Where predictive analytics delivers results

After a decade of BI platforms that promised results but didn’t always deliver, CSPs have every reason to demand better performance. Thankfully, predictive analytics meets their expectations, and companies that use it can expect:

• Enterprise-wide value. Predictive analytics organizes complex operations and customer data into a multidimensional model to facilitate sophisticated analysis so that users in any of a CSP’s departments can directly and easily consume data.
• Deep analysis. It also incorporates intuitive, three-dimensional data visualization, granting CSPs the power to drill down for rapid-fire analysis and problem identification.
• Data integration. By pulling from hundreds of sources, including billing, accounting, network feeds, inventory, customer relationship management (CRM), customer care and activation, enterprise resource planning (ERP), and pricing, predictive analytics reveals what a customer is worth at any given moment in time, whether now or in the future, thus completing the current “mission: impossible” of Big Data.

With these insights CSPs can make informed decisions that precisely determine the lifetime value of a customer, the likelihood of churn, when (or whether) to invest in product development and loyalty programs, and how much to invest by segment and individual customer, as well as the impact on the network, the expected return on investment (ROI) and profit potential.

Finding the perfect predictive-analytics opportunity

The challenge of where to start can slow a CSP down even when it recognizes the need to take action on its Big Data issues. Unfortunately, this inertia is exacerbated by vendors that flood the marketplace with inflated promises of analytics capabilities that generally fall short of what a true predictive-analytics application can and should do. To guide your selection process and avoid potential pitfalls, you need to know what to look for right off the bat:

• Flexibility and scalability. Insist on a cloud-based approach, as it provides maximum flexibility, meaning you can start small and expand, or contract as your needs change, without the risk of upfront capital requirements or licensing costs.
• Industry experts. It is essential that your predictive analytics application be backed by business experts with deep experience in the telecommunications industry and data experts who understand not only how to properly normalize data from the full array of data sources but also how to optimally leverage predictive analytics engines.
• The right questions. The application’s system and design must allow you to ask the right questions and perform specific analyses for your business via multivariate statistical capabilities.
• Operationalization. It must also be capable of becoming an integral part of your everyday business, and it’s imperative that such an application be fully operationalized to capture the full value needed by all teams across your organization.

The top analytics priorities for your chosen application should include:

• Statistical analysis. The application must use mathematical algorithms to identify hidden relationships between events, people or actions. This can take the form of root-cause identification, social-network analysis, behavioral segmentation, or semantic analyses.
• Propensity analysis. Statistical algorithms are essential for showing what will happen to an entity in the short term (e.g., within 90 days of today). In addition, propensity analysis can aid in the understanding of the behavioral drivers of such predictions.
• Forecasting. This capability uses econometric and statistical algorithms to forecast what will happen next week, next month or next year. It also enables “what if?” analyses to help with scenario planning.
• Optimization. This one leverages advanced statistical algorithms to identify the best possible outcomes when given a specific set of constraints.