In one of our previous blogs we talked about how machine learning can be used by Mobile Network Operators (MNOs) to provide you with a secure digital identity, analysing your behaviour to determine the likelihood of fraud and centralising your various forms of ID into one place. However, the capacity of machine learning in this market extends far beyond just security use cases and can be highly beneficial to solve problems within network service and improve overall customer experience.
Take the example of Reliance Jio, the world’s largest mobile data network operator, which generates four to five petabytes of data each day, equal to four to five thousand million bytes, or 4015 to 5015 bytes. While this number is massive on its own, it is worth noting that Reliance Jio has nearly 400 million customers across the world, meaning the true amount of data being produced every day is far greater when you include the millions of subscribers with other providers. With such an enormous swathe of data, there is great potential to use big data analytics and predictive analytics techniques to understand exactly what this data means and sort it into ways it can be used advantageously.
Mobile Network Traffic Analytics
For MNOs to be able to function effectively it is important for them to have specific details about when and where their services are most in demand. In the past, the only way to have a rough idea of this was to aggregate the data of their customers’ residences and use this to estimate the percentage of customers in each city/town. It also involved a lot of guess work to decide when and where there would be peak periods.
However, by using machine learning analytics this type of data can be more precisely leveraged for a variety of business use cases. In practical terms, this means that telecom companies can have a better understanding about when exactly they need to boost the network and in which areas of the country this is most important. For instance, if network capacity needs to be increased at 6pm in two big cities they can allocate resources based on the number of customers using the network in real time in each city. This reduces overall costs for the MNO as it gives them the decision-making power to ensure they don’t waste resources boosting the network when it is not absolutely necessary and can scale the network equipment up as required.
The telecom sector is experiencing a massive amount of change in a very short period of time since the COVID-19 pandemic escalated across the globe. One of the obvious impacts is how the network was utilised before compared to now: transformed from a more or less loosely centralised to hyper-distributed subscriber environment. Overnight, many countries have asked citizens to stay home and thus, companies have asked employees to work from home and students to study from home.
This huge shift related to how and where networks are accessed has literally changed the centre of gravity — from a highly concentrated set of larger corporate or enterprise-oriented networks to wholly residential. This has not only challenged businesses of all types as to how they will continue operations in this mode, but it has turned many upside-down.
Communications service providers (CSPs) are now facing a host of new traffic patterns and new levels of network utilisation with the emergence of new voice and video applications.
To assist their subscribers in this time of need, CSPs have been lifting network caps and allowing unlimited access to reserved bandwidth. Many are struggling with the new capacity demands, as capacity is often incrementally planned and deployed with gradual growth in the CSP business. In good times, MNOs would welcome millions of new subscribers (and their traffic) but in this case, this has caught many telecom companies by surprise.
Mobile Network Planning
With the proliferation of mobile devices and their capacity to use the internet in locations traditionally deprived of service, many of us now use our mobile phone to replace tasks we previously would have done on a laptop or desktop. And while in most cases there are no performance issues when we use our phones on the go, sometimes a change in cell towers or Wi-Fi hotspots can result in weak areas of coverage. This can mean that calls made over the internet suddenly lose connectivity at a particular point of your journey. Using machine learning analytics to analyse the geolocation data where multiple connections have suddenly weakened, provides key insights for MNOs. Not only are they able to know exactly where to repair the section of broken network, but they can spend time upgrading their service in the areas where their customers really need it, rather than spending money further improving areas that already receive acceptable signal quality, improving the overall reliability of the network. Here’s a video discussing the success story of how one of the largest mobile operators delivered a superior customer experience and intelligent automation.
Improving the customer experience
With consumers demanding personalized products and services, it has never been more important for MNOs to understand individual subscribers and respond effectively to their personal preferences. As your MNO has access to any anonymous information on your phone, for example what you have downloaded or visited online, they can use this as a base to provide better and more relevant adverts and offers.
Using machine learning, telecommunications companies can categorize this data into meaningful information. For instance, anyone who regularly uses their phone outside of the country they reside in might be offered a phone plan that allows them a certain number of international gigabytes per month for no extra charge, when they renew their contract. Using data to make these decisions means that MNOs do not have to rely purely on the demographics of their customers to tailor their product offering, while still respecting subscriber privacy and regulations, such as the General Data Protection Regulation (GDPR).
Your experience as a customer with your mobile service provider is often the most important reason behind your choice to remain on their network or find another provider. Nonetheless, you may be surprised to read that over half of customers, who experience issues with their MNO do not contact service and support teams. Whether your questions revolve around why your Wi-Fi Calling is using your network instead, or why your phone is not going back to 4G when you return to a 4G coverage area, it can be a frustrating problem that many of us would choose to just deal with, rather than go through the hassle of trying to fix it.
However, machine learning solutions are able to identify these service delivery issues in real time, enabling targeted and actionable alerts, and faster service restoration. For MNOs this increases customer loyalty, as well as increasing efficiencies throughout their network, without having to hire employees to spend time speaking directly to customers in order to tackle these issues.
Machine learning is a valuable asset for MNOs as a way to understand and use the data they are given by their customers. Its potential to lower costs, increase efficiencies, and dramatically improve the end-to-end customer experience across the world is very compelling. In a market that is undergoing significant development, gaining further insight into how subscribers use the internet on their mobile devices will be key to understanding how and where the next generation of network infrastructure, 5G, should be rolled out in the smoothest way possible. To learn more, check out our 5G eBook.