The brief and the plan

01The problem

This is the capstone of the Data Analyst track. There are no new functions to learn. Instead, you'll do what a junior analyst is hired to do: take a business problem and a pile of raw data, and come back with answers a leadership team can act on.

The company is Voltline Electronics, a chain of eight stores selling phones, laptops, accessories, home appliances and solar power equipment in Lagos, Abuja, Port Harcourt and Ibadan. The chief executive, Ngozi Afolabi, has sent you this:

"We grew 37% in the first half of this year and the board is delighted. I'm not sure I am. I want to know what's really driving it, which stores and products are doing well and which aren't, and whether our targets make sense. The data is whatever the tills export. I need something I can take to the board in a month, and I need to trust it."

The data comes straight from the stores' tills, with all the problems that implies. Nobody has cleaned it, and nobody has checked it. That's your job too.

02The concept

Every analysis project follows the same arc

Use whichever tools you like: Excel, Power BI, SQL or Python, or a mix. The lessons show the key steps in more than one. What's assessed is the quality of the answers, not the tool.

A plan starts from the questions, not the data

Turn the brief into specific questions you can answer with numbers:

1. How much of the 37% growth is real? (Price rise? The new store? More customers?)
2. Which stores are improving, and which are struggling? Why?
3. Which categories make money, not just sales?
4. Are the store targets fair and achievable?
5. Where is money being left on the table (stock-outs, returns, missed add-on sales)?

Definitions before numbers

Write these down before you calculate anything, because every number depends on them:

• Net sales = quantity × unit price − discount, with returns (negative quantities) included.
• Gross profit = net sales − quantity × unit cost, using the cost price in force on the sale date.
• Like for like = stores open for the whole of both periods being compared.
• The analysis period = 1 January 2025 to 30 June 2026. "H1" means January to June.

Deliverables

Agree what you'll hand over: a cleaned dataset with a cleaning log, a dashboard of two or three pages, a one-page executive summary with three recommendations, and your working (queries, workbook or notebook) so someone can check it.

03Example

A first look at the files. The dataset has six CSV files:

Look at the grain (what one row represents) of each file. sales_raw is one row per line of a till transaction, targets is one row per store per month, and cost_prices is one row per product per cost change. Combining files at different grains is where many analyses go wrong. You'll deal with it in lesson 4.

04Walkthrough

1. Download the retail dataset and open every file. Write down the grain of each one in a sentence.
2. Read the data dictionary on the dataset card. Note any column whose meaning you're unsure of.
3. Open sales_raw.csv and scroll. Without doing any analysis yet, write down three things that look odd.
4. Write your analysis plan (the task below): the questions, your definitions and your deliverables.
5. Set up a project folder: raw/ (never edited), clean/, analysis/ and report/. Keeping the raw files untouched means you can always start again.

05Practice

06Check your understanding