Level 2Lesson 14⏱️ 75 min

AI + Data Analysis

Read spreadsheets, spot trends, and turn numbers into insights—without being a data scientist

TeacherManagerDeveloperAnalystBusinessDoctorLawyer

1. You Don't Need to Be a Data Scientist

Most professionals say they "can't do data analysis." What they really mean: they don't know SQL, they haven't memorized statistics formulas, and they're intimidated by spreadsheets.

But here's the truth: Claude can do the technical work. Your job is to ask the right questions and interpret the answers.

Paste data into Claude. Ask what you want to know. Claude will find patterns, calculate trends, spot outliers, and explain what it means in plain English.

What's Possible

  • Spot trends: Growth month-over-month, seasonal patterns, anomalies
  • Summarize tables: Turn 500 rows into a 5-bullet insight
  • Calculate stats: Average, median, percentiles, growth rates, percentages
  • Find outliers: Which items are performing best or worst?
  • Write formulas: Excel, Google Sheets—Claude can build them
  • Compare periods: This quarter vs. last, this year vs. last year
  • Interpret dashboards: Paste a screenshot, ask "what should I worry about?"
⚠️
Critical Limitation: Claude cannot access external files, URLs, or databases directly. You must copy and paste the data into the conversation. For large files, use Claude's built-in file upload feature, or copy-paste the relevant rows. This is a security feature—your data stays in your conversation.

2. Working with Spreadsheets and CSVs

The workflow is simple: select data → copy → paste into Claude → ask a question.

Here's a real example. Paste this CSV data into Claude:

Sample CSV Data (Sales by Region, 2024)
Month,Region,Revenue,Units_Sold,Customer_Count,Avg_Order_Value Jan,North,145000,1200,342,121 Jan,South,98000,850,218,115 Jan,East,167000,1400,401,119 Jan,West,112000,920,256,122 Feb,North,156000,1280,358,122 Feb,South,105000,900,235,117 Feb,East,178000,1480,415,120 Feb,West,119000,960,268,124 Mar,North,198000,1620,445,122 Mar,South,142000,1200,310,118 Mar,East,215000,1780,485,121 Mar,West,156000,1280,350,122 Apr,North,172000,1420,395,121 Apr,South,128000,1080,290,119 Apr,East,189000,1550,430,122 Apr,West,134000,1100,300,122

Once you paste this, here are four different analyses you can run on the same dataset:

Analysis 1: Trends

Look at this sales data. Which region is growing fastest, and which is struggling? Show the month-over-month growth rate for each region.

Analysis 2: Outliers

Which region-month combination is an outlier? Which is performing best and worst? Explain why each might be an outlier.

Analysis 3: Summary Stats

For each region, calculate: total revenue, average order value, total units sold, and customer count. Which region is the biggest profit driver?

Analysis 4: Recommendations

Based on this data, where should we invest more sales effort next quarter? Which region is at risk? What's one action we should take?
💡
Copy from Excel or Google Sheets: Select the cells you want → Ctrl+C (or Cmd+C on Mac) → Paste into Claude. Claude will recognize the CSV format and work with it. No need to export to a file first.

Learn CSV basics: Microsoft Office CSV import/export guide

3. Asking the Right Data Questions

Bad prompt: "Analyze this data."

Good prompt: "In this dataset, which month had the highest growth rate? Show your calculation. What might explain this growth?"

Use the DATA framework:

D = Describe the data

What does each column represent? What time period? How many rows?

A = Ask a specific question

Don't say "analyze it." Say "Which product had the highest growth rate?" or "Is there a seasonal pattern?"

T = Tell Claude the output format

"Show results as a table" or "Give me a bulleted summary" or "Write an executive summary"

A = Ask for the reasoning

"Show your work" or "Explain the calculation" so you understand how Claude got the answer

Here are 5 before/after examples:

Example 1: Finance Data

Bad:

"Look at our expense data and tell me what's happening."

Good
Here's our monthly expense data from 2024. Our budget is $500K/month. - D: Columns are Month, Category (Salaries, Operations, Marketing, Tech), Amount - A: Which category is over budget? What's the total overage? Is spending growing or stable? - T: Show results in a table - A: Explain which budget line needs attention first

Example 2: HR Data

Bad:

"Analyze our turnover."

Good
Here's our employee data: hire date, department, tenure in months, and departures this year. - D: 150 rows, 4 columns. Data covers 2024. - A: Which department has the highest turnover rate? Is it above industry average (15%)? - T: Show as a ranked table with percentages - A: For the highest-turnover department, what patterns do you notice about when people leave?

Example 3: Marketing Data

Bad:

"What's our ROI?"

Good
Campaign data: channel (email, social, paid search), spend, impressions, clicks, conversions, revenue. - D: 12 rows (one per campaign), last 90 days - A: Which channel has the best ROI? Which is losing money? - T: Rank by ROI. Include cost-per-conversion and revenue-per-dollar-spent - A: Should we shift budget away from any channel? Where would it perform better?

Example 4: Operations Data

Bad:

"How are we doing on inventory?"

Good
Inventory data: SKU, reorder point, current stock, units sold last 30 days, lead time (days). - D: 200 SKUs tracked daily - A: Which items are at risk of stockout (below reorder point)? Which have excess inventory (3+ months of stock)? - T: Two lists with prioritization - A: For stockout-risk items, how many days until we run out at current sales velocity?

Example 5: Sales Data

Bad:

"Show me our pipeline."

Good
Pipeline data: opportunity name, stage, value, probability, close date, days-in-stage, owner. - D: 45 open deals, tracking from prospecting through closed - A: What's our expected revenue if 100% of deals close as-is? Which stage has deals that are stuck (> 90 days)? - T: Show deal summary + stuck deals list - A: Which owner has deals most at risk of slipping? What's the likely close date distribution?

4. Charts, Reports, and Presentations

Claude can also help you turn data into narratives. Once you have the insights, use Claude to write the story for slides, reports, and presentations.

Use Case 1: Write Chart Titles and Descriptions

I have a chart showing {describe the data}. Write: 1. A specific, actionable title (not just "Sales Over Time") 2. A 1-2 sentence description of what the chart shows 3. A one-sentence "so what?" (what action should the viewer take?) Example data: {paste numbers or description}

Use Case 2: Create a Data Story

Turn this analysis into a 2-paragraph narrative for a slide deck. Data: {paste your analysis or numbers} Audience: {executives, team, board, customers} Action: {what do you want them to do with this insight?} Format: - Paragraph 1: The situation (what happened?) - Paragraph 2: The implications (why does it matter? what's next?)

Use Case 3: Executive Summary

Write a 3-sentence executive summary based on this analysis. Data: {paste analysis} Format: - Sentence 1: The finding (what did we learn?) - Sentence 2: The implication (why should we care?) - Sentence 3: The action (what's next?)

Use Case 4: Turn Numbers Into Insight

Write the "key takeaway" sentence for a slide. Make it compelling and specific. Data: {paste raw numbers or analysis} Context: {who's the audience? why do they care?} Output: One sentence that captures the main point. Avoid "things changed" or "it's important." Be specific.

The Data → Insight Workflow

Raw Data
(CSV, spreadsheet)
→
Claude Analysis
(trends, stats, insights)
→
Written Insight
(narrative, key takeaway)
Slide / Report
(presentation-ready)

Learn more about data storytelling: Storytelling with Data blog

🖥️HANDS-ON EXERCISE⏱ 25 min

Hands-On: Analyze Real Data

In 25 minutes, take a real spreadsheet and turn it into actionable insights using Claude.

  1. Step 1: Open any spreadsheet you have at work (sales, budget, operations, inventory, anything)
  2. Step 2: Select 10-20 rows and copy them as CSV format (include headers)
  3. Step 3: Paste into Claude with your first analysis prompt
  4. Step 4: Ask a follow-up question based on the first result
  5. Step 5: Ask Claude to write a 3-sentence summary of what the data shows
  6. Step 6: Copy the summary into an email or Slack message and send it to your team

Template 1: Financial Data

For expense, budget, or revenue data:

Here's our {expense/budget/revenue} data for {time period}. - D: Columns are {list columns}, covering {months/quarters} - A: What's the biggest variance from budget? What's growing fastest? Any red flags? - T: Show as a summary table - A: Write one action we should take based on this data

Template 2: Operational Data

For inventory, production, or process metrics:

Here's our {inventory/production/process} data. - D: {describe what each column means} - A: Where do we have problems (bottlenecks, high costs, low efficiency)? What's working well? - T: List problems and strengths separately - A: If we fixed the top problem, what impact would it have?

Quick Reference: Data Analysis Prompts

Quick Reference
Find Trends

Which metric is growing? By how much month-over-month? Is the growth accelerating?

Spot Outliers

What's unusual or surprising in this data? What's the best and worst performer?

Calculate Stats

Show me: average, total, percentage change, growth rate, and ranking

Summarize Data

Write a 3-sentence summary of the key findings. What should I tell my boss?

Compare Periods

How does this quarter compare to last? Same time last year? Show percentage differences.

Write for Slides

Turn this analysis into a compelling narrative for a presentation. One key takeaway sentence.

← Lesson 13: AI Writing Lab