At its highest aspiration, the role of data is to model a "digital twin" of the real-world business. In this context, metric trees serve as the most crucial abstraction, acting as the system and process digital twins by capturing the intricacies of the business operations.
The concept of a digital twin originated in manufacturing and industrial applications, where it refers to a virtual representation of a physical system. This system is constructed by instrumenting and tracking extensive amounts of data from the physical world, and connecting them to provide a real-time, holistic view of how individual components interact to drive the larger system outputs. Such a virtual system enables accurate monitoring, ongoing performance analysis, and optimization.
Data professionals—such as platform engineers, analytics engineers, and BI developers—have been implicitly moving towards creating digital twin equivalents for the world of business operations. Let’s delve deeper into this analogy through the key concepts of a digital twin.
1) The Component Digital Twin
The first layer of a digital twin consists of individual components. In the data world, these are atomic observations and facts—raw data points that capture events and the properties of these events. For example, in a sales context, these might include an event like a product demo, characterized by attributes such as the type of account. Each observation forms a fundamental building block, akin to the physical components of a machine.
2) The Asset Digital Twin
The second layer is the asset digital twin, which integrates individual components into a distinct unit with a purpose. In data terms, this translates into processing, and transforming raw observations into meaningful metrics and segment slices.
For example, in a sales funnel, calculated metrics might span leads generated to contracts won, along with key financial outputs such as revenue and associated acquisition costs.
3) The System Digital Twin
The third layer is the system digital twin, which integrates assets into a well-defined system.
In the data realm, metric trees represent the closest concept, providing a structured representation of how different metrics interconnect to form a specific business process.
In the sales funnel example, the metric tree would encompass the entire journey from lead generation to contracts won and revenue generated, with conversion rates and other such metric equations stitching the process together over time. This metric tree can be further sliced by various dimensions, such as marketing channel or industry segment, offering the flexibility and power to generate multiple tree variants from the base system tree.
4) The Process Digital Twin
The final layer of a digital twin is the process digital twin, which captures the full process. In the world of data, this encapsulates the entire business model connecting metric trees across various functions into a unified business model tree. As an example, the sales funnel metric tree would be connected to outputs from other business functions and processes to ladder up to the growth and profitability metrics of the entire business.
Metric trees represent the essential final step to bring entire processes to life through data, advancing us towards the aspirational role of data to create an accurate, actionable “digital twin” of the business.
