Guavus Blog

All the latest from the world of Guavus

Tractica Research Report – Artificial Intelligence for Telecommunications Applications

by Mark Beccue, Principal Analyst & Aditya Kaul, Research Director at Tractica.

 

The content below is an excerpt from the Tractica Research Report on Artificial Intelligence for Telecommunications Applications, published in 3Q 2019. This Tractica report details the major market drivers and barriers, technologies, key players, and forecasts related to eight telecom AI use cases.

Guavus sits squarely in the customer experience/service delivery solution space. The company was acquired by Thales in 2017. According to Thales’ acquisition press release, Guavus supports more than 20 major telecom operators around the world, including Verizon and Jio.

“Our products provide customers with two main things, operational intelligence and end-user intelligence,” said Stephen Spellicy, vice president of Marketing and Products. “We help them cut costs or help them make money through new services, through understanding their customers user patterns and behaviors. We look at service assurance, customer care, subscriber profiles, and subscriber behavior on the network.”

AI plays significant roles in a range of Guavus solutions, including the Reflex Analytics Fabric (RAF) and products called Alarm IQ, Live Ops, and Marketing Insight. RAF is a platform that enables CSPs to develop customized analytics-based solutions without requiring the expertise of data scientists. In a focused On the Radar analysis, Ovum analyst Adaora Okeleke said:

RAF creates a bridge between big data platforms and data-science platforms, mapping the analytical algorithms and data models created offline in data-science platforms into production big data platforms, making the integration between these easier and quicker … Guavus Live Ops ingests network and customer data in real time, correlating these with real-time data obtained from other sources such as customer calls, trouble tickets, and network alarms. These data streams are normalized and joined with data at rest such as device type, firmware version, and network topology to create an enriched set of events and performance data … Guavus Alarm IQ correlates and prioritizes network alarms, identifies the root issues of the alarms faster, and predicts which equipment failures will cause the largest impact on a CSP’s customer base. 

Across these solutions, Guavus utilizes ML algorithms within its prebuilt applications to extend detection capabilities to unknown scenarios, as existing rules-based approaches are limited to known scenarios. 

Kent McIntosh, head of Marketing, gave some examples of how these AI-driven solutions are performing: 

Live Ops, which is being used across network maintenance and operations, customer care, dispatch centers and application/service support, addresses the whole notion of investigating a specific problem experienced by an individual or group of subscribers, and restoring service as efficiently as possible. For MSOs, it’s reducing care calls, truck rolls and trouble tickets. One operator has been reducing truck rolls by 4-7% and trouble tickets by 5-8%, which has translated into an annual seven-digit OPEX savings. With improved customer experience that company saw a big increase in their net promoter score. 

For Alarm IQ there are two ways to reduce alarm noise: auto- suppress alarms that correlate with planned maintenance and auto-suppress alarms that have a low probability of leading to a network incident. When predicting incidents, it can also prioritize and auto-escalate alarms that have a high probability of leading to a network issue. To understand the impact, we estimated one alarm per second received by the NOC for one of our customers. That’s 86,400 alarms a day and only 2-3% of alarms lead to incidents. 

The Marketing Insight product uses AI/ML to automatically build subscriber profiles from network interactions, with content consumed and apps used, and when, where and for how long. Those profiles also contain static stuff like demographic and socio-economic data, purchase history and service subscriptions. It applies behavioral analytics to those profiles to automatically classify subscribers by interests and affinities, so the output is target market segments. An operator can then use it with a campaign management system to automatically serve up personalized ads and offers to subscribers in real-time. For example, an operator was using demographic-based targeting, and only 6% of users were clicking through and downloading an app. Using the Marketing Insight solution, the operator increased downloads by 7X. 

To see the full Tractica report, please visit this page.