DTWS: Airtel CTO traces route to greater automation

India is a highly competitive market: 13 operators have consolidated to three in recent years, and the average revenue per user for Airtel’s 340 million customers is $2. Its network has 220,000 towers. Typically each has 12 radios because Airtel’s licensed spectrum is not in contiguous frequency bands, and 20,000 to 25,000 radios are added annually.

Airtel and Ericsson have jointly managed this network for 14 years and Randeep Sekhon, CTO of Airtel and Bradley Mead, Head of Network Managed Services at Ericsson, talked about their data-driven transformation during the third week of TM Forum's Digital Transformation World Series (DTWS).

“The only way an operator can win customer and revenue market share is to give a great experience to customers,” Sekhon explained. “With such a complex network, the only way you can give this experience is through automation.”

Airtel uses the Ericsson Operations Engine. Work began on the platform in 2018, based on input from Airtel, other operators, and partners including TM Forum. They collaboratively developed the model to make the shift from high-cost, reactive, incident-driven processes to predictive and proactive operations to avoid issues impacting customers.

Mead said the engine’s capabilities are built around people and skills because domain expertise is key, plus data and automation engineering. It draws on insights from data analytics and AI to support automation.

“The vision is to drive everything possible through automated processes, including to predict, prevent, and take closed-loop action,” he stated.

The aim is to achieve the lowest total cost of ownership; better customer experience; new revenue streams in from 5G; efficient transformation; and a trusted, transparent, secure business.

Sekhon’s summarized a number of Airtel’s achievements as shown in the graphic below. He said the most important metric is the reduction in customer complaints through channels including social media, Airtel’s HelpMe app, the interactive voice response and call center.

He sees this as a considerable achievement in the face of the network’s growth and increasing complexity, and the rise in traffic.

Mead said the Operations Engine is using ever more AI. As one example, AI-powered dynamic traffic balancing has, on average, improved cluster throughput by 18%. The prediction model generates 7,000 predictions daily, 70% of which are resolved by closed-loop action. This led to a 17% reduction in customers’ complaints.

Mead also notes that Airtel and Ericsson collectively decided to move away from technology-centric metrics to measurements around service and customer experience.

Sekhon noted, “It’s easy to get confused with too many KPIs coming from the network… We focus on the metric of customer experience via customer complaints, but not at the absolute level, but by geography, by type of complaint.” For example, what is the user complaining about? Voice or data, and is that 2G or 4G voice or data? Are they complaining about speed because they can’t open a website or can’t get a game to run properly?

“If you go and measure the way the customer experiences the network using special tools to fix the problems, then watch [complaints] going down, this is the journey we’ve been on together,” he added. “Moving those metrics improves customer experience rather than moving KPIs. Using the voice of the customer as a metric has served us well.”

Mead concluded, “What you measure matters. It changes attitudes and the decision-making process and ultimately changes the culture.”

Watch Randeep's keynote on Airtel’s data-driven transformation journey here