Six lessons, about an hour. Every technique from modules 1–8 is wasted if the answer doesn't land — this module is about making sure it does.
Write the sentence first: "Ibadan Central's revenue fell 14% in August, driven mostly by three fewer trading days after a generator fault." Only once that sentence exists do you build the chart, the slide, or the dashboard designed to make that exact sentence obvious at a glance.
Build the chart first, and you'll spend twenty minutes reverse-engineering what point it's supposedly making. Write the sentence first, and the chart becomes a much smaller, much clearer design problem: "what visual makes this specific claim visible in two seconds."
A chart type is an argument, not a decoration. Choosing the wrong one makes true data look like it's arguing something else.
| Comparing categories | Bar chart — Chidi can see which branch is biggest without reading a single label. |
| Showing change over time | Line chart — the shape of the trend is the point, not any single value on it. |
| Showing composition | Stacked bar, used sparingly — a pie chart with more than four slices argues nothing clearly. |
| Showing a single headline number | A big number with a comparison, not a chart at all — sometimes the honest visual is no visual. |
A dual-axis chart, a 3D pie chart, or a chart with fourteen colours is rarely clarifying — it's usually a sign the underlying finding hasn't actually been simplified to one sentence yet.
Chidi reads the first two sentences of anything you send him, and decides in that time whether the rest is worth his attention. Structure accordingly: finding first, then the one piece of supporting evidence that matters most, then anything else — never build up to the finding at the end.
"I looked at August sales across all nine branches, broke it down by category and by week, and after controlling for seasonality and comparing against last year..." — the actual finding arrives in paragraph four, if Chidi is still reading.
"Ibadan Central's revenue fell 14% in August — mostly explained by three lost trading days after a generator fault, not a real demand problem. Details below if useful."
Module 5's statistics discipline showed you how to know when you're not fully certain. This lesson is about saying so out loud without it sounding like you don't know what you're doing.
"Revenue is roughly flat month-on-month — within the range Lekki normally moves week to week, so I wouldn't read this as a real trend yet. I'll have a clearer read after another two weeks of data."
This states the number, states the honest uncertainty, and states exactly what would resolve it — which reads as more competent, not less, than a false-confident single figure that later turns out wrong.
Module 6.7 covered this for a dashboard specifically; the same discipline applies live, in a meeting, when Chidi or a branch manager pushes back on a number you're presenting in real time.
"That's useful pushback — can you say what number you were expecting? ... Okay, let me check whether we're looking at the same date range first, because that's the most common reason these disagree. [checks] You're right, I had July, you meant the rolling twelve months — let me pull that specific number now / I'll have it to you by end of day."
Never defend a number reflexively in the room before actually checking it. Never capitulate and agree it's wrong before checking either. The credible move, every time, is visible verification — even if it takes an extra minute of silence while you check.
Sometimes the honest answer is "I can't answer that with what we have" — and the module 1 scoping habit applies here in reverse: rather than a flat no, offer the smaller, real question that the same data actually can answer.
Chidi asks: "Would opening a tenth branch in Ikorodu definitely be profitable?" You: "I can't give you a definite answer from our data alone — that needs a market study we don't have. What I can tell you: based on our existing branches' revenue-per-square-metre and how similar-density areas perform, here's a realistic range. That's a real input, not a full answer."
Have your module 6 dashboard findings or module 5 drill results handy — you'll be turning one real analysis into three different deliverables.
Take any real finding from earlier in this course — Ibadan Central's August dip, the Lekki promotions fan-out, or one of your own drill results — and produce it three ways.
Same underlying finding, three genuinely different artifacts — not the same paragraph reformatted three times.
Whether you can adjust register and depth for three different real audiences without losing the honesty or the certainty-hedging that modules 5 and 9 have built — a Slack message that's too formal, or slides with too much text, both fail this just as much as an inaccurate number would.
Module 10 turns to running yourself as an analyst long-term — documentation, request queues, and building a portfolio from work like this drill's.