Kimball:BI developers :: Metric Trees:Data Consumers
Metric trees and their methodologies have the potential to revolutionize data consumption, similar to how frameworks like Kimball transformed the work of data engineers and BI developers.
The most frequent operations for building reports involved
- aggregations to generate metrics and
- slices to dimensionally cut and disaggregate the same metrics
Recognizing this recurring pattern, a framework like Kimball established standards at the data modeling layer that made it significantly easier to build datasets tuned for BI reporting.
By using just three core concepts - facts, dimensions, and attributes, - the Kimball methodology has saved countless hours of repeated work for cohorts of data/BI developers. In fact, the modern cutting-edge metric layers borrow their abstractions from Kimball’s approach.
Now, when it comes to consuming and extracting value from data, the common analytical operations are:
- Root causing a metric change, whether due to external or internal factors.
- Treating segments as first class concepts in any decomposition.
- Assessing metric performance against targets.
- Re-assessing, pacing and forecasting metric trends.
- Simulating scenarios to inform tactical or strategic decisions.
- Establishing metric goals tied to the desired outcomes.
Recognizing these common operations across any organization, a graph of metric relationships (metric trees!) brought to life on top of the data platform can be immensely powerful.
The operations listed above and more could be streamlined, even automated via applications that operate directly on these trees.
This presents an exciting future where we continue to advance data standardization and democratic enablement in our ongoing pursuit to deeply understand and operate our systems.
