A GCC mobile operator jump-starts a big data program

Our client, a mobile operator in the GCC, was facing declining core market revenues, intense VoIP substitution, high churn levels, trouble monetizing data, decreasing SIM sales, and the need to optimize its bottom line. In the face of these pressures and in order to reverse this trend, the client decided to jump-start a big data and analytics program to mine information about customers, build real-time predictive and prescriptive capabilities, enable real-time performance management, and ultimately extract more value from customers, through average revenue per user (ARPU) uplift, churn reduction, up-selling and cross-selling, and profitability enhancement.

Strategy& was engaged to lay the foundations of the big data program and to spearhead it, working hand-in-hand with the client on the strategy, design, and implementation phases of the program.

How we helped

The Strategy& team developed the big data strategy covering identification and prioritization of big data use cases across core business, operational efficiency, and new business models, in line with the client’s strategic direction, and defined Big Data operating model, technology enablers, and corresponding investments, as well.

On the operating model front, we designed the big data organization structure; described new roles and responsibilities such as data scientists, big data translators, and data engineers; defined interaction models and decision rights, and sized the manpower requirements.

On the technology front, we designed the needed systems architecture based on a Hadoop distribution of a Big Data Lake. We determined the Big Data Lake hardware dimensioning as well as software and tools specifications. We also helped source the Big Data Lake vendor, and detailed post-deployment operational set up and system integration requirements and road map.

We modeled and tested three big data use cases, namely Churn Prediction, Next Best Offer, and Social Network Analysis, leveraging data from various sources including the client’s data warehouse, CDRs, CRM, billing, network signaling, payment channels, and the like. The Churn Prediction model provided the likelihood of a customer churning, the value at risk, and the probability of retention; the Next Best Offer model provided likelihood of micro-segments positively responding to campaigns; and the Social Network Analysis model identified influencers and communities within the customer base using eigenvector centrality, degree, closeness, and betweenness centrality measures across the client’s customer base.

Finally, on the visualization front, we developed automated dashboards using Power BI to track results and performance in real-time, and used the developed dashboards to extract insights and develop an action plan.

The client’s senior leadership endorsed the required investments needed for the program. Operating model changes were approved and enacted. The Big Data Lake was sourced and set up. At the close of our engagement, the Churn Prediction, Next Best Offer, and Social Network Analysis models had been deployed and were fully operational. At that point, the client was in the process of developing recurring, targeted, and multi-wave predictive below-the-line campaigns based on our action plan. Finally, dashboards are now used on a daily basis by various marketing, product, retail, distribution, and customer care teams.