Cost overruns are not a contractor problem. They are a governance problem. By the time a project is twenty percent over budget, the contractor has usually been signaling stress for months — in pay applications, in schedule updates, in change order velocity — and the owner's controls failed to convert those signals into action. The five controls outlined below are not novel. They are, however, the controls we most consistently find missing or under-resourced when we are asked to forensically reconstruct why a program failed.
The discipline of project controls is fundamentally about converting noisy field data into the small number of executive-grade indicators a steering committee actually needs. Programs that do this well rarely surprise their boards. Programs that do not, surprise them at the worst possible moment.
Control #1: A Cost Breakdown Structure That Mirrors the Schedule
The Cost Breakdown Structure (CBS) is the chart of accounts for the project. Done well, it maps one-to-one with the Work Breakdown Structure used in the CPM schedule, which means every dollar earned in the field can be tied to an activity, and every activity has a budget against which to measure earned value. Done poorly, the CBS is a finance-department artifact that tracks invoices by trade but cannot answer the question 'what did we get for this money?'
The fix is to build the CBS jointly with the scheduler before the GMP is finalized. Granularity should follow the eighty-twenty rule: the twenty percent of activities that drive eighty percent of cost get individual budget lines; the rest can be aggregated. Anything more granular than that creates tracking overhead without improving decisions.
Control #2: A Change Order Process With Real Gates
Most change order processes look rigorous on paper and behave loosely in practice. The contractor submits a Potential Change Order, the owner's team negotiates, the change is approved under time pressure, and only later does anyone notice the contingency line is depleted. The fix is to install hard gates at three points: entitlement review before pricing, pricing review before approval, and a monthly contingency reconciliation that forces the steering committee to acknowledge what has been spent.
Entitlement review asks one question: does the contract entitle the contractor to additional compensation for this work? If the answer is no — because the work is within the original scope, or because the contractor accepted the risk in the GMP — the conversation ends there. Pricing review then evaluates quantum: are the labor hours, equipment rates, materials pricing, and markups consistent with the contract and with market benchmarks? Only changes that pass both gates are forwarded for approval.
Control #3: Forecasting That Looks Forward, Not Backward
Budget-to-actual reporting tells the steering committee what already happened. It does not tell them what is about to happen. Forward-looking forecasting uses earned value metrics — Cost Performance Index, Schedule Performance Index, To-Complete Performance Index — to project where the program will land if current trends continue, and pairs those projections with a contingency burn-down curve that shows whether reserves are adequate for the risks remaining in the schedule.
The discipline that makes this work is monthly variance explanation. Every line on the forecast that has moved more than a defined threshold from the prior month must have a written explanation from the cost engineer. Over time, those explanations build institutional memory about what drives variance on the program — and that memory is what allows the next forecast to be more accurate than the last.
Control #4: Procurement Validation Against Budget Assumptions
When a CM at-Risk firm buys out the trade packages, the spread between the budget assumption and the actual subcontract value is one of the earliest and most reliable indicators of where the program will land. Tight buyout — say, within two percent of budget — generally signals a well-estimated GMP. Loose buyout — five percent or more, with multiple packages over budget — signals that the GMP was optimistic and the contingency line is going to absorb the difference.
The control here is a buyout log, updated weekly, that tracks every package against its budget assumption with a variance percentage and a written explanation for any variance over a defined threshold. The log is reviewed monthly by the steering committee. Buyout discipline is one of the few project controls that produces actionable information in the first hundred days of construction, when there is still time to course-correct.
Control #5: Integrated Schedule and Cost Reporting
Schedule slippage almost always carries cost implications, but on most programs the schedule report and the cost report live in separate documents reviewed by separate audiences. The result is that the steering committee sees a yellow schedule indicator and a green cost indicator in the same packet, without anyone explaining that the yellow schedule will turn the cost indicator red within sixty days.
Integrated reporting fixes this by presenting schedule variance and forecast cost impact on a single page, with critical-path activities highlighted and the cost consequence of each delay quantified. The reporting cadence should match the decision cadence of the steering committee — typically monthly for the full report, with a one-page exception report any week a critical-path activity slips by more than five days.
Making the Controls Stick
Installing the controls is the easy part. Making them stick across a multi-year program — through staff turnover, contractor changes, scope additions, and the inevitable pressure to relax discipline when the project is going well — is the hard part. The organizations that succeed share three habits.
First, the controls are owned by a named individual on the owner's side, not delegated to the CM. The CM may produce the data, but the owner's controls lead validates it, presents it to the steering committee, and is accountable for the integrity of the reporting. Second, the controls are reviewed at a fixed monthly cadence that does not move for any reason other than a board-approved holiday. Third, exceptions are documented in writing and escalated; nothing material is decided in a hallway conversation that does not later show up in the meeting minutes.
Cost overruns are preventable. They are prevented by governance, not by software, not by spreadsheets, and not by hope. The five controls above, applied with discipline by a senior owner's team, will not eliminate every variance — but they will ensure that no variance becomes a surprise.