KPI Narrator
Auto-explain what likely moved your metrics this week
Feeds on your KPI data and writes the narrative your stakeholders need — what changed, which segments likely drove it, and which movements look unusual. Turns numbers into a review-ready draft instead of a blank page.
INGREDIENTS
PROMPT
Create a skill called "KPI Narrator". I'll give you KPI data (as a table or SQL query results) for the current period and the comparison period. Generate a narrative summary that: (1) States each KPI's current value and period-over-period change. (2) Breaks the change down by key dimensions (segment, region, product) to identify likely drivers visible in the data. (3) Flags changes that appear statistically significant or anomalous based on the history available. (4) Calls out results that should be reviewed by a human before they are presented as explanations. (5) Uses business language, not statistical jargon ("Enterprise revenue drove most of the growth" not "the Enterprise segment coefficient was 1.12"). If I provide a sample narrative from a previous report, match the tone and structure. Keep narratives concise — 2-3 sentences per KPI, not paragraphs.
How It Works
Executives don't want a table of numbers — they want to know what changed,
what looks unusual, and which dimensions likely drove the movement. Writing
those narratives is time-consuming and repetitive. This skill generates a
draft narrative from your KPI data so you start from something useful instead
of a blank page.
What You Get
- Period-over-period comparison for each KPI
- Narrative summary explaining what changed and the most likely drivers visible in the data
- Statistical significance or anomaly flagging, depending on the inputs available
- Segment-level drill-down ("Enterprise was up, SMB was down")
- Consistent language aligned with previous reports
- Callouts for results that need a human review before sharing
Setup Steps
- Ask your Claw to create a "KPI Narrator" skill with the prompt below
- Provide your KPI definitions and the queries that produce them
- Optionally provide a sample narrative so it matches your style
- Run after each data refresh to get an auto-generated summary draft
Tips
- These narratives are drafts and hypotheses, not proof of causality
- The significance flagging helps prevent overreacting to tiny movements
- Feed it the same KPIs weekly and it maintains consistency across reports
- Pairs well with Report Autopilot for end-to-end reporting workflows