Case Study: Banco BMG S.A.

CUSTOMER PROFILE

Banco BMG wanted to take a proactive approach to scoring customer relationships that could be headed for legal or regulatory action. The company applied NiCE Nexidia Analytics to flag conversations with these characteristics, then created a special agent team and workflow to try to resolve the difficult situations.

THE BEFORE

A quest for transformative ideas

Banco BMG began working with NiCE Nexidia Analytics and customer experience consultant in 2018 to find opportunities to transform its business and find novel ways to improve services. Speech analytics have been responsible for a number of innovative improvements since then, and the company wanted to expand on these early successes to improve other business outcomes.

DESIRE TO CHANGE

Staying in good graces

The bank knew that, hidden in the more than 550,000 calls it fields every month, a small number of customers are dissatisfied enough to escalate their complaints to a lawsuit or regulatory filing. By understanding the customer journey up to that breaking point, Banco BMG reasoned that it could reduce the number of high-risk outcomes by taking swift action to address those customers’ concerns. In effect, the bank wanted a new KPI in its call center: verbal dissatisfaction cues from callers.

THE SOLUTION

Building a Lawsuit Prediction Model from past experience

Banco BMG identified nearly 25,000 customers who had initiated legal proceedings over an 18 month period, and analyzed the more than 170,000 interactions associated with those customers. That analysis became the basis for a new predictive model which is now applied to all new interactions.

The organization wanted more than the KPI and more than a predictive model to provide early warnings. The real goal was to reduce the overall incidence and cost of these highrisk situations. The bank picked two of its best customer service agents, provided additional training, and gave them a new exclusive duty: to make things right with these customers on the edge.

THE RESULTS

Strong warning signals and a high success rate

The Nexidia analysis flagged several strong indicators of trouble, including:

• Multiple calls to deal with the same problem
• Direct threats of escalation to an outside authority
• Direct mentions of dissatisfaction
• Remarks showing frustration or loss of confidence, such as “I give up,” “I’m tired,” or “I can’t stand it anymore.”
• Complaints about missing information
• Complaints about slow resolutions or a lack of response
• Requests to see the legal documents governing an account

From these indicators, Banco BMG generates daily alerts of callers who meet one or more of these criteria. These alerts are forwarded to the specialists, who quickly analyze the cases and call the customers with a resolution or a targeted request for more information.

This specialist team reaches out to hundreds of clients each month. Early results show that the specialists resolve 62% of these high-risk issues, referring the rest with a refined and high-priority support ticket for further action by the bank.

THE FUTURE

More to learn, more to improve

As the program continues and the bank gains more long-term data about the effects of the early intervention in high-risk situations, it will be able to study the impact on customer satisfaction and Net Promoter Scores. From the initial analysis and early months of interventions, Banco BMG has already identified opportunities for improvement in several departments including the contact center, marketing, ombudsman, and product development functions. And the Lawsuit Prediction Model program is being considered for other Banco BMG business units, such as insurance.

The project will also further explore the differences in voice profile between customers who actively mention taking legal or government action while on the call from those who never mention those topics but use phrases that indicate high levels of frustration, impatience, or loss of confidence. The hope is to identify even more potential threats which might otherwise sound like normal or neutral interactions.