11 questions for building an analytics program

11 questions for building an analytics program

  • Share

Increased competition and other market forces are driving insurers to embrace advanced analytics. However, many insurance analytics projects have failed to produce expected results because they have been viewed only as technology implementations.

Analytics success requires a comprehensive approach that addresses strategic, operational and cultural factors across the entire business. To get the greatest return on an analytics investment, it’s crucial to ask the right questions throughout the four phases of an analytics project: prepare, design, implement and monitor.


The prepare phase focuses on properly setting the context and scope of the work to follow.

  1. Are we ready for advanced analytics?

    Assess current levels of analytics literacy, skills and resources. Also, ensure the company’s financial position is stable enough to adopt an advanced analytics project.

  2. What is possible?

    Clarify the business problems that need to be solved and innovation opportunities to be seized and how advanced analytics can help.

  3. What operating assumptions are holding us back?

    Identify the assumptions being applied to the business and their effects on key operational decisions.

  4. How does analytics help us win?

    Consider how advanced analytics will help the company alter the competitive landscape and drive market success.


This phase addresses the quantitative elements of a project.

  1. Do we have the data we need?

    Analytics teams and tools need speedy, reliable access to high-quality, well-structured and granular data from operational and transactional systems.

  2. What’s the right technical math to analyze the problem?

    Following proven practices and iterating the design appropriately will produce a model with the correct technical math. This is critical for testing hypotheses and demonstrating how to improve performance.

  3. What’s the correct business math for the decision-making process?

    The key to determining the best business math is to connect the outputs from technical models to business decision-making processes. This step will yield a compelling business case.


Use this phase to put into place a practical and actionable rollout plan.

  1. How do we generate buy-in and “commercialize” the analytics?

    Full support and involvement from both management and operations is critical to success. “Commercialization” comes into play when the project gains agreement on the scope, approach, metrics and responsibilities.

  2. How do we orchestrate an effective rollout?

    Deployment of new processes must be coordinated with other business changes and infrastructure upgrades. New metrics and measures associated with the affected operations should be set, and proper internal and external communication, training and change management techniques must be applied.


This phase should incorporate the individual project and new analytics capabilities instilled in the organization.

  1. How we do measure and define success?

    Key performance indicators and supporting metrics must be established, and the aggregate impact or value-change to the entire business – not just the narrow function implementing the change – should be determined.

  2. What can we learn from our successes and failures?

    Successful and failed projects should be monitored with the same scrutiny. Failures are often attributed to faulty model designs when the real problem may lie with implementation.