skills / data

data-storytelling

what it does

Transform data into compelling narratives using visualization, context, and persuasive structure. Use when presenting analytics to stakeholders, creating data reports, or building executive presentations.

Verified Before and after: 2 models 1 try per case View source Download .zip

Compare the answers

Answers from Claude Sonnet 5.5 and GPT-6.1 Sol · try 1 of 1

Same request, same model, once with the skill and once without. Read both and judge for yourself.

“My slide titles are boring. Can you rewrite them so each one says what the slide shows?”

Source material given with the request

Slide 1 title: "Website Traffic Q3" Data: sessions 182,000 in Q2 -> 241,000 in Q3 Slide 2 title: "Conversion Rates" Data: trial-to-paid 11.8% in Q2 -> 9.2% in Q3 Slide 3 title: "Support Metrics" Data: CSAT 91% -> 93%; median first response 3.1 h -> 1.4 h Slide 4 title: "Pricing Test" Data: annual-plan uptake 18% (control) vs 27% (2 months free on annual)
with the skill
# Rewritten slide titles | # | Old title | New title | |---|-----------|-----------| | 1 | Website Traffic Q3 | **Q3 Traffic Jumped 32%, from 182K to 241K Sessions** | | 2 | Conversion Rates | **Trial-to-Paid Conversion Fell from 11.8% to 9.2% in Q3** | | 3 | Support Metrics | **Support Got Faster and Better: First Response Time Cut by More Than Half, CSAT Up to 93%** | | 4 | Pricing Test | **Offering 2 Months Free on Annual Plans Lifted Annual Uptake from 18% to 27%** | ## Shorter alternatives (if titles wrap) 1. **Q3 Traffic Up 32% to 241K Sessions** 2. **Trial-to-Paid Conversion Dropped 2.6 Points in Q3** 3. **Response Time Down 55%, CSAT Up to 93%** 4. **"2 Months Free" Boosts Annual Uptake by 9 Points** ## Math behind the numbers - Traffic: +59,000 sessions, which is +32% (241 / 182 = 1.324) - Conversion: -2.6 points, or about -22% relative - First response: 3.1 h to 1.4 h, which is -55% - Annual uptake: +9 points, or +50% relative
Show the rest (6 more lines)
## Storytelling suggestions - **Slides 1 and 2 are your tension.** Traffic is up while conversion is down. You could make that explicit by retitling slide 2 as *"But Conversion Fell from 11.8% to 9.2% as Traffic Grew."* Avoid claiming the traffic caused the drop unless you've checked, for example by looking at whether the new visitors came from lower-intent channels. - **Slide 4 can serve as your resolution.** It reads as a possible lever for revenue per customer. A title like *"One Lever to Test: 2 Months Free Lifts Annual Uptake from 18% to 27%"* ties it to the narrative. - **Check slide 4 before presenting.** A bold title like this invites the question "is it significant?" Be ready with the sample size and test duration. Also consider whether the discount's cost offsets the uptake gain. If you don't have that data, soften the title to "…Lifted Annual Uptake in Test." If you tell me your audience (execs, board, team), I can adjust the tone and the level of detail in the titles.
without
Here are the rewrites: **Slide 1:** Q3 sessions grew 32% to 241,000 *(was "Website Traffic Q3")* **Slide 2:** Trial-to-paid conversion fell from 11.8% to 9.2% in Q3 *(was "Conversion Rates")* **Slide 3:** Support got faster and better: first response time more than halved while CSAT rose to 93% *(was "Support Metrics")* **Slide 4:** Offering 2 months free on annual plans lifted annual uptake from 18% to 27% *(was "Pricing Test")* Notes: - **Slide 1 math:** 182,000 to 241,000 is +32.4%. - **Slide 2 alternative:** "Conversion fell 2.6 points" is punchier, but the before and after numbers are more concrete. A 22% relative drop is another option, though it can sound more alarming than the point change. - **Slide 3:** Median first response went from 3.1 h to 1.4 h, a 55% reduction. I avoided saying faster responses *caused* the CSAT gain, since the data doesn't show that. - **Slide 4:** "Lifted" assumes the test was properly randomized. If it wasn't, use "was associated with" or "coincided with." - **Slides 1 and 2 together:** Traffic grew while conversion dropped, which may be worth addressing. For example, Slide 2 could read "Conversion fell to 9.2% as traffic growth brought in less-qualified visitors," but only if you have evidence for that.

Answers are shown exactly as the model wrote them, from the first try, and nothing is picked by hand.

The models

2 models

Claude Sonnet 5.5

made by Anthropic

claude-sonnet-5-5 · revision not reported by the provider

1 try per case · tested 3 Nov 2026

GPT-6.1 Sol

made by OpenAI

gpt-6.1-sol · revision not reported by the provider

1 try per case · tested 3 Nov 2026

The tasks

written by Carte Blanche
allhands-nps-not-comparable

“Draft the three slides for Friday's all-hands about our jump in customer satisfaction.”

board-story-opt-in-feature

“I'm presenting this to the board on Thursday. Can you make the story more compelling?”

ceo-update-segment-split

“Turn this into a short update for our CEO.”

chart-for-channel-mix

“I need one chart for the board deck showing how our signup mix shifted this year. What chart should I make and how should I set it up?”

slide-titles-takeaways

“My slide titles are boring. Can you rewrite them so each one says what the slide shows?”

Install

Download the .zip and unzip it. Put its files, in a folder named data-storytelling, here:

Claude Code~/.claude/skills/Every project. For one project only, use .claude/skills/ inside it.
Codex~/.agents/skills/Every project. For one repository only, use .agents/skills/ inside it.
Cursor~/.cursor/skills/Cursor also reads the Claude Code and Codex folders above.

Then start a new session and ask for what the skill does. Other agents that read SKILL.md have their own skills folder; check their docs.

Coming after launch: get skills from inside your agent, no download needed.