Business Metrics
Beginner 12 min read Business Metrics

Business Metric Fundamentals

A number is not a metric until it is defined, owned and gated

A business metric is a number with a definition, a grain, an owner and a deadline. Metrics become KPIs when they gate decisions; the discipline is documenting how each is computed.

metricKPIgraindefinitionbenchmarkleading indicatorlagging indicator
Before we start

What you will learn

  • Define what makes a number into a metric
  • Distinguish metric, KPI and benchmark
  • Write a metric dictionary entry for one metric
Introduction

What is this?

A business metric is a count, rate or amount with a precise definition, a grain (per what time and entity), a source of truth and an owner. A KPI is a metric tied to a target that a team is accountable for. The distinction matters because dashboards full of uncowned numbers quietly become fiction.

Motivation

Why do you need it?

  • Disagreement about 'revenue' is an analytics failure, not a business discussion.
  • Definitions drift: the same column named differently gives different numbers month to month.
  • Benchmarks are meaningless without matching grain and time range.
  • Metric dictionaries turn arguments into lookups.
Think about it

Mental model

Imagine

Recipe and ingredients

Think of

Metric and data

A recipe defines the dish: ingredient list, quantities, method and serving size. A metric is the same — you must know its ingredients (fields), method (formula), and serving (grain/time) before anyone can judge taste.

Mechanics

How it works

  • A metric statement names: what is counted, in which grain, over which time window.
  • The source of truth identifies the table/query that produces it.
  • An owner ensures the definition stays current and the pipeline stays healthy.
  • Targets convert metrics into KPIs with compare/baseline expectations.
  • Reviews compare actuals to target and decide action.

From raw number to governable KPI.

Raw event
Definition
Metric
Target
KPI decision
Hands-on

Examples

A metric dictionary entry

A metric dictionary entry
Metric: Monthly Recurring Revenue (MRR)
Grain: per active subscription in calendar month
Definition:
  SUM(plan_price) over active subscriptions
  contracts starting on/before month end,
  excluding paused and refunded lines.
Source: sus_width subscriptions_fct
Owner: finance-analytics (subdomain)
Target: within 3% of plan for monthly cadence

This entry fixes the grain, the formula, the source table and the owner — so any two reruns of the metric agree, and responsibility for freshness is explicit.

Real world

In production

  • A SaaS team ships a metric dictionary where 'Monthly Recurring Revenue' has a single ironic definition all dashboards share.
  • A marketplace defines 'GMV' with an explicit deduction for refunds so finance and product read the same number.
  • An exec cadence reviews 12 KPIs each with a named owner and a red/yellow threshold, replacing a wall of competing dashboards.
Guidelines

Best practices

  • Document every metric in a dictionary with definition, grain, owner and SQL.
  • Attach SLA dates to KPI reviews so accountability is time-bound.
  • Split leading vs lagging indicators in the cadence.
  • Version the definition when it changes; old dashboards can age out.
Watch out

Common mistakes

  • Counters anchoring to 'the dashboard number' without a definition.
  • Two teams computing 'repeat rate' differently and defending both.
  • KPIs without owners — nobody is accountable, so nobody fixes the pipeline.
  • Comparing metrics across different grains (weekly vs monthly) in one chart.
Root cause analysis

Common issues

Error
Two dashboards show different revenue for the same month
Root cause

Different filters, joins or timezone cutoffs

Fix

Converge on one metric dictionary definition and query that uses it.

Error
A KPI crosses a threshold yet nothing happens
Root cause

No owner assigned or review cadence missing

Fix

Assign an owner and a review date to the KPI.

Error
Monthly metric jumps 20% with no explanation
Root cause

Definition change or source pipeline refactor

Fix

Version the definition and attach changelog notes to the pipeline build.

When things break

Debugging tips

  • When numbers disagree, diff the WHERE clauses behind each query first.
  • Check grain: deduplicate by the primary key before comparing totals.
  • Ask 'who owns this number' before 'what is it' — ownership exposes the caretaker.
  • Replay a KPI for one small period with a hand-written query to catch drift.
Interviews

Interview checkpoint

Practice

Hands-on lab

Write Two Metric Dictionary Entries
20 min · beginner
In progress
Objectives
  • Define revenue and signups accurately
  • Attach grain, owner and target
  • Prove them with a SQL query

Write Two Metric Dictionary Entries

20 min · beginner

0/5 steps
0%
sandbox · Linux sandbox
$
Try these commands
Test yourself

Quick quiz

0/2
1

Which element turns a metric into a KPI?

2

Every number displayed on a dashboard is automatically a metric.

Go deeper

Official documentation

Guide to business metrics (Stripe)
docs
What is a KPI? (KPI.org)
docs