Your first AI-powered analysis

Let's do a real analysis together, the reliable way. We'll use the pattern that works in any code-executing AI tool (ChatGPT Advanced Data Analysis, Claude, Excel Copilot). Grab a small real dataset — a spreadsheet of sales, expenses, survey responses, anything.

Step 1 — Upload and orient. Load your data into a tool that runs code. Start by asking it to understand the data: "Here's my data. Describe what's in it — the columns, the number of rows, and any obvious data quality issues." This gives you (and the AI) a shared picture, and surfaces problems before you analyze.

Step 2 — Clean, visibly. Ask it to fix issues as steps you can see: "List the data quality problems and write the code to fix them (handle the missing values, standardize the date formats)." You want the cleaning to be visible and reversible, not silent. Review the steps.

Step 3 — Ask a specific analytical question. Vague questions get vague answers. Instead of "analyze this," ask something concrete: "What were total sales by month, and which month was highest?" or "What's the average order value by region?" Specific, answerable questions get computed, useful answers.

Step 4 — Confirm it's computing, not guessing. Watch for the tool to run code (you'll see "analyzing…" and can open the code). The number in the result comes from real computation. If instead it just typed a number in prose with no code, ask it to "calculate this using Python/a formula and show me" — force the reliable path.

Step 5 — Visualize. Ask for a chart: "Make a bar chart of sales by month." The AI writes real plotting code and produces the chart. Check that the chart matches the numbers (right axes, right aggregation).

Step 6 — Verify. Before trusting a key number, sanity-check it (Module 3 covers this properly): does the total seem right? Do the parts add up? Reproduce one number independently if it matters. "AI says 12,000 total orders — does that roughly match what I'd expect?"

Step 7 — Summarize for humans. Ask it to explain the findings in plain language: "Summarize the key insights from this analysis in a few bullet points a manager would understand." Feed it the computed numbers and let it describe them — it's great at turning results into clear prose.

What you just did: a complete analysis — understand → clean → question → compute → visualize → verify → communicate — with AI doing the heavy lifting and real code producing the numbers. Notice how much faster it was than doing it manually, and that you can trust the results because they were computed and checked.

Two beginner habits from this: always work in a tool that executes code (so numbers are real), and ask specific questions (so answers are useful). Do that, and even your first AI analysis is both fast and trustworthy. The rest of the course deepens each step — better cleaning, sharper questions, honest statistics, and rigorous verification.