Skip to main content

Regional Bank Identifies over $100M in At-Risk Deposits through Machine Learning

 

Challenge

According to the CDC/NCHS National Vital Statistics, banks lose about 10 percent of their account deposits due to customers closing their account. While half of that loss is attributed to factors out of the bank's control, such as death, divorce, or displacement, the other half of loss is attributed to the customer's dissatisfaction with the bank's fees, rates, products, or lack of convenience. For a bank that handles nearly $700 million in deposits annually, losing $70 million due to customer loss will seriously curb their ability to grow. 

A regional bank wanted to reduce customer attrition and learn how to predict which customers were likely to close their checking accounts within 90 days. This would give them the opportunity to take action to retain the customer.

Their traditional analytics tools didn't have the capacity to uncover patterns, and human analysis certainly won't work to analyze billions of data points in real time. The bank knew they could use machine learning to analyze their existing data to reveal trends and insights and predict future behaviors and outcomes, but they hadn't used it before.

For expert help with this project, they turned to New Era Technology. Our team had worked with them several times in the past and knew how to help them get the insight they needed. 

Solution

Most machine learning projects take several months before seeing results. However, our client wanted to see a proof of concept quickly to determine how this would work and whether it would help them fulfill their goals. With a New Era Technology Machine Learning Jumpstart, we could help them meet this objective.

Machine Learning Jumpstart

During our Machine Learning Jumpstart, we assess the key business problem, assess the data, and build a machine learning model that delivers answers fast. To do this, we follow a specific, four-step plan that includes: 

  • Use case identification. Prior to beginning our technical work, we explored several use cases in a workshop with the bank’s business and technical stakeholders. Our team rated each potential use cases on different criteria, including the complexity, availability, and value impact of the data. We agreed upon the deposit attrition proof of concept, deciding it would drive maximum predictive value with minimal risk.  
  • Data processing. We inventoried and sourced the existing data, then cleaned and loaded it in the target on-premises environment where the models would be developed. We provided the option of loading the models in the cloud to enable additional ML models and more complex computations.
  • Machine learning model development. Within three weeks, we began engineering the machine learning models, choosing the subset of data most relevant to the question, “Which checking accounts are likely to close in the next 90 days?” We selected the machine learning algorithms, then trained and tuned the model. 
  • Model insights integration. With a model in place, we met with with the bank’s stakeholders to present metrics to measure the model’s success, focusing on KPIs. The model returned valuable insight into the accounts at high risk of closing which allowed the bank an opportunity to refer customers to the bank’s retention team
This end-to-end process generates daily predictions using real-time data in less than an hour with the bank's existing infrastructure.
 

Educating the Team

While providing the machine learning model was our highest priority, we also wanted to help educate the bank's team on key elements of machine learning, such as the process for training models and the process for generating predictions. This would occur organically by working side by side with their team.

Optimizing Results

The bank wanted to use machine learning to predict customer attrition more effectively than how they could do so through traditional analytics. They also wanted to understand the key metrics for evaluating machine learning models. 

In the secondary phase, we could optimize the model to capture model efficacy. This will allow the bank to optimize and expand the use case or use the model as a template that can be modified for one of the other use cases. 

The entire proof of concept took eight weeks, and the bank is now in possession of machine learning models that can be implemented in marketing campaigns in the next phase.

Outcomes

Using machine learning to predict consumer behavior has significantly improved the bank's ability to predict attrition. Currently, the models are set to identify customers who will leave within 90 days at 83% precision. This allows them to take action needed to retain at-risk customers and continue growing their organization. 

With machine learning models in place, the bank’s immediate goal is to develop a deeper understanding of the technology. In the next phase, they will automate the entire marketing pipeline and feed predictions to Salesforce or another enterprise platform.

Currently, the models are set to identify customers who will leave within 90 days at 83% precision. In addition, the bank’s team now understands all the key elements of machine learning and has a meaningful list of indicators of attrition.

Improved data security
Jumpstart Defined Attributes of Accounts Likely to Close
Improved data security
ML Models Predict Customers Likely to Leave with 83% Precision
Improved data security
Enabled Opportunities to Increase Capital

Ready To Talk?

Let us know how we can help you out, and one of our experts will be in touch right away.