For teams
Analysts need to know which number carries the answer
A good analysis separates what is known from what is assumed. A spreadsheet erases that line: a measured figure and a hopeful guess sit in identical cells, formatted to the same two decimals. The analyst's real job is to find the input that decides the answer and say how sure anyone can be about it.
Find the load-bearing input
Move each input across a plausible range and watch the result. Most inputs barely shift it. One or two swing the answer from yes to no, and those deserve nearly all of the scrutiny.
Analysts call this sensitivity analysis, and the chart that ranks inputs by their swing is often drawn as a tornado. The habit matters more than the chart: never present a result without naming the input it rests on.
Measured, estimated, hoped for
Label every material figure by its source: a system of record, a quote, a survey, a forecast, or a conversation. A forecast from the team that wants the project weighs differently from last year's audited cost, even when both sit in the same column.
Precision can mislead as well. A return of 31.7 percent suggests a confidence the inputs rarely support, and a range with its reasons is more honest.
When the evidence ties
Sometimes the truthful finding is that the evidence does not yet separate the options. Saying so, and naming the cheapest fact that would separate them, is stronger analysis than a ranking built on a coin toss.
An analysis in a Decision Room
Each case gets its own Champion, including the dull one where nothing changes. The Challenger goes after the input your sensitivity work flagged. Inspector marks each claim as supported, partly supported, unverified, or contradicted, so the ranking shows what it stands on. The room works the sums exactly, from your own figures.
The Verdict states its confidence as the strength of the record, never as a probability that the option is right. When the record is thin, it can end with No Strong Winner Yet.
Which number decides a $500,000 project
A project costs $500,000 and is expected to return $180,000 a year for five years. At a 10 percent discount rate its net present value is about $182,000, a clear yes. Push the first year of cash back by twelve months and the value falls to about $120,000, still a yes. Lower the yearly cash to $130,000 and the value turns slightly negative, about minus $7,000. The project survives a delay but not a smaller result, so the load-bearing claim is the $180,000, with break-even near $132,000 a year. The Challenger presses there. If the figure comes from the sponsor's forecast and no comparable project has delivered more than $150,000, the Verdict lowers its confidence and names the evidence that would restore it.
Questions
- Does the room replace my model?
- Your model supplies the figures. The room argues over which of them deserve trust and records that judgment beside your numbers.
- Who signs off on the conclusion?
- The person or team accountable for the choice. The Human Decision is recorded separately from the Verdict, so your analysis and their call can both be read later.
Further reading
Bring Argumont a Decision worth challenging.
Start with a paid 60 day pilot on your own real Decisions.