Data & Spreadsheets skills
Skills that analyze data and build spreadsheets, models and dashboards.
We gave six data skills the same three briefs, all built by the same base model, and each produced a real .xlsx workbook. Every workbook was rendered and scored blind by two independent judges. Because the model is held constant, the differences reflect the skill.
Leaderboard
Happycapy · Analysis that must persuade, it won the financial model and topped the arena overall.
MiniMax · Formatted analysis and dashboards, it won the data analysis and the KPI dashboard.
wshobson · Financial models and unit-economics workbooks.
wshobson · Exec KPI dashboards and metric scorecards.
Anthropic · Dependable, editable spreadsheets across task types.
Jeffallan · Programmatic analysis and data wrangling.
Overall = mean of the per-task overall scores from 2 blind judges.
Every result, side by side
Every skill built the same brief on the same model. Tap any result to open the live page.
SaaS financial model
Winner: data-storytellingA 12-month SaaS financial model workbook, an assumptions sheet, a monthly MRR and customer build with formulas, and a summary (ARR, customers, CAC payback) with a chart.
Polished pricing-tier assumptions, cohort-style MRR build with clear headers, LTV/CAC + payback KPIs and a strong multi-series growth-trajectory chart; most complete and professional.
Strong narrative summary with LTV/CAC and payback KPI callouts plus a clean 12-month MRR chart; polished but the monthly build is comparatively lean.
Clean three-section flow (assumptions, monthly build, summary) with a real growth chart, but the summary insight is thin and some sections feel cramped.
Solid assumptions and FY-total callout cards with a monthly build and MRR chart, but the summary is generic and layout is less refined than C/D.
Covers assumptions, monthly build and exec summary, but layout is cluttered, formatting is busy/inconsistent, and the MRR chart area appears empty/underdeveloped.
Dense wall-of-rows assumptions, weak monthly build, and a broken-looking chart (sharp decay spike); least readable and least trustworthy numbers.
Sales data analysis
Winner: minimax-xlsxAnalyze a provided sales dataset into a workbook, raw data, revenue and units by month and category, top categories, an insights sheet and a chart.
Clean single-flow dashboard with clear monthly/category summaries, top-category shares, tiled insights, and a readable chart; strongest all-around.
Comprehensive: month+category summary, two charts, market-share top categories, detailed color-coded insights; very complete, minor visual density.
Well-organized raw/summary/chart/insights with growth metrics and tidy tagged insight cards; solid and complete if slightly less rich than A.
Moderate coverage with stacked category charts and insights table, but disjointed spacing and middling formatting/depth.
Analytically ambitious (unit economics, cohorts) but layout is broken with huge whitespace and garbled/empty charts, hurting readability badly.
Sparse: raw data, one small table, one chart, a best-month mini-table; large empty regions and missing category depth make it weakest.
KPI dashboard
Winner: minimax-xlsxA KPI dashboard workbook for a subscription business, a metrics data sheet and a one-screen dashboard tab with KPIs, month-over-month deltas, conditional formatting and charts.
Dense, multi-block dashboard: KPI cards, MoM+6-mo change tables, computed KPI ratios, status column, MRR bar + trend charts; most complete and analytical.
Attractive, well-spaced dashboard with clear KPI cards, deltas, conditional-formatted table, a written insight callout and two clean charts.
Cohort/finance depth (CAC, LTV, burn, runway), scorecard with GOOD/OK signals and two clear charts; strongest analytics, slightly heavy for one screen.
Clean exec dashboard with KPI cards, MoM deltas, status flags, two charts and a data sheet; Notes tab is filler and insight is thin.
Readable KPI panel and two charts, but a 6-month summary row mixes rates oddly and the second chart area is nearly empty; feels unfinished.
Three charts plus KPI strip, but raw data is duplicated in redundant side tables and MRR values are inconsistent across sheets.
How we scored data & spreadsheets
Two independent blind judges scored every output 1 to 10 on each of these axes, then we averaged across judges and tasks. These axes are specific to data & spreadsheets, while other arenas use their own rubric.
- Structure
- Sheet organization and logical layout.
- Formatting
- Readability, headers, number formats and color.
- Analysis
- Depth and usefulness of the summary and insight.
- Correctness
- Plausible, internally consistent numbers and clear formulas.
- Completeness
- Covers every sheet and element the brief asked for.
Limitations. This arena is a single run on Claude Sonnet 5 with 3 tasks, scored by AI judges rather than crowd votes. Treat the results as directional evidence, not a definitive ranking; re-test on your own workload before committing.
How each skill performed, and who it’s for
data-storytelling
Turns data into a clear narrative: headline findings, annotated summaries and an exec-ready story.
Best for: Analysis that must persuade, it won the financial model and topped the arena overall.
Watch out: Narrative-first; lighter on heavy spreadsheet mechanics and formulas.
minimax-xlsx
Generates polished, well-formatted .xlsx workbooks with clean layouts and charts.
Best for: Formatted analysis and dashboards, it won the data analysis and the KPI dashboard.
Watch out: A workbook builder, not a modeling methodology.
startup-financial-modeling
Startup financial-modeling methodology: assumptions, unit economics, LTV/CAC and payback.
Best for: Financial models and unit-economics workbooks.
Watch out: Strong on models, weaker on general data analysis or dashboards.
kpi-dashboard-design
KPI dashboard design guidance: metric selection, deltas, signals and layout.
Best for: Exec KPI dashboards and metric scorecards.
Watch out: Design-guidance oriented; layouts can get cramped one-screen.
xlsx
Anthropic's official Excel skill: builds .xlsx via openpyxl with formulas and formatting.
Best for: Dependable, editable spreadsheets across task types.
Watch out: Produced plainer workbooks here, less standout formatting and insight than the top skills.
pandas-pro
pandas-centric data analysis: dataframes, aggregations and transforms.
Best for: Programmatic analysis and data wrangling.
Watch out: Analysis-first; its workbooks were the least formatted and complete here (lowest score).
The tasks, head-to-head
How each skill scored on the three briefs, and how quality trades off against token cost.
| Skill | SaaS financial model | Sales data analysis | KPI dashboard | Avg |
|---|---|---|---|---|
| data-storytelling | 9.0 | 8.0 | 7.5 | 8.2 |
| minimax-xlsx | 3.5 | 9.0 | 8.0 | 6.8 |
| startup-financial-modeling | 8.5 | 4.0 | 7.5 | 6.7 |
| kpi-dashboard-design | 7.5 | 5.5 | 6.5 | 6.5 |
| xlsx | 5.5 | 7.5 | 5.0 | 6.0 |
| pandas-pro | 6.0 | 3.5 | 4.5 | 4.7 |
Cell = the skill’s score on that task (darker = higher). A ring marks the task winner.
Token cost & efficiency
Every skill ran on the same base model, so token use reflects the skill itself: how much it writes (generated) and how large its SKILL.md is (context processed). Summed across the 3 tasks; lower is cheaper.
Context includes the skill’s SKILL.md re-read into the model each turn, so a larger skill file drives up cost, while the no-skill baseline is cheapest.
What we learned
- data-storytelling won overall (8.17) and the financial model, framing numbers as a clear narrative with headline findings beat raw spreadsheet mechanics.
- minimax-xlsx won the data analysis and the KPI dashboard, it produces the cleanest, best-formatted workbooks with charts.
- Specialist methodology skills split by task: startup-financial-modeling is strong on models, kpi-dashboard-design on scorecards, but neither generalizes across all three.
- Anthropic’s xlsx is a dependable middle, correct and editable but plainer; pandas-pro analyzed well yet shipped the least polished, least complete workbooks (last place).
- For data work, presentation is a real differentiator: the skills that formatted and narrated the output beat the ones that just computed it.
Frequently asked questions
What is the best AI skill for spreadsheets and data analysis?+
In our blind test data-storytelling (Happycapy) scored highest overall (8.17/10) by turning the numbers into a clear narrative, while minimax-xlsx produced the best-formatted workbooks and won the data-analysis and KPI-dashboard tasks. The best pick depends on whether you want a persuasive analysis or a polished workbook.
Which AI skill is best for financial models, dashboards, or data analysis?+
In our tests data-storytelling won the SaaS financial model, and minimax-xlsx won both the sales-data analysis and the KPI dashboard. Specialist skills helped in their lane (startup-financial-modeling for models, kpi-dashboard-design for scorecards) but did not generalize across all three.
Do data skills produce real editable Excel files?+
Yes, every workbook in this arena is a real .xlsx built with openpyxl (formulas, formatting and charts). Anthropic’s xlsx and minimax-xlsx are the most spreadsheet-native; others (data-storytelling, the modeling and dashboard skills) shape the content and still emit a workbook.
How is the Data & Spreadsheets arena scored, and is it fair?+
Every skill builds the same three briefs with the same base model, so results isolate the skill, not the model. Each workbook is rendered, anonymized (A to F), and scored 1 to 10 by two independent blind judges on structure, formatting, analysis, correctness and completeness, then averaged across judges and tasks.
Where can I find and install these data skills?+
All are openly available: data-storytelling (Happycapy), minimax-xlsx (MiniMax), xlsx (Anthropic), startup-financial-modeling and kpi-dashboard-design (wshobson), and pandas-pro (Jeffallan). On Happycapy you can install any of them and run them in your browser with no setup.
Build with these skills, no setup required
Every skill in this data & spreadsheets arena installs on Happycapy in one click and runs in your browser on the same model. Start free, or explore 2M+ skills in the Skill Store.
