ATM Disloyalty Prediction Model Case

Data Science SWAT team: ATM disloyalty prediction model for a banking group

About This Project

Our client, a leading global banking group was seeking to understand the ATM disloyalty patterns in order to reduce the annual fees paid to third-party institutions and improve customer services.

Supported by our advanced analytics team, Altair gathered insights and provided the analytical muscle to help the client develop a more precise and robust prediction model which results in significant savings in operating expenses.

Challenge

In 2016, the banking group paid extremely high fees to third-party ATM networks. They understood that a small number of customers accounted for a disproportionate amount of usage and the existing disloyalty prediction system had much room for improvement. Leadership believe that an advanced prediction model would improve their customer service level and optimize operating expenses.

The key challenge Altair faced was how to design Machine Learning algorithms effective enough to obtain faster and more accurate insights than traditional methods.

Altair’s Solution

We assembled a customized dataset from multiple sources, performed data-driven analyses, and developed a test and learn capability for a robust prediction model.

 

Approach

The methodological approach focused on features and transactions at the customer and ATM level. Various analytical techniques were utilized to gain a wide understanding of the ATM customer behavior in order to generate a prediction model.

 

  • Customer segmentation:

The featuring engineering process has pointed out various findings that are key to understanding ATM behavioral patterns of customers. In turn, a segmentation has been performed to find out if different patterns of expenditure are related with ATM disloyalty. The segmentation analysis has provided 8 groups of customers with different characteristics of usage.

 

  • Data collection:

Other data points such as ATM locations, average balance account, ATM usage, online banking activities and UX at third party ATMs were incorporated into the analyses and several correlation patterns identified.

 

The Prediction Model

This framework allows for a more robust prediction model to determine disloyal behavior by reducing data noise, and label customer for specific actions. Based on the customer grouping, Altair also recommended actionable measures to optimize ATM footprint and minimize costs.

Outcome

Machine Learning techniques showed many advantages:

  • Data driven insights: more reliant results than subjective analysis
  • Speed: after clearly defined the objective, criteria and framework, all results have been obtained in three weeks by a team of two consultants, moving key resources to understand the strategic alignment and recommend practical actions.
  • Internal capabilities: tools used are open-source and available in current architecture
  • Transparent solutions: all code and methodologies applied are available for the client to use in other business cases

In addition, we believe that an organization needs to embed analytical tools and insights into its day-to-day operations to deliver the most value. To that end, we reviewed and defined the organizational requirements needed to support a robust big data analytics capability in general, setting the company on course toward their goal.

Country

United Kingdom

Date
Category
Analytics, Banking and Financial Services
Tags
Analytics, ATM, Banking & Financial Services, Disloyalty Prediction

We use cookies to give you the best online experience. By agreeing you accept the use of cookies in accordance with our cookie policy.

Privacy Settings saved!
Privacy Settings

When you visit any web site, it may store or retrieve information on your browser, mostly in the form of cookies. Control your personal Cookie Services here.

These cookies are necessary for the website to function and cannot be switched off in our systems.

In order to use this website we use the following technically required cookies.
  • wordpress_gdpr_allowed_services
  • wordpress_gdpr_cookies_allowed
  • wordpress_gdpr_cookies_declined
  • pll_language

Decline all Services
Accept all Services