The useful number is not what the project has spent. It is what the project is now expected to cost when everything still outstanding is included.
Forecast final cost combines cost already recognised with the cost still expected to finish the job. A basic conceptual formula is:
The exact treatment of claims, commitments and actual cost depends on the project system, but the commercial principle is consistent: do not count the same obligation twice, and do not pretend an unbought scope has disappeared merely because there is no invoice yet.
The budget is a baseline. Once tender returns, awarded packages, material quotes, variations or site conditions provide better information, the forecast should use that information. If a package is expected to exceed budget by $25,000, leaving the forecast at budget does not protect margin. It merely delays the admission.
A tendered subcontract or quoted material gives you a known procurement forecast. Any budget that remains genuinely uncovered still needs an allowance until there is a defensible basis to release it. This prevents a favourable quote in one area from being treated as a saving while other scope remains unprocured.
Pending variation cost, known scope gaps, likely rework and other identified risks belong in the forecast where the project is reasonably expected to incur them. They should not wait for an invoice to become real.
Whole-project totals can hide problems. Review forecast variance by trade or cost category, identify the largest adverse movements, then record the reason and action. That is where the forecast becomes management information rather than arithmetic.