Among significant technologies today, and those foremost in a marketer’s arsenal, are Big Data analytics, artificial intelligence (AI), and machine learning (ML).
Whilst these technologies aren’t new, their effective usage is still embryonic. In common with other technology advances, their commercial value lies in how they’re applied. The challenge for the communications service provider (CSP) is how to tap the new opportunities they afford.
Given that AI and ML in particular can be leveraged to significantly enhance marketing programs, the CSP marketer’s initial challenge is to identify their potential impact on service innovation. To do this, two separate considerations must be taken into account, says Adhish Kulkarni, SVP customer value management and loyalty at Evolving Systems.
First, in commercial / business terms, can AI/ML help to resolve a particular problem more effectively than it has been addressed in the past? Progressive technology must be able to resolve business problems in a more sophisticated way and to provide outcomes that have not been possible before.
Second, the technology itself. The business impact sought above will need to be embodied as models, logical regression tables, rules and algorithms (like naive Bayes rules, decision trees, support vector machine, KNN, sentiment analysis, and so on) in order to deliver a result likely to improve on past performance. Can this be done?
With AI and ML, CSPs are well placed to innovate. However, though the progressive business manager does not need to master writing code he/she will need to be able to understand the concept, requirements, and Use Cases that relate to these new technologies in order to use them effectively.
This being the case, the telecommunications and Internet industries must more effectively leverage the large volumes of unstructured data they generate and this is where AI and ML come in. Indeed, as the industry continues to grow, yet more complex unstructured data will be created. It will be imperative to comprehend these mammoth volumes of data if CSPs want to lead their markets by delivering better customer services, identifying needs and offering solutions based on effectively utiliSing what is initially little more than a Big Data repository.
This means that to compete successfully today’s CSP has to store, manage, and manipulate vast amounts of disparate data at the right speed and at the right time. To gain the right insights, Big Data can be understood in the context of four characteristics: The four ‘Vs’:
Volume: With the concept of unlimited calling and the growth in usage of mobile data and broadband services, we have seen the volume of data created increase. With the arrival of an operator like Jio, global analysts forecast a demand for mobile data at 500-600 crore GB / month in India alone. Thus CDRs (Call Detail Records) are increasing and the resulting volumes of data cannot fit an SQL or Excel query anymore. Big Data analytics must be used to derive more meaningful information from the humongous amount of raw data being generated.
Velocity: How fast data is processed. A CSP customer may call his service provider’s customer care helpline multiple times, for instance to activate a service or to connect or disconnect an add-on. When this happens, the marketing or customer service team must be able to act immediately. With customers having so many options available, he or she may be rapidly lost if response is slow, and it is the leveraging of Big Data that can ensure this doesn’t happen. Big Data helps to combine numerous data sources and generate meaningful information which can lead to action quickly, for instance through a map-reduce format.
Variety: Composing useful data. A CSP customer may order a third-party or partner product, for instance from Amazon, use the big basket website for another transaction, take an uber to work, download the latest version of PUBG mobile, start a Spotify premium subscription and watch a favorite show on mobile on Netflix. Besides this, the customer may send occasional SMS’, make calls and receive calls from a single number plus have subscribed to a caller tune. How can meaning and direction be gleaned from these seemingly disparate activities? There is a gold mine of information at hand that can deliver commercial success, but only if the right data monetiSation processes are in place.
Veracity: The accuracy of data is critical. Wrong data means wrong insights and thus selling the wrong products. Big Data and Machine Learning algorithms can easily detect anomalies in data points for a course correction at the earliest opportunity.
With the above in mind, how can telco operators actually leverage Machine Learning? Here are a few approaches to consider:
Offer personalised content to customers: Personalisation is a prime driver of marketing success nowadays. Advances in technology, data, and analytics must be used to enable marketers to create more personal and “human” experiences.
Telecom operators can personalise their offerings by directing targeted communications to users with certain profiles. Moreover, they can tailor their overall offerings, such as types of subscriptions to certain segment of users, which can be used to optimise campaigns and help in targeting customers effectively.
Combat customer attrition: Predicting attrition can help optimise retention results and achieve ROI targets. It is almost always the case that acquiring new customers is more expensive than retaining existing ones. As a result, preventing customer churn is a recurrent concern for mobile operators.
Predicting churn in an accurate and timely way can have a significant impact on customer retention. Machine Learning can help ensure that data-driven recommendations reach the right audience at the right time and achieve revenue uplift.
Deep dive into the customer behaviour: Analysing customers’ transaction patterns, demographics, past trends, and other attributes enables operators to devise effective strategies for engaging them further. In a Telco with various product offerings, the focus should be on customer characterisation to ensure ease of targeting, marketing, and offering personalised products to retain profitable customers and increase ROI across the base.
Evaluate loyalty programs’ impact: Measuring loyalty program performance is essential to understanding its benefits, and to taking corrective action where required. In such contexts the research question is typically whether the program caused an observed behaviour of interest to occur.
While this is ideally addressed with experimental designs, forming randomised comparison groups is often unachievable in practice. A Machine Learning approach boosts the effectiveness of such analyses and provides valuable insights into the program impact.
To summarise, the amount of data telecommunication service providers produce is far higher than in past and is still increasing rapidly. CSPs must urgently start using Big Data and Machine Learning technologies to simplify and derive ‘usable’ (meaningful) information from the new data points to achieve superior decision-making that positively impacts revenues and customer satisfaction.
Machine Learning opens new horizons for sets of information, creating models that can add a new dimension to decision-making. The iterative aspect of Machine Learning is particularly important because, as models are exposed to new data, they can independently adapt. They learn from previous computations to produce reliable, repeatable decisions and results. Doing so is a science that’s not new, but one that has gained fresh momentum.
The author is Adhish Kulkarni, SVP customer value management and loyalty, Evolving Systems.
Follow us and Comment on Twitter @TheEE_io