Where should a Nigerian consumer business expand next?
A dashboard can show two states. A decision screen has to explain which one fits the decision, what evidence would reverse the answer, and where the data stops.
This is illustrative. It walks through how a Decision Screen is built. It is not a client case study and it is not evidence that ONYX has already completed this engagement. No company is named, and no ranking, saving, figure or outcome appears anywhere on this page, because none has been produced.
Start with a real choice
A screen begins with a decision that is already in motion, not a topic. For this question that means two candidate expansion markets, one launch window and one operating model — held fixed so the comparison stays like-for-like. The alternatives below are labelled A and B because naming a market here would be inventing the answer.
Expansion market A
Assessed on the same evidence lines, over the same launch window, under the same operating model.
Expansion market B
Assessed on the same evidence lines, over the same launch window, under the same operating model.
Fixing the window and the model is what makes the two columns comparable. Change either one and the screen is answering a different question.
Test the alternatives against the evidence that matters to that choice
Not everything that could be measured about a market is relevant to this decision. A screen works from a short set of evidence lines chosen because each one can change the ranking. Each line below states the question it settles, not the answer.
- 01
Market size
How many people the operating model can actually reach in each candidate market, and how that catchment is defined.
- 02
Cost pressure
What is happening to the input and consumer prices the model is sensitive to, and over what window that movement is measured.
- 03
Fiscal capacity
Whether the receiving government can sustain the spending and services the expansion assumes will be there.
- 04
Local economic intensity
How much activity the market carries relative to its size, so a large population is not mistaken for a large market.
- 05
Operating risk
What could interrupt the launch window or the operating model once committed, and how observable each of those risks is in advance.
Every input carries its source, period and method
The evidence register is the spine of the deliverable. One row per input, and each row has to survive being read on its own. Below is the shape of a register row — the fields it must state, and what each one is for.
| Field | What it has to state |
|---|---|
| Input | The specific quantity used, not the topic it belongs to. |
| Source | The named publisher the value came from, and the release it came from. |
| Period | The period the value actually describes — which is often not the period it was published in. |
| Method | How the value was derived, including any normalisation, rebasing or unit conversion applied to it. |
| Proxy | Whether the input stands in for the thing you care about. If it does, it is labelled a proxy and stays labelled a proxy. |
| Vintage | Whether the evidence is a live signal or older structural evidence. The two are never presented as the same thing. |
Two rules do most of the work here. Older structural evidence is not dressed up as a live signal. And a proxy is labelled as a proxy — including in the recommendation that leans on it.
What comes back
The output is not another country overview. It is four things, and the last two are the ones that make it a decision screen rather than a report.
A written recommendation
Which alternative the evidence supports, stated plainly, with the reasoning that gets you there rather than a score you have to take on trust.
An evidence register
Every input used, one per row, each carrying its source, period and method. It is the part you can hand to someone who wants to argue with the conclusion.
A comparison model
The alternatives set against the same evidence lines, so the comparison is like-for-like and you can change an assumption and see what moves.
Confirming and overturning signals
A short list of what to watch after the decision, split by which way each one cuts.
What would reverse the answer
A recommendation that cannot be overturned is not a recommendation, it is an opinion. So the screen names the signals that would confirm the result and the signals that would overturn it, and for each one it says which input it feeds, which direction it has to move to matter, and where you would see it move.
That list is what you monitor after the decision. It is also the honest test of the work: if nothing could change the answer, the evidence was never doing any work in the first place.
Where the data stops
Every screen ends with the limits stated in the same document as the recommendation, not in a footnote. That means naming which inputs are proxies, which are structural rather than current, where a series is thin or lagged, and which parts of the question the available evidence cannot reach at all.
Where the evidence cannot settle something, the screen says so and says what it would take to settle it. It does not fill the gap with a number.
Bring your own decision
One defined Nigeria question, two or more plausible alternatives, and a deadline or trigger that makes it real.
Start a Decision Screen →