Turn raw numbers into actionable intelligence. Claude reads your data, identifies patterns you might miss, generates the right analytical questions, and produces clear narrative insights — without a data science degree required.
Claude can process and analyze data you share in text format — pasted CSV data, table formats, JSON, or structured text descriptions of datasets. It reasons through patterns, suggests interpretations, and generates both insights and the code to visualize those insights.
📈
Business KPIs
Revenue trends, funnel metrics, retention curves, CAC/LTV analysis — interpret what your numbers mean for the business
📋
Survey Responses
Quantitative and qualitative survey data — synthesize open-text responses, score distributions, demographic breakdowns
Identify outliers, unusual patterns, data quality issues, and unexpected drops or spikes in any metric
🐍
Analysis Code
Generate Python (pandas, matplotlib, seaborn) or R code to reproduce, visualize, or extend your analysis
📖
Data Narrative
Convert raw numbers into the story your leadership needs to hear — plain-English insights with business implications
How to Share Data Effectively with Claude
How you share data significantly affects the quality of analysis. Best practices:
Paste CSV/Table format: Copy data directly from Excel/Google Sheets as tab-separated or comma-separated values. Claude reads tabular data well.
Include headers and units: "Revenue ($)" is clearer than "Column B". Always include column names and measurement units.
Provide context: What is this data? What time period? What business unit? What was the business context when this data was generated?
State your question upfront: Don't just paste data and wait. Tell Claude what business decision this analysis informs.
Upload files directly: If your data is in a CSV or XLSX file, attach it — Claude will process it directly (Pro plan supports file uploads).
Business Metrics Analysis
Business Metrics Analysis Prompt
Analyze this business performance data. I want to understand what's driving the numbers and what to do about it.
My role: [e.g., Head of Growth at a B2B SaaS company]
Decision this informs: [e.g., "Whether to double down on our current acquisition channel or diversify"]
Data (paste your metrics here):
[MONTH] | [METRIC 1] | [METRIC 2] | [METRIC 3] | [METRIC 4]
Jan 2025 | ...
Feb 2025 | ...
[continue for all periods]
Context:
- What changed in the business during this period: [any product launches, campaigns, pricing changes, team changes]
- Known external factors: [seasonality, market events]
- Targets we were working toward: [goals vs actuals]
Analyze:
1. TREND ANALYSIS: What are the dominant trends? What's growing, declining, or flat?
2. ANOMALY FLAGS: Are there any data points that deviate significantly from trend? What might explain them?
3. CORRELATION PATTERNS: Do any metrics move together or against each other?
4. LEADING vs LAGGING: Which metrics are likely predictors of future performance?
5. BUSINESS DIAGNOSIS: What does this data suggest about the health of the business?
6. RECOMMENDED ACTIONS: What should I do differently based on this data?
Format: Start with a 2-sentence "headline" of what this data tells me. Then detailed analysis sections.
Survey & Qualitative Data Analysis
Survey Analysis Prompt
Analyze these survey results. This is [employee engagement / customer satisfaction / product feedback / market research] data.
Survey context:
- N respondents: [total count]
- Collection period: [when conducted]
- Population: [who responded]
- Response rate: [if known]
Quantitative data:
[Paste your scores, rating distributions, ranking data]
Open-text responses (sample):
[Paste a representative sample of verbatim open-text responses — ideally 20-50]
Analyze:
1. QUANTITATIVE SUMMARY: Key scores, distributions, and notable ratings (table format)
2. OPEN-TEXT THEMES: Group qualitative responses into 5-7 themes with frequency counts
3. SENTIMENT ANALYSIS: Overall sentiment breakdown by theme (positive/neutral/negative)
4. STRONGEST SIGNALS: What do respondents feel most strongly about (positive and negative)?
5. SEGMENT DIFFERENCES: Any notable differences by [demographic / role / tenure] if data provided?
6. CORRELATION INSIGHTS: Do low scores on any quantitative items align with specific qualitative themes?
7. PRIORITY MATRIX: Rank issues by: (frequency × intensity of feeling × actionability)
8. RECOMMENDED ACTIONS: Top 3 specific, actionable changes based on this data
Include direct quotes to illustrate each theme. Preserve the respondents' actual language.
Financial Data Interpretation
Financial Statement Analysis Prompt
Analyze these financial statements for [COMPANY/BUSINESS UNIT].
I am [ROLE — e.g., "a small business owner / CFO / investor / first-time user of this data"].
[PASTE FINANCIAL DATA — P&L, Balance Sheet, or Cash Flow]
Provide:
1. HEADLINE HEALTH CHECK: In 2 sentences, is this business financially healthy? What's the most important thing to know?
2. KEY RATIOS: Calculate and interpret [gross margin, operating margin, current ratio, burn rate, runway, etc. — specify which matter to you]
3. TREND ANALYSIS: What's improving, declining, or concerning over the periods shown?
4. RED FLAGS: Are there any numbers that should concern me? What might they indicate?
5. STRENGTHS: What does this data show about what's going well financially?
6. QUESTIONS TO ASK: If I were meeting with leadership, what 5 financial questions should I ask?
7. PLAIN ENGLISH SUMMARY: Explain what this financial picture means in language a non-finance professional can act on.
Note: I am not relying on this as professional financial advice — I'm using this to become a more informed [role].
Generating Analysis Code
For analysts and data scientists, Claude generates production-quality Python or R code to perform the analysis you've described:
Analysis Code Generation Prompt
Write Python code (using pandas and matplotlib/seaborn) to analyze this dataset.
Dataset structure: [describe columns, dtypes, row count]
Analysis goal: [what you want to discover or visualize]
Generate code that:
1. Loads the CSV at [filepath] and does basic data validation (null checks, dtype verification)
2. Calculates: [list the specific metrics or transformations you need]
3. Creates visualizations: [list the charts — e.g., "time series line chart of revenue by region", "correlation heatmap"]
4. Outputs a summary statistics table to console
5. Saves all charts to /output/ directory
Code requirements:
- Pandas + matplotlib/seaborn only (no external packages beyond these)
- Add comments explaining each major step
- Handle the case where data is missing for a date range
- Make chart titles and axis labels clean and presentation-ready
Then: Explain what each visualization shows and how to interpret the output.
Data Narrative & Storytelling
One of Claude's most powerful data use cases is converting analysis into a compelling narrative for non-technical audiences — turning tables of numbers into a story that drives action:
Data Narrative Prompt
Convert this data analysis into a compelling narrative for [AUDIENCE — e.g., "our board of directors / non-technical department heads / investors"].
Raw analysis: [paste your analysis or data]
The audience cares about: [what they prioritize — growth, efficiency, risk, etc.]
The decision this needs to inform: [specific decision or action]
Tone: [e.g., "confident and data-driven but acknowledge uncertainty"]
Write:
1. THE HEADLINE: One sentence — the single most important thing this data tells us
2. THE STORY: A 3-paragraph narrative that flows logically — context → what happened → why it matters
3. THE IMPLICATION: What does this mean for [specific decision/priority the audience cares about]?
4. THE ASK: What specific decision or action do you want from the audience?
Rules: No technical jargon. No raw tables in the narrative (refer to "our conversion rate" not "CR_v2_col_G"). Every number that appears should have a "so what" immediately after it.
Know the Limits
⚠️ Data Analysis Limitations
Large datasets: Claude can process moderate datasets pasted as text but isn't built for million-row datasets. Use proper tools (Python, SQL, Tableau) for big data and bring Claude the summarized results.
Real-time data: Claude has no live data connection. All data must be provided by you in the conversation.
Statistical rigor: Claude can write statistical code and describe methods, but verify statistical claims for academic or high-stakes business use with a qualified data scientist.
Correlation ≠ Causation: Claude will present patterns but can't prove causation. Always ask Claude to distinguish what the data shows from what it might mean.
💡 The Best Data Analysis Workflow
Use Claude for: understanding what analysis to run, interpreting what numbers mean for the business, generating visualization code, and presenting findings to non-technical stakeholders. Use proper tools (Excel, Python, SQL) for the actual number-crunching on large datasets. Claude bridges the gap between raw output and human understanding.