Skill Locker
All skills
v1.3.03 loop iterations

Data Storyteller

Data analysts find truth. Data storytellers make truth matter. The Data Storyteller takes an analysis nobody would finish reading and builds a narrative with one core message, three key numbers, and a recommendation that flows naturally from the data instead of arriving in a separate slide.

What this skill does

The graveyard of business is full of brilliant analyses that changed nothing because they were delivered as tables and bullets to executives who checked their phones. This skill takes the analysis you've already done and structures it as a narrative — situation, complication, action — that carries the insight to the moment the audience has to decide. The chart is the evidence. The narrative is the argument.

The SCA frame runs the structure. Situation establishes the baseline everyone agrees on, anchored by a single number. Complication introduces the tension — what changed, what surprised, what broke — and is where the data insight lives, usually framed as expected versus actual. Action resolves the tension with a recommendation that flows naturally from the complication rather than appearing in an unrelated final slide. If your data doesn't fit SCA, it usually means you've got a report, not a story, and the skill will tell you so.

Three numbers, not thirty. The Rule of Three Numbers is enforced because every additional number dilutes the ones you already had. Each survivor gets the RAMP test — Relatable (translate £4.2M of waste into "the salary of 42 employees"), Anchored (every number has a comparison point), Memorable ("nearly 3 out of 4" sticks where "73.2%" does not), and Pointed (does this number advance the story or is it ego). Numbers that don't pass get cut, no matter how hard you worked to calculate them.

Delivery format changes the shape of the narrative. Live presentation gets one finding per slide with insight-as-title headlines and chart callouts on the moment that matters. Written memo leads with the ask in the first sentence and runs the full SCA arc in under 500 words. Verbal/meeting versions come in two lengths — the 30-second hallway version with three numbers and the 2-minute full-arc version for when they want more. Dashboard annotations are one sentence per metric, no arc. The skill writes the version that fits the channel.

You get the story brief (audience, core message, ask, leave-behind number), the full narrative with hook, context, insight, evidence, implication, recommendation, ask, and a supporting notes block with anticipated pushback and backup data. If asked to "make bad data look good," the skill refuses to spin — it reframes constructively (problem + credible recovery plan) but never misrepresents. Honesty survives the next board meeting; spin does not.

When this triggers

  • ·You've done the analysis and now have to present it to a board, executive, or client who won't read tables
  • ·Your last data presentation got 'interesting, thanks' and zero decisions
  • ·You need to brief someone in five minutes and the spreadsheet is fifteen tabs
  • ·You're writing a memo or update where the numbers must do the persuading
  • ·Findings are mixed or negative and you need to present them honestly without losing the room

Example

Trigger

User: 'I need to present this to our board: we hit our user target (12,400 vs 10,000) but 73% churned within 14 days. Help me structure it.'

Output

Story Brief Audience: Board (5 people, 10 minutes) Core message: We won the wrong race — acquisition target hit, retention reality bad enough that net new users were negative. Ask: Approve £85K reallocation from paid acquisition into onboarding fix for next quarter. Leave-behind number: "We added 12,400 and kept 3,400." Narrative (SCA structure) Hook "We hit 124% of our acquisition target last quarter and the user base shrank. Here's why." Context Q3 target was 10,000 new users at <£18 CAC. Board signed off £180K of paid spend on that basis. Insight Three numbers tell the story: · 12,400 new signups — 24% above target · 73% churned in the first 14 days · Net retained users: 3,400 — well below the 8,000 implied by the original plan Evidence [Cohort retention curve, with day-14 cliff annotated. Single callout: "Drop-off concentrated in onboarding step 3."] Implication Continuing the current playbook spends £180K next quarter to add ~3,000 net users at an effective CAC of £60 — three times the planned £18. Recommendation Pause £85K of paid spend. Redirect to fix the step-3 onboarding drop. Conservative model: retention to 50% lifts net users to 6,200 at the same total spend. Ask Approval to reallocate £85K within Q4 budget. Anticipated questions · "Is the data clean?" — Yes, 14-day cohort, n=12,400 · "Why step 3 specifically?" — Drop-off rate jumps 3x there · "What if the fix doesn't work?" — Trigger to reverse: if day-14 retention isn't above 40% by week 4, we revert spend

Get this skill + 8 more

Get the full Data & Analysis pillar (9 skills) or the complete library.

Get the full stack — $299

What you get

  • 170-line SKILL.md, ready to drop into ~/.claude/skills/
  • Tested through 3 Karpathy-loop iterations (versions v1.0.0 → v1.3.0)
  • Triggers automatically when relevant — no command to remember
  • Lifetime updates as the skill is refined further

More from Data & Analysis

Browse the full library

297 skills across 31 categories. One purchase, lifetime updates.

See all bundles