Stop 12 · AI & Modern BA PracticesTopics · 3/7

AI for Data Analysis & Reporting

In short

AI can write your SQL, suggest a chart and explain a spreadsheet in plain English. That makes you faster. It does not make the numbers right, so you still need to know when an answer smells off.

Using AI to speed up data exploration, generate first-draft reports, or explain a dataset in plain language — extending Phase 5's data analysis skills with a genuine productivity multiplier.

The purple blob at a kirana counter compares a giant calculator with an open ledger book.

Think of it like this

The kirana calculator

The shop owner punches totals into a calculator because it's quick. But when it says 40 kg of sugar sold on a Tuesday, he frowns and checks the ledger. He knows his shop. The calculator doesn't. That's you and AI with data.

Watch it happen

Asking AI for a sales query

AI assistant

Database

Schema + question

1/5The BA shares table and column names only, plus the question: revenue by city last month.

The big ideas

What you need to know

Drafts queries and formulas

Describe what you want and AI writes SQL, XLOOKUP or pivot steps. You read it before you run it.

Suggests charts and summaries

It can propose a chart type and write a first summary paragraph. You decide what the story really is.

Confident and wrong

AI may misread a column, ignore nulls or make up a trend. Always tie at least one number back to a trusted source.

Share the shape, not the data

Give column names and a few fake rows instead of real customer records, unless your tool is approved for that data.

See it in code

Walk through the code

sql

1-- AI draft, with the BA's fix on line 72SELECT city,3       SUM(amount) AS revenue4FROM orders5WHERE order_date >= '2026-09-01'6  AND order_date <  '2026-10-01'7  AND status = 'paid'8GROUP BY city9ORDER BY revenue DESC;

1/4One row per city with total revenue.

The AI wrote most of this. The BA added the status filter after noticing cancelled orders were counted as revenue.

Save yourself the pain

Common mistakes

Oops

Pasting an AI's chart and its summary into the report without checking a single number.

Do this instead

Reconcile at least one total with a trusted source, like the finance report, before anything goes out.

Oops

Uploading a full customer export with phone numbers to get a quick analysis.

Do this instead

Remove personal columns or use sample rows, and only use tools approved for that kind of data.

Your mission

Catch the AI's mistake

~35 min

Quick check

AI says 'Sales grew 300% in March'. What's your first move?

Go deeper

1Use an AI tool to summarize a sample dataset and identify a notable pattern.
2Compare the AI-generated insight against your own manual analysis for accuracy.
3Learn to verify any AI-generated numeric claim independently before presenting it to stakeholders.