About this skill

Problem

When CSV or Excel data is passed to an LLM, analyses often rely on column names, estimate metrics without checks, treat correlation as causation, skip missing-value or anomaly handling, and fail to provide a practical chart plan. This skill turns data analysis into an auditable workflow instead of jumping straight to conclusions.

How It Works

Data understanding: confirm row count, column meanings, field types, time granularity, and the business question.
Quality checks: profile missing values, duplicates, and anomalies, then recommend handling options.
Descriptive analysis: calculate totals, averages, medians, max/min values, growth rates, distributions, and preliminary correlations.
Insight extraction: produce 3–5 key findings, labeled hypotheses, actionable recommendations, and risk notes using a data → insight → recommendation chain.
Visualization plan: choose line, bar, pie, histogram, or scatter charts for trends, comparisons, shares, distributions, and relationships, with usage and variables noted.

Use code tools such as pandas for complex calculations; correlation should not be treated as causation.

Boundaries

Best for structured tabular data with a clear business question. Be cautious with small samples, unmasked sensitive fields, strict causal claims, or high-compliance data before relying on the conclusions.

Screenshots

Use Cases

After receiving sales CSV data, profile missing values and anomalies, summarize growth, averages, and channel share for review.
When product event tables are pasted in, split usage by date, spot variable relationships, and suggest line or bar chart plans.
After exporting multiple Excel files, align field definitions, calculate conversion and retention metrics, and produce a risk-noted report.
When explaining early patterns to business users, summarize 3–5 findings, hypotheses, and follow-up validation suggestions.

Best For

Operations analysts turning sales or ops Excel files into conclusions and chart recommendations
Product analysts processing event CSVs to identify usage trends and variable relationships
Business data consultants reviewing data quality and explaining metric definitions before delivery
Consulting analysts preparing management reports with risk notes and hypotheses to validate