Halfway through a capital project, most of the reported financial position is history. Spend to date is a fact. Commitments are a fact. The only number that carries information about the outcome is the cost-to-complete, and it is routinely the least rigorous figure in the pack.
Two ways to produce the number
The first method is subtraction: approved budget, less spend to date, less commitments, equals contingency remaining, therefore the forecast is the budget. This is fast, requires no judgment, and is what a finance system produces by default.
The second method is a build-up: what scope remains, what it will cost at today's rates, what the open change position is likely to settle at, what the remaining program costs in time-related overheads, and what residual risk is still carried. This requires the project team to form a view and defend it.
The two methods agree while the project is going well. They diverge exactly when the answer matters.
Subtraction assumes the budget is right. A forecast tests whether it is. The moment those two stop agreeing is the moment you needed to know about six months ago.
Where the subtraction hides the problem
Subtraction conceals overruns through three mechanisms.
Unpriced change is the largest. Instructed variations that have not yet been agreed sit outside the commitment position. On a project with an active change environment, the unagreed pipeline can be a material fraction of the remaining budget, and it is invisible to a method that only counts what has been signed.
Productivity is the second. Where installed quantities are running behind planned rates, the remaining work will cost more than the estimate assumed. This is visible in the field long before it is visible in the ledger.
Time-related cost is the third. Preliminaries, site overheads, supervision, and hire accrue per week. A schedule slip converts directly into cost, and a subtraction-based forecast will not show it until the invoices arrive.
What a defensible build-up contains
Remaining scope, quantified — not percentage complete, which is an opinion, but measured quantities remaining.
Current rates rather than estimate rates. Where the market has moved since the estimate was prepared, the difference is real money and should be stated.
The open change position, with a probable settlement value. Carrying instructed-but-unagreed change at zero is not conservatism; it is omission.
Time-related cost against the current, honest schedule — not the published one, if those differ.
Residual risk, drawn from the risk register rather than from a percentage.
The percentage-complete trap
Physical progress reported as a percentage is the least reliable input to a forecast and the most commonly used.
The number is usually derived from cost incurred against cost budgeted, which makes it circular: it reports that the project has spent what it has spent. Where it is derived from earned value against a properly quantified baseline it means something, but that requires a baseline detailed enough to measure against — which many projects do not have.
The practical test is to ask what the percentage is calculated from. If the answer is cost, it carries no independent information about progress.
When to insist on an independent read
Three conditions justify commissioning an independent cost-to-complete.
When the reported forecast has not moved for several consecutive periods despite known change. Stability of that kind is either genuine or a sign the forecast is being managed rather than produced.
When contingency drawdown has outpaced risk closure.
When the change pipeline has grown faster than it is being settled, which indicates the commercial position is accumulating rather than resolving.
None of these prove a problem. All of them justify asking for a number built from the bottom up, by someone whose reputation does not depend on the answer.
