Step 3: Create key metrics & glossary terms


Metrics documentation

A metric or Key Performance Indicator (KPI) is a quantifiable measurement used to evaluate the performance or success of an organisation, process, team, or individual.

💡 Putting metrics under a domain or a team bears the risk of adding multiple definitions for the same business concept. For that reason we recommend having a separate section for metrics that is either flat or is only grouped by the type of metric (and not by domain, team etc).

For example you can group MAU, WAU, Number of events etc under Adoption. But we don’t recommend grouping them under Product for example as other departments / teams might play a part in influence them.

When documenting you should consider what questions you want the documentation to answer. Some question examples can be found below:

  • Why was the metric created? Describe the purpose of the metric

  • Who should use it? Target audience

  • What does it measure? Describe where is it commonly used

  • How and when? How is it calculated, when and how often is it refreshed

  • Anything special we should know? `Eg always uses a specific filter

Metrics and KPIs will be present in one or more dashboards, datasets / semantic models & fields and can be sourced from one or two tables and columns. We recommend using the Pinned Assets in order to highlight these relationships.

Each metric should have at least one owner and corresponding domain / team tags.

Each metric should have at least one pinned asset to either dashboard or source table used in the calculation for the metric

Glossary Documentation

Under glossary (or business concepts) you can add any of those business terms that don’t fit under the metric umbrella.

Some examples could be:

  • acronyms: eg GDPR, QBR, etc

  • business groupings: eg territory - EMEA, APAC etc

  • business terms: eg active employee, commercial / enterprise account , customer etc.

The documentation of these terms should include a business definition, a technical one and possibly the source of the data

Last updated

Was this helpful?