The feasibility number becomes the budget
An early cost figure is produced to test whether a project is worth studying. It survives as the number the project is held to, long after the basis for it has gone.
Every capital project produces a cost figure long before it has a design. It is produced quickly, from limited information, to answer a screening question: is this worth spending money to study.
That figure then does something it was never built to do. It gets repeated in a paper, quoted in a board minute, written into a funding request, and eventually becomes the number the project is judged against. Nothing about the estimate changed. Its status did.
What an early estimate actually is
An early figure is a statement about a class of projects, not about this one. It is derived from a capacity metric, a rate per unit, a comparable asset, or a factored allowance, applied to a scope that has been described in a paragraph.
Its accuracy range is wide and, in a competent organization, stated. A range of plus or minus a third is normal at that stage, and it is honest. The estimate is fit for its purpose, which is to distinguish projects worth developing from projects that are not.
The failure is not in the number. It is in what happens to the range.
How the range disappears
The range is lost through ordinary organizational behavior, not through anyone's error.
A range is difficult to put in a table, so the midpoint is used. A midpoint in a table looks like a figure, and a figure in a board paper looks like a commitment. Once it has been approved in any form, restating it upward requires someone to explain why the number has gone up — which is a harder conversation than not mentioning it.
Meanwhile the project's definition improves. Capacity is confirmed, redundancy is chosen, the site is selected, utility constraints appear, and the scope acquires the things a paragraph could not describe. Each of these is a legitimate cost. None of them was in the original figure, and collectively they are usually not small.
The overrun did not happen during construction. It happened at the moment a screening figure stopped being described as a screening figure.
Re-baselining is a governance act, not an admission
The remedy is unglamorous and rarely resisted once it is proposed properly: the estimate is re-stated at defined points, with the basis recorded, and the difference is explained by scope rather than by performance.
That requires three things to be written down at each stage. What is included and excluded. What has been assumed where a decision is still open. What the accuracy range is, and why.
An estimate carrying those three things can be superseded without anyone losing face, because the reason for the change is legible. An estimate carrying none of them cannot be superseded at all, because there is no way to distinguish a scope increase from a cost failure.
Reading an estimate you have been handed
Four questions separate an estimate that can be relied on from a figure that has simply survived.
What is the basis of estimate? If there is no document naming the source of each major quantity and rate, the figure is an opinion with decimal places.
What is the stated accuracy range, and at what confidence? A number without a range is being presented as more certain than the information behind it permits.
What is excluded? Owner costs, land, utility contributions, escalation, financing costs, and commissioning support are the usual omissions, and any one of them can be material.
When was it produced, and what has been decided since? An estimate is a snapshot of a project definition. Where the definition has moved and the estimate has not, the gap between them is unquantified risk sitting in plain sight.
The organizational fix
Projects that avoid this pattern do one thing differently. They keep the estimate and the definition attached to each other, formally, so that the number cannot be quoted without the scope it belongs to.
It is a documentation habit rather than a technique, and it removes the most common way a capital program arrives at its first crisis: not by spending too much, but by having been measured, from the beginning, against a number that was never intended to be a target.
