorithms, including Gradient Boosting Decision Trees (GBDT), Naïve Bayes, and Classificational Neural Networks. Through careful assessment, we determined that the GBDT method yielded the highest metrics for the original task.

Model development

The system was continuously evaluated, refined, and tested during the model development phase. We fine-tuned the models using multiple iterations and validation techniques to achieve the highest performance in AI predictive analytics.

Model deployment

As part of the deployment step, we integrated the developed model into the bank's system, incorporating it as part of the users' key metrics. This process involved close communication between Innowise teams and the bank's IT department to ensure seamless integration. By following this structured approach, Innowise delivered an effective AI-driven predictive banking solution, addressing our client’s specific challenges and enabling them to make data-driven decisions for improved performance and customer satisfaction.

Team

Project Manager

Data Scientists

Data Engineers

Back-End Engineers

Front-End Engineers

QA Specialist

Results: Increased customer lifetime value and re-activation of churned customers with AI in banking and finance

The implementation of AI in banking and finance delivered remarkable results for our client. The bank experienced a significant increase in customer lifetime value, unlocking new revenue opportunities and fostering long-term relationships with its valuable clientele by deploying targeted retention strategies.

One of the most noteworthy achievements of the system was the substantial reduction in customer churn rates and successful re-activation of 17% of inactive customers. By identifying customers who are likely to leave the bank’s services in advance, the system enabled the bank to proactively address their concerns and provide personalized retention initiatives based on insights provided by the AI-driven predictive banking software solution. Through targeted communication and tailored offers, the bank successfully retained a larger number of customers, ensuring their continued loyalty and contributing to the overall growth of the institution.

Project duration

November 2021 - December 2022

Related cases

Contact us!

Book a call or fill out the form below and we’ll get back to you once we’ve processed your request.

What happens next?

Having received and processed your request, we will get back to you shortly to detail your project needs and sign an NDA to ensure the confidentiality of information.

After examining requirements, our analysts and developers devise a project proposal with the scope of works, team size, time, and cost estimates.

We arrange a meeting with you to discuss the offer and come to an agreement.

We sign a contract and start working on your project as quickly as possible.