Every estimator has watched a carefully built range collapse to its midpoint the moment it reached a slide. The fault is rarely the audience. It is that the range was presented as a caveat rather than as the answer.
The work behind the range was real: line ranges, a risk register, a simulation, a distribution. Then the presentation opens with one number in seventy-two point font and the range appears underneath in grey, looking like an apology. The room does what rooms do: it writes down the big number and moves on. Four years later the big number is still being quoted, and the range that would have explained everything died in a footnote.
Why ranges die in the room
Three forces kill them. A single number is quotable and a range is not, so the minutes, the press release and the next slide deck all reach for the point. Precision reads as competence, so the estimator who says “212.4 million” sounds more certain than the one who says “180 to 260”, even though the second one is telling the truth and the first one is not. And a range presented without structure sounds like hedging: the room hears “we do not know” instead of “here is exactly what we know and what we do not”.
None of these forces can be argued away. They have to be out-presented. The range has to arrive as the deliverable, with a shape and named parts, the way it does at concept stage where the range is the whole answer: see A Class 5 estimate is a screening tool, not a budget.
The problem is recognised well beyond any one estimating desk. AACE International Recommended Practice 104R-19, Communicating Expected Estimate Accuracy (RP 104R-19), exists precisely because ranges fail so often in the telling, and RP 41R-08, Understanding Estimate Ranging, covers the thinking that produces a defensible range to begin with. Both are available to AACE members at web.aacei.org. What follows is the presentation discipline that makes the analysis survive the meeting.
Lead with the decision
Before any number appears, say what confidence this decision needs to carry. A screening decision between concepts can run on P50: an even chance of doing better or worse is fine when the question is which idea to study. A budget can be set at P50 with an owner who understands what that means. A funding request usually needs P80, because a four-in-five chance of staying inside the ask is what boards and lenders can defend to their own stakeholders. The choice of confidence level belongs to the owner, and where that number comes from is its own subject: see Contingency is not padding.
Once the decision’s confidence level is agreed, the range stops being a hedge and becomes the machine that produces the number. The room is no longer being asked to tolerate uncertainty. It is being shown where its number came from.
Three numbers, one chart
The presentation needs exactly three points, on one simple horizontal band, each with its story attached.
- P10: the optimistic case, and what has to go right to get there. Not a fantasy: a list. Ground as expected, quotes holding, no schedule slip.
- P50: the base, with the largest three drivers named. This is the working number, and the drivers are what make it credible.
- P90: the case to be funded against, with the risks that push it there. Each risk named, each with an owner and a response.
The script, in one paragraph
Here is what it sounds like when it works. “The project is 180 to 260 million, and the most likely outcome is around 210. The low end requires the ground to match the boreholes and the equipment quotes to hold, and we would not plan on it. Three things drive the spread: ground conditions at the river crossing, the electrical package market, and the permit date. The 260 case is those three going against us together; each has a response plan, and the register behind them is in the appendix. We recommend funding at 240, which the analysis puts at P80.” Ninety seconds. Every number has a reason, every risk has a name, and the room has been given something better than certainty: a map.
Compare the alternative: “The estimate is 212.4 million.” Faster, cleaner, and wrong in a way that will not surface until it is expensive.
A range with named drivers is a plan. A range without them is a shrug.
What to leave out
The histogram, the tornado diagram and the correlation matrix. They belong in the appendix, ready for the one director who asks. The board needs three numbers, the reason for the spread, and what is being done about the top risks. Every chart beyond that spends attention the drivers need.
Where this goes wrong
The midpoint in the headline. The slide leads with the P50 and mentions the range verbally. Whatever is in the biggest font is the number the organisation remembers; put the range there.
False precision. A number quoted to the decimal, at a stage where the honest range spans forty percent. The decimal does not add information; it adds a promise the estimate never made.
P90 sold as “safe”. Funding every project at P90 is not prudence; it is capital locked up against risks that mostly will not happen, starving the next project. The confidence level is a portfolio decision with a cost, and presenting it as free makes the estimator sound naive to a good CFO.
Drivers unnamed. A range with no explanation of what widens it invites the room to negotiate it, and a range that can be negotiated was never an analysis; it was an opening offer.
The moving target. P50 in one meeting, P80 in the next, with no note of the change. The numbers differ, someone spots it, and the credibility of both dies in the reconciliation.
The point of the whole exercise
Decision-makers do not actually want one number. They want to not be surprised. A range with named drivers is the only format that delivers that: it tells the room today what the trend report would otherwise tell them in eighteen months, at negotiating distance instead of at crisis distance. Present it as the answer, and it gets used as one.
Emerald Group prepares estimates and risk analyses for board and funding decisions, and presents them in the room when owners want the estimator answering the questions. If your next number has to survive a board meeting, get in touch.