When discussing the last mile of data delivery and consumption, particularly the goal of “democratizing access,” it's essential to consider two distinct personas: the analyst and the decision-maker.
Over the past decade, significant strides have been made in generating data assets for the technical analyst—individuals who can access datasets from various sources and perform flexible computations. While this persona typically resides within the data team, some members of finance, product, and growth teams also fulfill this role.
The expansion of tools and processes for data pipeline monitoring, observability, cataloging, and a renewed focus on data modeling have all contributed to the creation of reliable assets primarily designed for the analyst.
Once these data assets are available, the analyst is tasked with transforming the outputs into metrics and dashboards for the decision-maker. However, this workflow faces three key challenges:
- Incomplete View: Given the evolving nature of business processes and the complex, combinatorial aspects of metric calculations, the dashboards produced lack completeness for the decision-maker. Simply put, it is impractical to compute every possible calculation upfront.
- Information Overload: Paradoxically, even with the filtered dashboards currently generated, decision-makers may not have the time or energy to sift through a swamp of tables and charts. Consequently, the analyst is now tasked with distributing relevant information and insights, which may be less effective as they lack the rich business context.
- Evolving Use Cases: The operational workflows around metrics are expanding. It’s no longer just about observing metric changes; the traditional “dashboard” interface is inadequate for exploring deep segmentation insights or determining necessary actions. This inadequacy often leads analysts to spend a significant portion of their time on “ad-hoc” analytics work.
To address these challenges, we must advance the field of data modeling by focusing on metrics and their inter-relationships through the use of metric trees. Furthermore, we need to develop applications that leverage this metrics back-end to facilitate repeatable workflows for users across the organization.
These applications should ideally be low/no-code, enabling users from both data and business teams to extract meaningful insights regarding business performance and operate effectively.
By combining metrics modeling with streamlined metrics workflows, we can unlock the last mile of efficient data delivery and enhance effective consumption.
