If you start researching this topic, there seem to be various views on what exactly a Metrics Store is. First of all, it is about an additional, intermediate area between the data ware-/lakehouse and other systems, esp. BI/analytics solutions.
Why is such an additional area useful?
Metrics are usually defined in data warehouses and business intelligence systems, but this can lead to problems, especially with increasing data volume and complexity. These include inconsistencies in key metrics definition or the inability to reuse previously defined metrics.
A metrics store is a different story. It is a centralized and managed place for organizations to store and access KPIs/metrics. Like a “corner store”, this data access layer provides employees with important KPIs.
Now what’s the challenge?
Metrics are usually defined in data warehouses or BI applications. However, this can present several challenges. If metrics are defined at a high level and do not provide a way to analyze the underlying data, it can be difficult to identify specific areas for improvement or diagnose problems. This is compounded by a lack of flexibility.
A too inflexible or non transparent organisation of key figures can lead to many individually defined figures. In the worst case, it can lead to different teams getting completely different reporting numbers for simple business questions, and no one can tell exactly which number is correct.
Overall, defining metrics in data warehouses or BI applications require careful thought and planning to ensure they are accurate, align with business goals and provide meaningful insights.
So a metrics store – what for?
A key feature of metrics stores are their flexibility. They should support a wide range of data formats and sources, which enable organizations to collect and store data from various systems, applications, and devices. This makes it easier to integrate data from different sources and to perform cross-system analysis. They are highly scalable and allow organizations to store and analyze vast amounts of data over time, without having to worry about performance or capacity constraints.
What’s your definition of it?
Companies like Airbnb and Facebook already work with metrics stores. This is probably why it’s becoming a buzzword in the US. Other companies also already work with a metrics store, but maybe under a different name. And there are overlaps with other existing concepts such as data catalogs, data marketplaces or semantic layers. Companies should start with a look at their current situation around definition and usage of the metrics in their BI & analytics systems to evaluate whether they have room for improvement and a concept such as a metrics store could help to improve the situation.
