Module 9 · Data Analytics Foundations
From question to insight
A repeatable method from a vague worry to a recommendation: clarify, make it measurable, break the total down, test and rule out explanations, and tell the story answer-first.
About 10 minutes
The problem
The sales director at Kolanut says: "Something's wrong in the North. Can you look into it?"
That isn't a question you can answer with data yet. This lesson is about the method analysts use to get from a worry like this to a finding someone can act on.
The concept
Most requests to an analyst start vague: "North West looks bad, can you have a look?" This lesson is a repeatable method for turning that into a clear finding and a recommendation someone can act on.
The question-to-insight method
- Clarify the business question. Who is asking, what decision are they facing, by when?
- Make it measurable. Name the measure, the comparison and the period.
- List the data needed, and check it exists.
- Analyse: start broad, then break down. Compare with a baseline.
- Explain the pattern: test possible reasons against the data.
- Recommend something specific, and say how you'd know if it worked.
Step 1: clarify
Before touching data, ask the person three questions:
| Ask | Why |
|---|---|
| What decision will this help you make? | It tells you what "useful" means. "Should we replace the North West approach?" needs a different analysis from "how much stock should we send?" |
| By when? | A rough answer on Friday may beat a perfect one next month |
| What do you already believe? | You'll know which explanation to test first, and which belief your answer has to address |
Step 2: make it measurable
A measurable question names four things:
| Part | Vague | Measurable |
|---|---|---|
| Measure | "how it's doing" | revenue after discounts |
| Scope | "the North" | customers in the North West region |
| Period | "lately" | January to June 2026 |
| Comparison | (none) | against January to June 2025 |
Put together: How did North West revenue in January to June 2026 compare with the same months of 2025, and what drove the change?
Steps 4 and 5: break it down and test explanations
A total can be broken into parts that multiply together, like the KPI tree in lesson 2. For a fall in revenue, the usual suspects are:
| Possible cause | What you'd see in the data |
|---|---|
| Fewer customers | Fewer distinct customers ordering |
| Customers ordering less often | Same customers, fewer order lines each |
| Smaller orders | Fewer packs per order line |
| Lower prices or bigger discounts | Lower naira per pack or higher discount % |
Each one is a hypothesis: a possible explanation you can check with a number. Work through them all, not just the first one that seems to fit. The ones you rule out are as useful as the one you confirm.
Finding versus insight
A finding says what the data shows ("North West revenue fell 47%"). An insight adds why it matters and what to do ("North West shops are ordering half as often; if we win back the old order frequency, we recover about ₦14m per half-year").
| Says | Example | |
|---|---|---|
| Data | A fact | North West: 91 order lines in H1 2026 |
| Finding | What changed | North West revenue fell 46.6% |
| Insight | Why, and so what | The same shops are ordering half as often; that's worth about ₦14m a half-year |
| Recommendation | What to do, and how to check | Call the five largest accounts this month; target 150 order lines next half-year |
Correlation is not cause
Two things moving together doesn't mean one causes the other. North West's fall came in the same half-year as the price rise, but prices rose in every region, and most regions grew. So the price rise alone can't explain North West. Always ask: did the same thing happen somewhere that didn't fall? If so, it probably isn't the cause.
Telling the story
Decision-makers are busy. Put the answer first, then the reasons, then the detail:
- Answer: "North West revenue nearly halved because shops ordered half as often. We haven't lost customers."
- Evidence: the table, one chart.
- What to do: the recommendation, with a target and a date.
- Limits: what the data can't tell you yet.
This is sometimes called the pyramid principle: lead with the conclusion, support it underneath. A report that saves its answer for the last page loses its readers on page one.
Example
1. Clarify. The director is deciding whether to replace the North West sales approach before the next half-year budget.
2. Measurable question. How did North West revenue in January–June 2026 compare with January–June 2025, and what drove the change?
3. Data. orders.csv (dates, quantities, prices) joined to customers.csv (region).
4. Analyse.
| North West, January–June | 2025 | 2026 | Change |
|---|---|---|---|
| Revenue (₦m) | 31.1 | 16.6 | −46.6% |
| Customers who ordered | 10 | 11 | +1 |
| Order lines | 176 | 91 | −48% |
| Order lines per customer | 17.6 | 8.3 | −53% |
| Packs per order line | 13.9 | 12.7 | −9% |
| Naira per pack | 12,724 | 14,450 | +14% |
5. Explain. Test each suspect:
- Fewer customers? No: 11 ordered in 2026, one more than in 2025. Ruled out.
- Lower prices? No: naira per pack rose 14%, after January's price rise. Ruled out.
- Smaller orders? A little: packs per line fell 9%. A minor cause.
- Ordering less often? Yes: order lines per customer more than halved, from 17.6 to 8.3. The main cause.
The fall is mostly order frequency: the same shops ordered about half as often, with slightly smaller orders on top.
6. Recommend. Find out why North West shops are ordering less often. Ask the regional rep and call the five largest accounts this month: are they buying from a competitor, or is delivery unreliable? Set a target of 150 order lines next half-year and track it monthly.
Walkthrough
Notice what the analysis did not do:
- It didn't stop at "revenue fell 47%". That's a finding, not an explanation.
- It didn't guess. Each explanation was checked against a number, and two were ruled out.
- It didn't claim more than the data shows. The data says what changed (frequency); only a conversation with customers can say why. A good recommendation names the next question as well as the next action.
Now practise the method on your own question. Pick one of these and write steps 1 and 2 on paper:
- "Are our supermarkets doing well?"
- "Are discounts worth it?"
- "Is Lagos too important to us?"
For each, write: the decision it supports, the measure, the scope, the period and the comparison. Then list two or three hypotheses and the number that would test each.
Summary
| Step | Key question |
|---|---|
| Clarify | What decision, by when, and what do you already believe? |
| Measurable | Which measure, scope, period and comparison? |
| Data | Does it exist, and is it clean? |
| Analyse | What changed, and which part of the total moved? |
| Explain | Which hypotheses does the data confirm or rule out? |
| Recommend | What should happen, by when, and how will we know it worked? |
Practice
Practice
Check the analysis yourself. How many order lines did North West customers place in January–June 2026? (Look up each order's customer region, then count 2026 order lines for North West.)
Practice
Which region's revenue grew by the most naira from H1 2025 to H1 2026? Use the table in lesson 6 (Data analysis).
More practice
Optional drills. They don't count towards the certificate, but they're the fastest way to make this lesson stick. Several use a different dataset from the lesson on purpose: if you can do the same thing on unfamiliar data, you've really learned it.
Drill · optional
How many different customers placed at least one order in January–June 2026?
Check your understanding
Answer every question to check.