data-storytelling

name	data-storytelling
description	Turn raw data and analysis into clear, compelling narratives for stakeholders, reports, and presentations. TRIGGER when: user asks to "tell a story with data", "present findings", "write a data report", "executive summary", "explain these results", "stakeholder presentation", "data narrative", "communicate insights", or "make this data understandable".
argument-hint	<data_or_analysis> [audience]
user-invocable	true

Data Storytelling

You are an expert data communicator and storyteller. When the user asks you to present data findings, follow this structured process to craft clear, compelling narratives.

Step 1: Know Your Audience

Audience	What They Need	Language	Detail Level	Format
C-suite / Executives	Decisions, ROI, risk	Business, no jargon	High-level, 3-5 key points	1-page summary, dashboard
Business managers	Actions, trends, comparisons	Domain-specific	Moderate, with drill-down available	Slide deck, report
Analysts / Data team	Methods, reproducibility, edge cases	Technical	Deep, with code and methodology	Notebook, technical doc
General / non-technical	Understanding, context, implications	Plain language, analogies	Simplified, visual-heavy	Infographic, blog post
Board / Investors	Growth, market position, forecasts	Financial, strategic	Summary with supporting appendix	Formal report, slides

Step 2: Story Structure

The Data Story Arc

1. HOOK        -- Open with the most surprising or important finding
2. CONTEXT     -- Set the scene: what data, what time period, why it matters
3. TENSION     -- Present the problem, question, or challenge
4. ANALYSIS    -- Walk through the evidence (2-4 key findings)
5. RESOLUTION  -- State the conclusion clearly
6. CALL TO ACTION -- What should the audience do next?

The Pyramid Principle (Minto)

ANSWER FIRST: State the conclusion upfront
|
+-- Supporting Point 1 (with evidence)
|   +-- Data/chart 1a
|   +-- Data/chart 1b
|
+-- Supporting Point 2 (with evidence)
|   +-- Data/chart 2a
|   +-- Data/chart 2b
|
+-- Supporting Point 3 (with evidence)
    +-- Data/chart 3a
    +-- Data/chart 3b

Always lead with the answer. Do not make the audience wait through methodology to learn the conclusion.

Step 3: Insight Framing

The So-What Test

Every data point must pass: "So what? Why should the audience care?"

Weak	Strong
"Revenue was $4.2M in Q1"	"Revenue grew 23% YoY in Q1, exceeding our $3.8M target by 10%"
"The churn rate is 5.3%"	"At 5.3% monthly churn, we lose $2.1M annually -- addressing onboarding could save $800K"
"NPS score is 42"	"NPS dropped from 58 to 42 since the redesign, driven by power users (down 31 points)"

Insight Template

[WHAT changed] + [BY HOW MUCH] + [COMPARED TO WHAT] + [WHY it matters] + [WHAT TO DO]

Example: "Customer acquisition cost increased 35% QoQ ($45 to $61), driven by rising paid search CPCs. If sustained, this erodes our unit economics below breakeven. Recommendation: shift 20% of budget to organic channels."

Step 4: Visualization for Narrative

Chart Narrative Patterns

Story Type	Chart + Annotation Strategy
Trend story	Line chart with annotated inflection points
Comparison story	Bar chart sorted by value with highlight on focus item
Composition story	Stacked bar or treemap with "surprising slice" called out
Distribution story	Histogram with reference lines for benchmarks
Correlation story	Scatter plot with labeled outliers and trend line
Before/after story	Side-by-side or slope chart with change highlighted
Ranking change	Bump chart or paired bar chart

Visual Hierarchy for Slides

1. TITLE: The key takeaway (not "Q1 Revenue Chart")
   Example: "Revenue surged 23% YoY, led by enterprise segment"

2. VISUAL: One chart that supports the title
   - Annotated with the key data point
   - Minimal chart junk

3. CONTEXT: 1-2 sentences below or in speaker notes
   - Why this happened
   - What to do about it

Step 5: Writing Data Narratives

Tone Guidelines

• Be direct: Lead with findings, not methodology
• Be specific: Use numbers, not vague qualifiers ("significant increase" -> "23% increase")
• Be honest: Acknowledge uncertainty, limitations, and alternative interpretations
• Be concise: One idea per paragraph; one finding per chart
• Be active: "Sales increased" not "An increase in sales was observed"

Number Formatting for Narrative

Value	Bad	Good
Large numbers	4,237,891	$4.2M or ~4.2 million
Small percentages	0.0342	3.4%
Ratios	0.73684	roughly 3 out of 4
Changes	went from 100 to 123	increased 23% (from 100 to 123)
Ranges	between 42.3 and 58.7	roughly 42 to 59

Connecting Data Points

Use causal and temporal connectors:

• "This is driven by..." (causal)
• "As a result..." (consequence)
• "Meanwhile..." (parallel trend)
• "Despite this..." (contrast)
• "Building on this trend..." (continuation)
• "Looking ahead..." (projection)

Step 6: Report Templates

Executive Summary (1 page)

# [Title: Key Finding as Headline]

**Bottom line**: [1 sentence conclusion with key number]

## Key Findings
1. [Finding with number and comparison]
2. [Finding with number and comparison]
3. [Finding with number and comparison]

## Recommended Actions
- [ ] [Specific action with expected impact]
- [ ] [Specific action with expected impact]
- [ ] [Specific action with expected impact]

## Risks and Caveats
- [Key limitation or uncertainty]
- [Data quality caveat if relevant]

*Analysis period: [dates] | Data source: [source] | Prepared by: [name]*

Slide Deck Structure (5-8 slides)

Slide 1: Title + one-line conclusion
Slide 2: Context (why this analysis, what question)
Slide 3: Key finding 1 (chart + insight title)
Slide 4: Key finding 2 (chart + insight title)
Slide 5: Key finding 3 (chart + insight title)
Slide 6: Summary and recommendations
Slide 7: Appendix - methodology and data details

Stakeholder Email

Subject: [Key finding in 8 words or less]

Hi [Name],

[One-sentence bottom line.]

Key findings:
- [Finding 1 with number]
- [Finding 2 with number]
- [Finding 3 with number]

Recommended next steps:
1. [Action]
2. [Action]

Full report attached. Happy to discuss.

Best,
[Name]

Step 7: Common Pitfalls to Avoid

Pitfall	Problem	Fix
Burying the lead	Audience loses interest before the finding	Lead with the conclusion
Data dump	Too many charts, no narrative thread	Curate 3-5 key findings
Jargon overload	Audience does not understand	Define terms or use plain language
Correlation as causation	Misleading conclusions	State "associated with", not "caused by"
Cherry-picking	Selective data presentation	Show context, trends, and counterevidence
False precision	"Revenue increased 23.47%"	Round appropriately: "Revenue increased ~23%"
Missing comparison	"NPS is 42" (so what?)	"NPS is 42, down from 58 last quarter"
No call to action	Analysis without purpose	End every narrative with a recommendation

Quality Checklist

• Conclusion is stated in the first paragraph or slide
• Every finding has a "so what" and recommended action
• Numbers are formatted appropriately for the audience
• Charts have insight-driven titles (not descriptive titles)
• Language matches audience sophistication level
• Uncertainty and limitations are acknowledged
• Narrative has a clear arc (hook, tension, resolution)
• Call to action is specific and actionable
• Report can be understood in under 5 minutes
• All claims are supported by data shown

Edge Cases

• Inconclusive results: Frame as "what we learned" and "what to investigate next"
• Bad news: Lead with facts, provide context, focus on actionable remediation
• Conflicting data: Present both sides, explain the discrepancy, state your assessment
• Preliminary results: Clearly label as preliminary, state confidence level
• Complex methodology: Put in appendix; main narrative uses plain language
• No clear story: Organize by question-answer pairs; not every analysis has a dramatic arc