Step 1: Clean Your Data So AI Can Read It
Open your data in Google Sheets and do these three things first:
- Make sure the first row is a header: Every column needs a clear label (e.g. “Date”, “Revenue”, “Customer Source”)
- Remove blank rows and merged cells: AI struggles with messy formatting
- Standardize formats: Dates as YYYY-MM-DD, amounts as plain numbers (no currency symbols or thousands separators)
Once clean, select the data range and copy it. If you have more than 500 rows, just copy the first 500 — Claude’s free tier has input length limits, and most analyses work fine with a sample.
Pitfall warning: The most common disaster is hidden outliers in your data. If one day’s revenue is 100x the normal amount (likely a data entry error), Claude will treat it as real and draw wild conclusions. Run =MAX() and =MIN() in Google Sheets first to scan for extreme values and manually exclude obvious errors before handing data to AI.
Step 2: Use Claude for Deep Analysis
Paste your data into Claude with this prompt:
You are a senior data analyst preparing a monthly report for company leadership.
Here is our [business type, e.g. "e-commerce sales"] data covering [e.g. "January–March 2026"]:
[paste data]
Please complete the following analysis:
1. Data overview: Summarize overall performance with 3–5 key metrics (month-over-month and year-over-year changes)
2. Trend analysis: Identify upward/downward trends, flag inflection points and likely causes
3. Anomaly analysis: Flag data points outside normal range with possible explanations
4. Attribution analysis: Which factors had the biggest impact on core metrics?
5. Action recommendations: Provide 3 specific, actionable next steps based on the data
Requirements:
- Back every conclusion with specific numbers (e.g. "March grew 23% month-over-month")
- Action recommendations should specify who does what, and expected outcome
- Business-professional style, appropriate for a Director or VP audience
Claude will return a structured analysis report. Then ask it to generate chart data:
Based on the analysis above, please generate data tables for the following charts in a format I can use directly in Google Sheets:
1. Monthly trend line chart for core metrics
2. Channel/category breakdown pie chart data
3. Year-over-year comparison bar chart data
Pitfall warning: Claude occasionally makes arithmetic errors, especially with percentage and month-over-month calculations. Always spot-check 2–3 key numbers by verifying them in Google Sheets with formulas. One wrong number caught by your manager is enough to undermine the credibility of the entire report.
Step 3: Build Charts in Google Sheets
Back in Google Sheets, use Claude’s data tables to create charts:
- Enter the chart data Claude provided
- Select the data → Insert → Chart
- Choose the right chart type for your data:
- Trends over time → Line chart
- Proportions → Donut chart (more modern than a pie chart)
- Comparisons and rankings → Horizontal bar chart
3 tips for making charts look professional:
- Use no more than 5 colors; use light-to-dark variations within a single color family
- Remove gridlines; keep only data labels
- Use conclusion-driven titles, not descriptive ones (write “March Revenue Hits Record High, Up 23% MoM” not “Monthly Revenue”)
If you need more visually polished output, Canva Pro ($12.99/month) has data visualization templates where you can drag in your data and get publication-ready charts instantly.
Step 4: Assemble the Full Report in Google Slides
Open Google Slides and build your deck with this structure:
| Slide | Content | Notes |
|---|
| S1 | Cover | Report title + date range + your name |
| S2 | Executive Summary | 3–5 key numbers in large text |
| S3–S4 | Trend Analysis | Line chart + annotated inflection points |
| S5 | Attribution | Bar/donut chart + brief interpretation |
| S6 | Recommendations | 3 action items, each with owner and timeline |
One core message per slide. Have Claude write the copy for each:
I'm building a 6-slide monthly data report for company leadership.
Based on our earlier analysis, please write for each slide:
1. Slide title (under 10 words, conclusion-style)
2. 3–4 bullet points (one sentence each, including specific numbers)
3. Speaker notes (2–3 sentences on what to say when presenting this slide)
Style: concise, data-driven, action-oriented.
Copy charts from Google Sheets into Slides — they stay linked to the source data, so updates flow through automatically.
Pitfall warning: The three things executives hate most in reports: ① Dense tables of raw numbers (replace them with charts); ② Data without conclusions (every slide needs a “so what?” sentence); ③ Analysis with no next step (beautiful analysis that doesn’t recommend action is wasted work). Run this three-point check on every slide before moving on.