Purpose
Turn raw dashboard numbers into a concise, decision-oriented written narrative that tells stakeholders what changed, why it likely changed, and what to do about it — eliminating the gap between a dashboard full of charts and an executive who needs the story in 90 seconds.
When to Use
Invoke this skill when the user asks to:
- Write an executive summary of a dashboard, KPI report, or metric snapshot
- Turn a table of metrics or a data export into a stakeholder-facing narrative
- Prepare a weekly, monthly, or quarterly business review write-up
- Explain what a set of trends means and what actions to take
- Add a commentary or "insights" section to a recurring report
- Flag anomalies or notable changes in a batch of metrics for leadership
Inputs Required
Before starting, collect:
- Metrics data: The numbers — a table, export, or list of KPIs with current values and (ideally) a comparison period
- Comparison basis: What "prior period" means — week-over-week, month-over-month, vs. target, vs. last year
- Audience: Who reads this — executive team, board, department head, cross-functional stakeholders
- Business context: What the org does and what these metrics drive (revenue, retention, efficiency, etc.)
- Reporting period: The date range this summary covers
- Known events: Launches, campaigns, outages, seasonality, or one-time factors that could explain movement
- Desired length: Short (one paragraph), standard (half page), or detailed (full page)
If comparison basis is missing, default to the most recent equivalent prior period and state the assumption. If no known events are provided, note that drivers are inferred, not confirmed.
Instructions
Step 1: Classify Every Metric by Movement
For each metric provided, tag its change against the comparison basis:
- Direction: Up / Down / Flat
- Magnitude: Compute the percentage change and absolute change
- Materiality: Significant (>10% move OR crosses a target/threshold) vs. Minor (routine fluctuation)
- Directional quality: Good / Bad / Neutral — a rising churn rate is a bad "up"; a falling CPA is a good "down"
Do not treat all movement as equal. A 3% dip in a stable metric is noise; a 3% dip that breaks a 6-month uptrend is a story.
Step 2: Identify the Single Headline
Determine the one thing leadership most needs to know. Choose the metric or pattern with the highest materiality and business consequence. This becomes the opening line — a plain-language takeaway, not a number dump.
- Frame it as a statement of consequence: "Revenue growth stalled this month, driven by a drop in new customer acquisition."
- If multiple stories compete, lead with the one tied most directly to revenue, risk, or a stated goal.
Step 3: Rank Trend Callouts (Top 3–5)
Select the 3–5 most decision-relevant movements. For each callout, write:
- What moved: Metric, direction, and magnitude (% and absolute)
- Why it matters: The business consequence, not just the number
- Likely driver: Tie to known events where provided; otherwise offer the most plausible inference and label it as inference
Order callouts by business impact, not by dashboard layout order.
Step 4: Flag Anomalies and Risks
Call out anything that warrants attention beyond routine trends:
- Metrics crossing a target, SLA, or threshold
- Sudden breaks in an established trend
- Metrics moving in a contradictory direction (e.g., traffic up, conversions down)
- Data quality concerns — implausible values, gaps, or suspicious flatlines Distinguish confirmed anomalies from "worth watching" early signals.
Step 5: Recommend Actions (Ranked by Impact)
Produce 2–4 concrete recommended actions tied directly to the callouts. Each action:
- States what to do, who likely owns it, and the metric it aims to move
- Is ranked by expected impact, not effort
- Avoids generic advice ("monitor closely") in favor of specific next steps ("investigate the 18% drop in mobile checkout conversion — likely a UX or load-time issue")
Step 6: Calibrate Tone to Audience
- Executive/board: Lead with consequence, minimize jargon, quantify everything, keep it tight
- Department head: More operational detail, name specific segments or channels
- Cross-functional: Define metric terms briefly, connect movements to shared goals
Output Format
- Structure: Headline takeaway → Trend callouts → Anomalies & risks → Recommended actions
- Length: Short (~120 words), Standard (~300 words), or Detailed (~500 words) per the requested depth
- Format: Prose narrative for the headline and callouts; bulleted list for anomalies and recommended actions
- Sections: (1) Headline Takeaway — 1–2 sentences; (2) What Changed — 3–5 ranked trend callouts; (3) Anomalies & Risks — flagged items; (4) Recommended Actions — 2–4 ranked next steps
- Tone: Analytical and direct — the voice of a trusted analyst briefing a busy executive; confident about numbers, honest about inference vs. fact
Quality Checklist
Before delivering output, verify:
- [ ] Every material metric change includes both a percentage and an absolute value
- [ ] The headline states a business consequence, not just a number
- [ ] Trend callouts are ordered by business impact — not dashboard layout order
- [ ] Inferred drivers are labeled as inferences, not stated as confirmed fact
- [ ] Each recommended action names a specific next step and the metric it targets
- [ ] Recommended actions are ranked by expected impact
- [ ] Tone and jargon level match the stated audience
Notes
- The most common failure in dashboard summaries is describing what the dashboard already shows. The value is interpretation and recommendation, not restatement — always answer "so what?" and "now what?"
- When known events are absent, resist inventing causes. A well-labeled inference ("possibly seasonal, though the prior year showed no similar dip") is more credible than a confident guess.
- If data lacks a comparison period, say so plainly — a snapshot with no baseline yields observations, not trends, and the summary should not overstate movement.
- For recurring reports, keep the section structure identical period-to-period so stakeholders can scan the same place for the same information each time.
How to use this skill
This file is a Claude Code skill in Anthropic's standard format — markdown with YAML frontmatter at the top.
Installing in Claude Code:
~/.claude/skills/<folder-name>/SKILL.md— auto-loads in every session anywhere on your machine (recommended for skills you'll use across multiple projects)<your-project>/.claude/skills/<folder-name>/SKILL.md— auto-loads only when working in that project (recommended for project-specific skills)
Restart Claude Code or open a new session to register.
Using in Claude.ai, ChatGPT, Gemini, or other AI tools:
- Drag this file into a new chat, or copy and paste its contents
- Send your task or request as your next message — the AI will apply the skill where it fits
For Claude Code skill details: https://docs.anthropic.com/en/docs/claude-code/skills