metric-definition

Use when an analysis depends on a KPI, metric, ratio, column label, or business definition that could be interpreted in multiple valid ways across datasets, dashboards, or reports.

Metric Definition

Purpose

Force precise metric definitions before interpretation, reporting, or comparison.

When to use

Use this skill when:

  • a metric name could mean several things
  • a file, chart, or dashboard contains labels like revenue, usage, activation, churn, or cost that are not yet formally defined
  • the user asks for analysis of a dataset but the main KPI could be interpreted in multiple valid ways
  • teams are debating a KPI
  • an answer depends on denominator choice
  • a dashboard label is too vague to trust

When not to use

Do not use this skill when:

  • the metric has an agreed canonical definition already documented and confirmed
  • the task is pure arithmetic on an already defined metric

Required thinking discipline

  • Never treat a metric label as a definition.
  • Identify numerator and denominator clearly.
  • Specify the entity, grain, timeframe, and filters.
  • Surface definition risks that could change interpretation.
  • Evidence constraint: Every conclusion must cite specific data — a number, a rate, a segment, or a timeframe. Do not speculate without evidential basis. If data is insufficient, state what is missing rather than asserting an unsupported inference.

Workflow

  1. Restate the metric name.
  2. Define what is counted.
  3. Define what is excluded.
  4. Define denominator logic if the metric is a rate or ratio.
  5. Specify grain, timeframe, and segment rules.
  6. Note data source or implementation assumptions.
  7. Produce a metric spec before any narrative interpretation.

Output format

  • Metric name
  • Business intent
  • Formal definition
  • Numerator
  • Denominator
  • Entity and grain
  • Timeframe logic
  • Filters and exclusions
  • Known ambiguities

Good example

Metric:

Activation rate

Good response:

Activation rate = share of new signups who complete project creation within 7 days of signup.
Numerator: distinct new users who create at least one project within 7 days.
Denominator: distinct new signups in the cohort.
Exclude internal accounts and spam-flagged users.

Bad example

Activation rate is how many users become activated.

Why this is bad:

  • numerator is undefined
  • denominator is missing
  • time window is missing
  • exclusions are unclear

Practical notes

  • If a metric is controversial, show 2 possible valid definitions and explain the consequence.
  • For executive readouts, include one plain-English definition plus one formal definition.
  • In generic data-analysis asks, use this skill whenever a column name or KPI label could silently change the conclusion.

Optional variants

  • Growth metrics: be strict about attribution windows and exclusion rules.
  • Finance metrics: be strict about recognition timing and currency handling.
  • Operations metrics: clarify service level thresholds and exception handling.