Blogs

Voice of the client: Argentina’s Banco Macro makes gains with AI

Creative and Editorial Lead, Data and AI, IBM

As banks worldwide continue to increase their digital offerings over the next few years, industry leaders are looking to increase their AI capabilities and apply them to the overall digital experience. Banco Macro is set on outpacing competitors and has embraced the challenge to climb the AI ladder with the help of the IBM Data Science and AI Elite team.

Banco Macro is based in the South American country of Argentina, a place where a  bumpy economy and socioeconomic divides provide additional challenges and opportunities for the bank’s AI aspirations.

You won’t any longer see wheelbarrows full of cash that were commonly associated with Argentina’s period of hyperinflation. But still, instances of financial flux can influence how much clients are saving, borrowing and spending. That’s kept Banco Macro motivated to find new ways to drive more efficiencies, from optimizing cash flow management in branches and ATMs to preventing liabilities such as fraud— all through applying AI to its customer data.

Nicolas Martins leads data and AI projects at Banco Macro. He joined forces with the IBM Data Science and AI Elite team to help develop two pilots. Banco Macro took advantage of the expertise of the IBM Data Science and AI Elite team to tap into Watson Studio to develop, deploy and maintain their machine learning models and integrate the results into their daily workflow.

Senior analysts from Banco Macro were paired with data scientists from the IBM Data Science and AI Elite Team, who undertook three, two-week sprints in Buenos Aires to develop two use cases.

The first use case proved that machine learning models could increase the likelihood of clients accepting the next best action. The second model was able to detect activity on accounts held by recently deceased individuals. Both pilots proved successful – and are paving the way for two additional use cases, risk analysis for scoring and debt liability reduction.

“With these models, we could theoretically predict that we will be able to increase by three times the response rates of those campaigns. The idea is to optimize who we are going to contact and what are we going to offer to them,” said Martins. Customers who were identified by the models were also 25 percent more likely to accept the next best action.

More complex algorithms required more compute power. So while Banco Macro was running on IBM SPSS Modeler, the bank upgraded to Watson Studio and Power AI. The bank’s data scientists were especially fond of all the coding languages that Watson Studio supported, especially Python. The platform opened up other opportunities to optimize through additional library access.

“We find the platform very friendly. It’s very collaborative. It helps you track what you are doing and the results,” he said. “You can set the comments and it works with the most popular coding languages. It has a lot of ways to integrate with current and future data that we might need to include in future models.” 

Based on the success of the models, Martins says Banco Macro will be using the models in the short term to serve other business areas—not only in commercial areas but also for regulatory and compliance issues.

Find out how the IBM Data Science and AI Elite team can help you build out your AI strategy.

Accelerate your journey to AI with a prescriptive approach. Visit ibm.com/data-ai to learn how to modernize, collect, organize, analyze and infuse all your data.

IBM Data and AI Elite team members Ainesh Pandey, Anna Hazard, Morten Vester Pedersen and Robert Uleman
Topics:
Data Science