Free lesson

Build Your Data Analysis Assistant

Watch the video, then turn a real dataset into three findings and a short narrative your audience actually needs to hear.

What does the video cover?

  • Five practical uses: summarizing data, writing and explaining formulas, building tables, creating charts and pivot tables, and automating repetitive spreadsheet work.
  • Where AI sits: in a chat with an uploaded file, or right inside Excel and Google Sheets.
  • What to watch out for: messy data, fragile spreadsheets and outputs you have not checked.

The exercise: turn a real dataset into findings

You already know what is buried in your spreadsheets; you just do not have time to dig it out. This exercise puts AI on that job, so you walk away with findings written for your audience.

  1. Pick your tool. Upload your data to ChatGPT or Claude, or work inside your spreadsheet with Copilot in Excel, ChatGPT in Excel or Gemini in Google Sheets.
  2. Grab a dataset you already work with: a CSV export, a sales report or a tracking sheet with enough rows to be worth analyzing.
  3. Paste the prompt below with your details filled in, then attach or paste the data.
  4. Check the "so what." If the narrative misses what your audience cares about, tell the AI more about the audience and ask it to rewrite.

The prompt template

Prompt template
Character: You are a world-class data analyst specialized in [DATASET TYPE] who's equally sharp at finding patterns and framing them for a specific audience.

Objective:
I'm a [ROLE] at [COMPANY] in [INDUSTRY], and I need to analyze and communicate findings to [MY AUDIENCE].
Analyze the attached/linked/pasted dataset, surface the three most important patterns, and flag anything concerning. Then write the narrative my audience needs: not a data dump, the story of what matters, what it means, and what to do next.

Do:
- Lead with "so what," not methodology
- Match tone to the (implied if I didn't define it) audience: a board wants risk flags; a client wants value; a team wants next steps
- Keep the narrative to 3–5 sentences, ready to paste into a message or slide

Don't:
- Don't write like a report ("This analysis examines...")
- Don't explain what data analysis is; just do it

Output:
- Three key findings of one paragraph each
- A 3–5 sentence narrative written for my specified or implied audience, ready to send

How do you make this prompt yours?

The template works as it is, but it does not know you yet. Five changes turn a generic result into one you trust.

  1. Name the dataset type. "Quarterly cloud spend by service" gives the AI far more to work with than "a spreadsheet."
  2. Name one audience per run. A CFO, your team and a client need different stories from the same numbers; run the prompt once for each.
  3. Say what decision is on the table. For example: "The board will decide next month whether to renew this vendor."
  4. Clean the data first. Use clear column headers, remove merged cells and say what each column means if the names are cryptic.
  5. Ask it to show its work. Add: "For each finding, tell me which columns and rows it is based on," so you can check the numbers.

Working in Excel or Sheets? Our guide to AI for data analysis covers Copilot, ChatGPT, Claude and Gemini side by side.

AI Leader Advanced
Want this prompt written for you?

Personalizing a prompt by hand works, but it takes a few rounds. In AI Leader Advanced, LeadOS interviews you before the course starts and builds a personalized library of exercises and prompts around your role, your workflows and your goals. Your first analysis already knows your role, your audience and the decisions they face.

See AI Leader Advanced