Forget Big Data – Telcos Can Use Machine Learning For Future Success

On top of this are issues around viewing big data as a source of knowledge – a way to generate insights that enable employees to make better decisions. Many CSPs aggregate, assess and examine the datasets to uncover useful business information – but this is then used to support human decision-making. While this application of big data can serve as a tool for strategic decision-making, it an insufficient use of the volume of data at hand. Using big data to support human decision-making – based on analysis or not – relegates its usefulness to supporting infrequent, localised, biased and late decisions. This makes it very difficult to evaluate the return on investment (ROI) of the implementation of big data solutions and, more importantly, decreases overall returns.

To stop under-utilising the potential of their big data, CSPs need to change their approach to it. This is where machine learning comes into play. It allows routine decisions to be made efficiently on a large scale and supports a wide range of use cases for marketing, risk management, and operations purposes.

Machine Learning for future success

Machine Learning is the technology that can bring real disruption to CSPs. It is a complementary technology used to analyse large datasets, identify underlying patterns, action prescribed outcomes and actively learns from the outcomes to feed future automated decision-making. Machine learning’s greatest advantage over previous forms of data analytics is that it does not require the operator to have a deep understanding of the domain itself, nor to have data interpretation capabilities. So how can CSPs use machine learning to improve customer experience?

When applied to a use case like personalisation for churn prevention, a machine learning model can have a rapid and positive impact on a customer’s lifetime value. Consider being able to identify the warning signs of a churning customer. Machine learning allows the operator to foresee when an individual customer is about to churn, automatically deduce an informed response to prevent this from happening, and then act upon it. This automated and intelligent action is based on how similar customers have acted, predicting a probability of conversion for each possible course of action.

Similarly, with cross-sell recommendations (SMS, on the website or via a contact centre) a machine learning model can recommend which products or services to display to each customer in order to maximise the total revenue, and increase the probability of him accepting the offer as relevant to him. This action is based customer profile data, billing and service usage data, as well as, data on active communications and responses. Unlike the more traditional analytics method, which grouped customers into a limited number of segments, each person gets individual recommendation that reflect their actual and most recent preferences.

While CSPs like Telefónica are taking controlled steps into big data analytics, more CSPs need to begin to embrace the machine learning technologies, fast. With digital services seizing an increasingly large share of the mobile spend, and subscriber data exploding, it would be self-destructive for CSPs to rely on technology just for collecting and extracting data and rely on humans to make decisions. Whether applied to marketing or operational use cases, machine learning technologies have immense potential in telecoms. But to earn the promised benefits, such as lower costs and increased revenues, CSPs must empower machines and intelligent algorithms to make routine business decisions, and importantly, learn from the outcomes.

Try our drones and robots quiz!