8x8: delivering customer insights from thousands of call hours using NLP

“You can't use a traditional data warehouse to do some of the things we're trying to do, such as offering actionable insights in real time.”

Solution 

Working with Cloudera, 8x8 is bringing together hundreds of terabytes (TB) of call data, including real-time stats and voice recordings, with millions of new call records being added daily, and applying machine learning to deliver intelligent interactive voice response (IVR), call routing, contact center, and fraud detection capabilities.

“With Cloudera, we can analyze large volumes of data in real time and make decisions about it,” said Mukerji. “These types of use cases just aren’t possible with traditional platforms.”

Added Dr. Ali Arsanjani, vice president of AI and Machine Learning at 8x8, “Through this work, we’re delivering new capabilities and deeper insights that innovate unified communications and contact center services and truly transform the customer experience.”

Implementation 

In implementing its modern data platform, 8x8 sought to deploy some uses cases on prem and others on the cloud. “Our goal is to move to a hybrid solution, where we have microservices and data stacks in both places so we’re not tied to one vendor or one approach,” said Mukerji. “Cloudera's approach makes it easy for us to deploy both on prem and on the cloud.”

Results

The end result of its new intelligent services, said Mukerji, will be “greater call center efficiency, less hold times for customers, and many more intelligent capabilities that help both the call center owners and the customer satisfaction rate.”

Take the company’s X Series platform, powered by Cloudera. This integrated solution for voice, meeting, team collaboration, and contact centers provides enterprises with context-rich engagements, collaborative contact center capabilities, and advanced customer journey analytics.  One of the many benefits: enterprises can easily analyze all employee and customer interactions from initial call to the contact center through issue resolution. It’s a significant step forward for continuously improving the customer experience. 

Using natural language processing (NLP), 8x8 can also analyze call center data and provide its clients insight into customer sentiment. “We’re trying to bridge the last mile when it comes to call center data and convert it to something actionable,” said Mukerji. “With some of our NLP technology, we can summarize for clients the key takeaways from thousands of hours of calls—are people happy, are they buying your products, are they complaining.”

Likewise, communications providers traditionally assess call quality after the call has ended. 8x8 is working towards a more proactive approach so they can influence the customer experience in real time. “We want to provide a seamless experience for our customers, and proactively take action while the call is still in process,” said Mukerji.

The company is also applying this same approach when it comes to fraud detection. For example, hackers often hijack phone accounts to rack up thousands of dollars in calls to premium rate 1-900 numbers they’ve created. The account holders end up with a hefty phone bill, while the criminals receive a tidy return on their investment. Using machine learning, 8x8 can identify and protect its customers against these types of schemes before they’re affected. “Fraud is a big problem for everybody in the phone space, and we're working to detect it in real time so we can stop it before it happens,” said Mukerji.