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

Recently, Telefónica took yet another step into the world of big data analytics – announcing plans to consistently track customers’ service experiences to improve Quality of Service (QoS). This move follows the launch of its Dynamic Insights Unit back in 2012.

To do this, Telefónica will deploy big data analytics tools within all of its service operation centres. The intention is to monitor customers’ network data to anticipate potential network issues, and proactively respond with tailored actions – including network repairs or optimisation. But is what Telefónica’s proposing really enough?

Big data analytics takes things a step beyond the traditional descriptive approach. It allows predictions to be made about future events based on the analysis of historical data, and recommends actions that maximise the probability of a desired outcome. Yet, what Telefónica describes as a “ground-breaking” project, many other big data industries are already doing and in far more advanced ways, using far more advanced tools – in particular, machine learning.

The battle for mobile spend

Communication Service Providers (CSPs) are facing fierce competition for customers’ mobile spend from digital services like Netflix and Spotify. To win back their share of the market, CSPs must focus on creating a personalised experience communicating value, increasing loyalty, and recognising the individual. This is something for which some digital services are already held up as a gold standard for by customers – a challenge understood only too well by operators.

CSPs acquire vast amounts of data from multiple sources on a daily basis – customer profiles, billing and payments, history of communications, geolocation and network information. If harnessed correctly, these data streams provide CSPs with a 360-view of customer behaviours, preferences and movements – enabling personalised engagement with customers. But isn’t using data to improve customer engagement and drive profits old news? Yes, and no. Yes, because the industry has been talking about it for many years. But no because the current analytics methods being used by CSPs are underutilising this immense amount of subscriber and network data and it’s potential.

Due the sheer size of the task of storing, processing and accessing large amounts of data, one issue many CSPs face is using limited data-sets to action decisions – often done on a “herd” basis, not a personal one.