Boards love a single number. It's clear, it fits in a budget line, and it feels like a commitment. The problem is what it leaves out.
A point estimate is one cost figure, such as the base estimate. A range estimate shows the spread of likely outcomes and how probable each one is. A single number says nothing about its chances of holding, and on most projects the honest answer would be: not great.
What a single number hides
Take the fictional Millbrook Connector, a road and bridge project estimated at $120 million. Every line in that estimate is a reasonable "most likely" value. Nobody cut corners. But once you simulate the uncertainty in each line, plus six known risks, the picture looks very different:
| Budget option | Amount ($M) | Chance of staying within budget |
|---|---|---|
| Point estimate | 120.0 | 1% |
| Point estimate + flat 10% | 132.0 | 33% |
| P50 | 135.8 | 50% |
| P80 | 143.4 | 80% |
None of the individual lines in that $120 million was wrong. The total was. Skewed cost items have a habit of overrunning far more than they underrun, so their most likely values don't add up to the most likely total. Add the customary 10% on top, and the board still has a two-in-three chance of coming back for more money. Our guide on why a percentage on top is not a risk budget shows the same effect on two other projects.
And this isn't some exotic case. Across 258 transport projects, nine out of ten went over budget, with an average overrun of 28% (Flyvbjerg et al., 2003).
Why ranges lead to better decisions
A range does three things a single number cannot.
- It makes risk appetite a real choice. Picking P50 or P80 is a deliberate decision about how much overrun risk you accept. With a point estimate, that decision gets made by accident. P50, P80 and P90 explained helps you pick the level.
- It shows you where to act. A range comes with its drivers. In the Millbrook example, the bridge costs and the risk of poor ground at the foundations dominate, which makes a strong case for more site investigation before the budget is fixed.
- It protects your credibility. A board that approved "$136 million at 50% confidence, $143 million at 80%" won't be shocked when the bill lands at $139 million. A board that approved $120 million will be.
More and more governments ask for ranges. The Australian Government requires P50 and P90 estimates for larger projects seeking Commonwealth funding. In the Netherlands, the forecast for the national flood defense program was reported as a P50 of €18.4 billion within a range of €14.3 billion to €22.5 billion (Horvat & Partners, 2024).
How to present a range to a board
The usual pushback is practical: "We still need one number for the budget." Fair enough. The answer isn't to hide the range. It's to build the decision around it.
- Show three numbers, not a chart full of percentiles: the P50, the proposed budget level (often P80) and a high case (P90).
- Name the top three drivers, and what is being done about each one.
- Ask for a clear decision. For example: "We propose a project budget of $136 million (P50) and a sponsor reserve of $7.6 million, bringing total funding to $143 million (P80)."
- Agree when you'll come back. "If the forecast goes above P80, we return to the board with options."
The board gets its single number, and the range becomes the reason behind it.
When a point estimate is fine
For small, routine work with plenty of comparable history, a point estimate plus a factor-based allowance can do the job. The more unique, larger or earlier the project, the more a range is worth.
Frequently asked questions
Isn't a range just a way to ask for more money?
No. A range often shows that part of the allowance isn't needed yet. It also lets you hold reserve centrally instead of padding every single estimate.
How wide should a range be?
As wide as the evidence supports. Early on, ranges of minus 20% to plus 50% are common. They narrow as design and procurement move forward.
Quantil's QRCIM turns a cost estimate and risk register into a board-ready range in Excel. Try the free demo.
References
- Department of Infrastructure, Australian Government (2023). Guidance Note 3B: Deterministic Contingency, v2.0.
- Flyvbjerg, B., Holm, M. K. S., and Buhl, S. L. (2003). How common and how large are cost overruns in transport infrastructure projects? Transport Reviews, 23(1), 71–88.
- Horvat & Partners (2024). Berekening bandbreedte prognose dijkversterkingsopgave, for the Dutch Ministry of Infrastructure and Water Management.