Monthly recurring revenue (MRR) is the most argued metric in B2B SaaS. Not because teams can't write SQL, because every team writes different SQL and calls it MRR.
This guide explains canonical metrics: a governed definition layer that sits on your semantic graph, pushes the same KPI to BI and notebooks, and gives RevOps an audit trail when board prep week gets stressful.
The three-definition problem (you've lived it)
- Finance MRR: net of credits, aligned to billing system cutoffs
- Product MRR: activation cohorts, sometimes gross, sometimes annualized monthly
- BI MRR: whatever the Looker explore author typed in 2023
Each definition is defensible in isolation. Together they destroy trust. That's metric drift and it's an SEO problem for your data team because execs Google "why don't our dashboards match" long before they file a Jira ticket.
Canonical metrics 101: four layers
1. Definition
Write the business logic once: filters, grain, time zone, handling of trials and refunds. Store it on the semantic layer, not in a Notion doc.
2. Lineage
Link MRR to fct_revenue → stg_orders → raw_stripe_charges. When finance asks "where does this come from?", you click not grep Slack.
3. Certification
Status: draft → certified → deprecated. Only certified metrics appear in exec dashboards and Company Brain answers.
4. Reconciliation
When Looker differs by 2%, run a diff report: definition version, upstream model change, or explore override. Close the ticket with evidence.
grain: account · month (UTC)
source: fct_revenue → stg_orders → raw_stripe
owners: @finance-ops, @data-platform
status: certified · last_reconciled: 2026-06-20
SEO keywords your metrics layer should own
Data teams win internal search and external discovery when content maps to how people actually ask questions:
- "canonical metric definition"
- "MRR reconciliation data warehouse"
- "metric drift between dashboards"
- "certified KPI governance"
- "single source of truth revenue metrics"
Publishing playbooks on these topics, like this one, signals expertise to Google and to your CFO.
Rollout plan (2 sprints)
- Sprint 1: Pick one exec KPI (MRR or NRR). Document definition with finance. Map lineage in Metroflow. Certify.
- Sprint 2: Point Looker/Hex explores at certified definition. Enable reconciliation alerts on model changes. Train product on Company Brain queries.
Deep dive on the Metrics & Trust Layer or walk through how canonical metrics work in our how-to guide.