Business
KPI Definitions Doc AI Prompt Template
Write a KPI definitions document with this AI prompt. One source of truth so every team computes metrics identically. No sign-up needed. Just copy and run.
The prompt
ROLE: You are a data governance lead who ends metric arguments with documentation. CONTEXT: Company: [COMPANY]. Metrics to define: [METRICS]. Data sources: [SOURCES]. Known definition disputes: [DISPUTES]. Audience: [AUDIENCE]. TASK: Write a KPI definitions document (metric dictionary). FORMAT: For each metric: 1. Name and one-line plain-English definition, 2. Formula (exact calculation), 3. Data source(s) and table/field if known, 4. Filters and exclusions (what is NOT counted), 5. Refresh cadence, 6. Owner, 7. Example calculation with sample numbers, 8. Common misinterpretations. Then: a Change Log process (how definitions get updated and communicated). End with the 3 most dangerous definition disputes to resolve first. CONSTRAINTS: Definitions must be computable by two analysts independently with identical results. Plain English first, formula second. Every exclusion must be justified; hidden exclusions are where metric gaming lives. If any variable is unclear or missing, ask up to 3 targeted clarifying questions before producing the output, and clearly mark every assumption you make with [ASSUMPTION].
Variables to fill in
Replace each [variable] in the prompt with your own details:
[COMPANY] [METRICS] [SOURCES] [DISPUTES] [AUDIENCE]
Example input
COMPANY=Finlytics, METRICS=MRR, churn rate, NRR, CAC, active user; SOURCES=Stripe, Salesforce, product DB; DISPUTES=churn: logo vs revenue, MRR: includes annual prepaid?; AUDIENCE=execs, finance, data team
How to use this prompt
Resolve the known disputes in a working session first, document the decisions here, and require sign-off from metric owners.
Tips for the best results
- The example calculation is the most valuable field; abstract formulas get misread, worked examples do not.
- Version every definition change; silent metric changes rewrite history and destroy trust.
- Review disputed metrics quarterly; business changes invalidate old definitions.
Frequently asked questions
Who should own metric definitions?
The business owner of the metric (e.g., CRO owns pipeline definitions), with the data team as steward of computation.
How detailed should definitions be?
Detailed enough that a new analyst reproduces the number exactly. When in doubt, add the worked example.