Ask someone to list life’s essentials without which they can’t survive, and chances are that their phone will feature near the top. With the increasing use of smartphones, the telecommunication & media industry has become an increasingly critical player in our lives, arguably representing the backbone of digital disruption. Yet service providers still have much to learn.
Imagine driving to work and your navigation app prompts you to change routes to avoid a traffic buildup. The source of this information is a combination of artificial intelligence (AI) and machine learning (ML). But if the app would divert all users in the vicinity to the same alternate route, the result would be a disaster. That’s why such apps use highly intelligent algorithms that allow them to manage the traffic flow. While by and large, AI is already well integrated with navigation apps, other clear gaps remain in our industry that AI can help fill.
When infrastructure optimization is lacking, and network performance is substandard, the result is a poor customer experience. To provide the levels of service that consumers expect, service providers must, therefore, optimize the network’s flow of data and calls on their grid. AI can help achieve this through algorithms that automatically increase and decrease network capacity, based on predicted usage patterns according to the time of day, location, and so on. As a result, smart algorithms can be employed to manage a cell phone tower’s network traffic – enabling service providers to plan the logistics for acquiring higher network bandwidth and building new towers where sporadically high patterns of traffic occur. With a smart algorithm that can predict such patterns, tradeoffs between locations that experience high traffic at different times of day can also be formulated so that resources are shifted to accommodate users at multiple locations, as the need changes.
Algorithms can also be applied to manage workforces that maintain cell tower infrastructure. Service providers need to manage maintenance personnel to ensure all their sites remain fully operational, as well as to manage the building of new sites. To maximize company resource utilization, algorithms can predict which locations should be serviced first in order to minimize disruptions to the maximum number of people. Furthermore, AI-powered video analysis based on surveillance camera footage can enable the triggering of an alert to send a maintenance crew as soon as an unanticipated breakdown occurs in a high-value area.
An additional area where AI can add value is customer service. By analyzing an individual customer’s usage trends, customized plans can be created to promote features they frequently consume, and leave out features that they will find less attractive. This also reduces overhead costs by channeling resources to improve popular features, while increasing customer satisfaction and empowerment through personalization.
Technical support provided to the customer too, can be personalized. When contacting customer support, an ML algorithm can predict potential reasons for an enquiry, based on the individual’s usage patterns, historical data and analyzing problems that similar users are experiencing. Then, AI based automatic reply systems like chatbots can provide the customer with rapid resolution, while saving company resources in the process.
If all else fails, AI can help prevent an unsatisfied customer from seeking greener pastures. Churn prediction models can help predict when a user is at high risk of discontinuing or downgrading their subscription. Armed with this information, the marketing team can then proactively target that customer and then, using another ML model, employ the best retention strategy for that specific user.
Creating all the above features requires business support systems (BSS) and operation support systems (OSS) to be optimized. End-to-end testing of all new applications before launch is therefore critical. Traditionally, this would occur according to a “quarterly” cycle. With the advent of DevOps and microservices, however, deployment windows have shrunk, meaning the testing cycle too needs to evolve. By introducing AI into the testing automation workflow, service providers can predict which tests are most critical prior to launching a feature, and direct resources accordingly when tradeoffs are required.
Finally, network operators already have access to a treasure of data. They collect data ranging from users’ locations to usage patterns and network performance. And while statistical analysis of past data is already commonly used for network planning, root cause analysis and so on, it is time consuming and not fully indicative of future trends. With ML, however, such analysis can create predictive models that are both forward-thinking and self-correcting.
While we’re only at the dawn of the digital age, it’s already clear that the possibilities of using AI in our industry remain endless. And over the coming decade, who knows what new surprises AI-driven innovation is likely to bring.
The author Jitendra Dutt Sharma is Head of Competency Centers – Amdocs Quality Engineering Services