Driven by the promise of sophisticated reporting and analytics capabilities that hold the potential to help identify actionable items to improve care quality and cost, raise consumer engagement and reduce overall disease risks, payers have invested significant resources and efforts to create large data repositories commonly referred to as lakes. These data lakes were supposed to ingest all forms of data – from complex clinical provider information to claims data to ever-increasing consumer generated data – and generate powerful insights for payers to improve business performance.
While today data lakes are increasingly prevalent, not all its users have become supporters. For some, they have caused significant problems, with little by way of results. One particularly frustrated senior executive referred to their effort as less of a lake and more a data swamp.
Though promising and powerful, investing in data lake technology is by no means sufficient to achieve results. In our experience, payers need to ensure three basics are in place before they undertake this complex endeavour:
Ensure you have a clear business need, with clear management ownership: Like virtually all technological innovations, chances of success are high only when innovations are applied in the context of a strong and clear business need. Before undertaking the creation of a data lake, payers need to clearly spell out what they are seeking to achieve. What reporting needs do they principally intend to fulfil? What analytics and business drivers do they intend to clearly understand? Will business users (e.g. senior case managers) be able to interpret the data (e.g. HL7 messages) effectively? Questions such as these are crucial to allocate resources, management support and funding.
Undertake the journey with a clear understanding of your current reality: Data lakes can enable extremely powerful analysis and reporting of business problems and performance. And yet, their value is heavily circumscribed by the robustness and reliability of your underlying data systems, input data completeness and quality. Data lakes can only enable you to understand answers to data you first own – they cannot provide answers for information that does not exist. Choose the ambition, scale and prioritisation of your needs carefully!
Ensure rigorous and robust data governance: With various types of data being brought together without transformation, health plans must ensure the overall management of the availability, quality and security of data. Payers must pay particular attention to tracking the evolution of data as it moves through the enterprise – for example, tracking claim amounts at a point in time for member-level wellness ROI while receiving claim adjustments.
In conclusion, while data lakes and data warehouses hold the promise of bringing tremendous analytical insight and clarity to business problems, their development must be undertaken with care and the presence of a well-defined roadmap is mission-critical. All in all, these are very interesting times, especially in the health plan industry and I look forward to helping customers overcome such obstacles and avoid unwanted vegetation in their lakes!