Top Telecom provider transforms Customer Service Operation with Machine Learning
As an established global leader in the highly competitive telecom market, this provider understands the importance of delivering sustainable digital experiences to its 400+ million customers. In late 2018, a critical challenge was connecting disparate technical expertise across its global operations for the local markets it serves – from India and Egypt to Europe to North America and New Zealand – to be data driven.
The company’s Technical Shared Services (TSS) group is responsible for overseeing operations for 26 local markets, which run largely as independent business entities. TSS was tasked with creating machine learning knowledge and making it a systemic capability across all local markets to improve customer experience, reduce operational costs, and standardize workflow processes.
The team sought-out Gigster for its expertise in software innovation processes, capabilities for handling large change initiatives, and track record for delivering successful enterprise outcomes.
With over 40,000 services and 9000 employees across all local markets, there are significant inefficiencies running as disparate business units. TSS focused on using machine learning to transform IT service operations. It tasked Gigster with building a “cognitive platform”, an intelligent, cloud-based environment for service failure prediction and more. Gigster approached the major initiative with a two-part plan:
First, silos between teams and technologies were eliminated, enabling a new standard for remote collaboration across the business. Gigster brings together the best people for every project, regardless of their location – including experts in ML and AI.
Second, Gigster leveraged a proven innovation process and plan-of-action to move the project forward from initial proposal through execution.
The first phase of the project – including successful launches in three large test markets – was completed in late 2019. Machine learning algorithms were employed to look at service log data and predict an outage in a particular region. The results demonstrated the merit of the project – the company predicted service failures two hours in advance with 70% accuracy, and one hour in advance with 80% accuracy.
According to the head of operations, engineering, and innovation, “[Gigster is] perceived as…a natural partner, because they make it simple, they make it flexible. They delivered something to us that we will use long-term, not just a one-and-done release,” he said.