AI Helped One Bank Capture the Voice of its Customer

How AI Helped One Bank Capture the Voice of its Customer

Businesses today understand that they can’t rely on customer satisfaction surveys alone to gauge contact center performance. Via the power of artificial intelligence (AI), data from every call can now be analyzed to produce a sentiment score that more accurately reflects the customer’s experience and provides real insights on how to improve it.

Cincinnati-based Fifth Third Bank, one of the largest regional banks in the Midwest, offers a case study in what can happen when companies embrace AI and speech analytics to analyze and improve customer interactions. The bank recently won “Trailblazer of the Year” honors at NICE’s CX Excellence Awards, part of the Interactions Live 2022 conference, the industry’s biggest CX virtual event.

Fifth Third Bank Vice President, Director of Consumer Ops Business Support, Michelle Grimm, and Speech Analytics Program Manager Kevin Anderson also led a best practice session during Interactions Live and spoke about how the company successfully expanded its use of NICE Enlighten AI and Nexidia Analytics from just a fraction of its overall agent pool to every agent in every department. The result: sentiment scores went up, and costs went down.

Putting AI to use

Prior to 2021, Fifth Third Bank relied on an average of 50 customer satisfaction surveys a week to measure the performance of its agents. The survey responses were skewed toward just a segment of its customers, however, leaving Fifth Third Bank with limited visibility into how it was serving a huge swath of its customer base.

Everything changed last year when the bank took NICE’s Nexidia Analytics live to thousands of agents; since then, the company has analyzed over 15.7 million interactions involving 2,300 agents. Of those interactions, 77% were automatically categorized, making it easier to identify trends and high-risk interactions that needed special attention.

“It almost speaks for itself,” said Grimm, a banking industry veteran. “Our supervisors and agents being able to see every one of their interactions―compared to just 50 surveys―tells a powerful story.” 

Sentiment over surveys

Fifth Third Bank’s transition away from the survey model didn’t happen overnight. It started with an identification of the shortcomings of the existing system:

  • Surveys provided only limited visibility into the customer experience.
  • Coaching conversations ended up focused on metrics instead of specific behaviors.
  • Outside of escalated complaints (when a customer asks to speak to a supervisor), there was no way to identify specific calls that deserve extra review.

The next step was to prove that customer satisfaction scores and sentiment scores were directly correlated, which the bank did by analyzing 8,000 interactions using both survey and sentiment. In the end, the scores were nearly a mirror image.

“This really confirmed our intuition that these metrics are directly aligned and indicative of one another,” Anderson said. “The study also showed that survey scores skewed heavily toward the positive, while sentiment scores provided a greater range of outcomes. It became clear that from a data perspective, the surveys were not providing a complete picture of the agents’ performance or the department’s performance.”

In the months following the changeover, Fifth Third Bank saw immediate benefits, including:

  • Improved employee productivity.
  • Higher employee compliance.
  • Lower costs (as speech analytics identified processes that could be automated).

For example, Nexidia Analytics was able to quickly ascertain that over 100,000 calls a month were coming from bank customers who needed help logging into their online accounts. Made aware of the trend, Fifth Third Banks’s digital team change the language on the website, and call volume fell off precipitously. 

“That allowed us to free up agents for calls where they could really make a positive difference,” Grimm said.

Better data = better coaching

Another goal of Fifth Third Banks’s AI experiment was to improve data collection and thereby deliver better coaching. Enlighten AI learned which agent behaviors had the biggest impact on CX sentiment and then scored each call, accordingly, allowing managers to instantly identify troubling trends or praise positive outcomes.

“(Supervisors) can pull up the dashboard and pick a specific call and view a specific issue to address,” Anderson said. “They’re able to say to their agents, ‘Hey, you need to do better with X, Y and Z. And oh, by the way here’s a call where you did those things really well and a call where you didn’t; let’s listen to those back-to-back.”

A 160-year-old institution, Fifth Third Bank’s longevity can be attributed to many factors, not the least of which is its willingness to embrace the future. Given its early success with AI and customer experience, the company plans to expand its use of speech analytics to support compliance and risk management goals next. 

To learn more, watch the on-demand session, “Transforming Voice of Customer at Fifth Third Bank with Nexidia Analytics and Enlighten AI.” and read the case study.