Week 1 — Questions, Data & Spreadsheets · ~20 min

Using AI in this course: the rules

Free preview

Why this lesson comes first

AI assistants like ChatGPT, Claude, Gemini and Microsoft Copilot can write an Excel formula, a DAX measure or a SQL query in seconds. Copilot is even built into Excel and Power BI. So why learn to write them yourself?

Because the formula was never the job. The job is knowing the number is right. An analyst who can't check a total will one day hand a manager a confident, beautifully formatted, wrong answer, and the manager will act on it. AI makes formulas cheap. It makes checking more valuable, because there are more unchecked numbers around than ever.

You can only check what you understand. So this course brings AI in gradually.

The three stages

StageWeeksWhat AI may doWhat you must do
Foundations1–6 (Excel, Power Query, modelling)Explain functions, error messages and concepts, like a patient tutor.Write every formula and transformation yourself.
Pair7–11 (DAX, dashboards, SQL)Suggest approaches, draft measures and queries.Check every result by hand against the practice data, and explain every line in a spot check.
Lead12 (SQL project)Anything a teammate could do.Own the result: checks, documentation, and a live defence.

Rule 1: Disclose your AI use

Every submission includes an AI-use note. There's a box for it on the submission form:

AI use: asked ChatGPT why my SUMIFS returned 0
What I checked: the criteria range and the sum range were different sizes — fixed
What it got wrong: it suggested SUMIF with the arguments in the SUMIFS order

"No AI used" is a fine answer. Honest disclosure never costs marks. Hidden AI use that we discover in a spot check is treated as not your work.

Rule 2: Never paste these into an AI tool

  • Real customer data: spreadsheets with names, phone numbers, addresses, account numbers, BVN, NIN, salaries.
  • An employer's or client's sales, finance or HR data, unless their policy explicitly allows that tool.
  • Passwords, database connection strings, API keys, Power BI credentials.

Under the Nigeria Data Protection Act 2023 (NDPA), personal data may only be processed for a lawful purpose and must be kept secure. Pasting a customer list into a public chatbot sends it to a third party — that is processing, and you need a lawful basis for it.

What to do instead: describe the structure ("a table with columns order_id, customer_id, order_date, quantity, unit_price"), or paste the fictional practice data, or invent sample rows. AI doesn't need real data to help you with a formula.

Rule 3: Verify every number before you trust it

AI is fluent and frequently wrong. In analytics that looks like:

  • formulas using functions your Excel version doesn't have (XLOOKUP in Excel 2016, TEXTSPLIT before 365),
  • SUMIF and SUMIFS with the arguments in the wrong order,
  • DAX that runs but answers a different question (wrong filter context, ALL where you needed REMOVEFILTERS),
  • SQL joins that fan out and double a total, or a WHERE that turns a LEFT JOIN into an INNER JOIN,
  • confident claims about your data it has never seen.

The defence is the habit this course drills from Week 1: check against numbers you already know. The practice data has 8 rows and total revenue 160; January 35, February 65, March 60; customer 5 has no orders. If an AI-written formula, measure or query doesn't reproduce those, it is wrong — no matter how good the explanation sounds.

Rule 4: You own what you ship

If a number in your report is wrong, "the AI wrote it" is not a defence — not in this course, and not at work. Your name is on the report.

Self-check

  1. In Weeks 1–6, may you ask an AI to write your cleaning formulas? What may you ask it?
  2. Name two kinds of data you must never paste into an AI tool, and what to paste instead.
  3. An AI gives you a SQL query that returns total revenue 320 on the practice data. What do you conclude?

Sign in to track your progress.