Michael Becker

Director, Analytics and Data Science, RingCentral, Inc.

Supernova Award Category: 

Data to Decisions

The Organization: 

RingCentral, Inc. is an award-winning global provider of cloud-unified communications and collaboration solutions, providing innovative and secure ways to connect employees, boost workforce productivity, share knowledge, and strengthen customer relationships. It provides an open platform that integrates with today’s leading business apps while giving customers the flexibility to customize their own workflows. RingCentral solutions empower today’s mobile and distributed workforces to be connected anywhere and on any device through voice, video, team messaging, collaboration, SMS, conferencing, online meetings, contact center, and fax. 

The Problem: 

RingCentral had several goals involved leveraging data insights but the volume of data and complexity of data formats made ingesting data, managing data pipelines and applying analytics a challenge.In order to get a 360degree view of a conference call they needed to address 3 mission-critical areas:Quality of Service: RingCentral handles 200+ million calls/month and it’s critical that call quality meets customer expectations.While RingCentral had capabilities to forensically assess call quality after the fact it needed a way to detect and remediate poor quality in real-time to quickly pinpoint problems whether in its system or within their providers’/customers’ networks.Fraud Detection:Fraud is an ongoing threat as malicious users attempt to hijack accounts and mount large charges on international carrier networks that adversely impact their customer base and brand reputation.Detecting fraud through real-time analysis of network usage and carrier records is key to shutting down bad actors but with legacy solutions RingCentral could only monitor a limited set of resources.Product Usage:RingCentral needed to understand how customers are using their technology in order to improve customer experience, as well as product stickiness and virality.Various departments across the company want to gain insights from data but RingCentral lacked a consistent way to stream the data and required many hours to create reports and run analyses.

The Solution: 

To address these real-time analytics challenges, RingCentral partnered with StreamSets and Cloudera to build out dataflows into an enterprise data lake to store and analyze its data. RingCentral manages their date pipelines with StreamSets DataOps Platform, which helps them ensure relevant and reliable data is immediately available to all areas of the organization. 

Quality of Service: RingCentral leveraged StreamSets Data Collector and scaled resources with StreamSets and Cloudera to build out dataflows into an enterprise so that call detail records (CDRs) are evaluated in real-time through data lake to store and analyze its data. 

Fraud Detection: The product usage data lake also alerts when usage and carrier rates exceed thresholds based on historical norms, which usually indicates a fraud incident. 

Product Usage: RingCentral created an enterprise data lake, which records all product usage and ties that information to a specific user, tracking daily activity.

The Results: 

By leveraging StreamSets and Cloudera, RingCentral has addressed its three targeted initiatives, while also creating new capabilities and opportunities for product and solution innovation:

Quality of Service: RingCentral now addresses quality of call service in real-time and can make immediate adjustments to their network and communicate issues to their carrier partners and customers. By detecting sub-optimal call quality, they can directly reach out to customers and get ahead of potential satisfaction issues. Given the importance of low churn rates to a services business this ability to attend to quality in real time can substantially impact revenue. 

Fraud Detection: RingCentral now monitors usage and carrier charges in near real-time allowing them to identify bad actors before they incur large charges for their customers. This helps them to more efficiently prevent fraud and protect their brand reputation.

Product Usage: RingCentral now has a single place to retrieve product usage information, with roll-up information based on each use case. Product usage is now analyzed in real-time and they continue to enhance this product usage database. Sales, marketing, engineering, operations and support all have self service access to the data for use in their specific analytics activities, complete with access controls.

Metrics: 

Leveraging StreamSets has allowed for increased fraud and active detection by the teams at RingCentral. The product usage data lake alerts when usage and carrier rates exceed thresholds based on historical norms, typically indicative of a fraudulent incident. RingCentral now monitors usage and carrier charges in near real-time with Cloudera and StreamSets, enabling them to identify bad actors before they incur large charges for their customers. With StreamSets, RingCentral is now able to identify fraud multiple times per week. They’ve gone from hundreds of open cases of fraudulent activity to nearly zero open cases and they’ve saved nearly 1 million dollars per year. The combination of tools has enabled them to more efficiently prevent fraud and protect their brand reputation.

The Technology: 

StreamSets transforms how enterprises flow big and fast data from myriad sources into data centers and cloud analytics platforms. Its DataOps platform helps companies build and operate continuous data flow topologies, combining award-winning open source data movement software with a cloud-native Control Hub. Enterprises use StreamSets to enable cloud analytics, data lakes, Apache Kafka, IoT and cybersecurity.

Disruptive Factor: 

RingCentral knew it needed a 360-degree view of a conference call, so it took the initiative to evaluate the problem and identify the three mission-critical areas. After implementing the technology and experiencing the results, RingCentral disrupted the status quo by changing its quality of service, fraud detection process and product usage analysis. 

Quality of Service: The distributed processing and elastic scaling capabilities of StreamSets’ Data Collector and Apache Spark allowed RingCentral to burst processing capacity to more frequent quality sampling so that analysis happens faster with higher fidelity. 

Fraud Detection: As attacks remain a threat, it can be a “make or break” type situation for a company. Leveraging the product usage data lake alerts allowed RingCentral to identify and take action against bad actors faster. 

Product Usage: RingCentral’s enterprise data lake gave the company full visibility on product trends and customer feature utilization. These insights are shared between departments to optimize a variety of processes and feed real-time reports and dashboards.

Shining Moment: 

RingCentral’s data lake architecture and DataOps platform combined allows them to run multiple analytics business imperatives on a single set of data across multiple analytics solutions. In addition to their data lake, RingCentral also does widespread ETL and data processing for a variety of data sources using StreamSets. They push that data to a Vertica database for access via Tableau for business intelligence. 

About Your Organization

RingCentral (NYSE: RNG) is a leading provider of global enterprise cloud communications, collaboration, and contact center solutions. More flexible and cost-effective than legacy on-premises systems, the RingCentral platform empowers employees to Work as One™ from any location, on any device and via any mode to better serve customers, improving business efficiency and customer satisfaction