Board approval is the point where accountability for a capital project changes hands. Before the vote, every assumption is negotiable and every alternative is open. After it, the board owns the consequence of a commitment made mostly on information assembled by the people who wanted the approval.
That asymmetry is the reason approval deserves more structure than it usually gets. The board's leverage is at its maximum in the meeting where its information is at its thinnest — and the remedy is not more diligence in general, but a defined package the project must produce before the question is called.
Approval is the point of maximum leverage
Everything a board later wishes it had required becomes progressively more expensive to obtain after approval. A basis of estimate reconstructed after the budget is public costs a quarter and an argument. A delivery structure revisited after contracts are signed costs a renegotiation. A governance design imposed after the project team is staffed costs a reorganization and some careers.
Before the vote, all of these cost a delay to one meeting. The board that understands this stops treating the approval as the end of a presentation and starts treating it as the one gate it fully controls.
The package to require
Six items make up an approval package a board can actually stand behind. None of them is exotic; what is rare is requiring all of them at the same time, in writing, before the vote.
A budget with its basis attached. A single number is an assertion. The board needs the basis of estimate behind it: what scope the number prices, at what stage of design maturity, on what unit rates, with what escalation assumption, and what is excluded. The exclusions are usually where the next hundred million lives.
A schedule that can show its derivation. The dates in the approval deck should be traceable to the constraints that actually govern them — the interconnection queue position, the permitting sequence, the long-lead equipment order dates, the commissioning duration. A date that cannot be traced to a constraint is a target, and targets have a way of appearing achieved in reporting long after they have failed in fact.
A delivery structure with the risk allocation written down. Design-bid-build, construction management at risk, design-build, EPC — the model chosen allocates cost and schedule risk for everything that follows, and the allocation should be stated on paper, not inferred from the acronym. The board should be able to see which risks the owner retains and what those retained risks could cost.
A contingency with a stated basis. Contingency set as a customary percentage is not analysis. The board should see what identified risks the contingency covers, what it excludes, and under whose authority it is drawn down — because an ungoverned contingency is not a reserve, it is a rounding of the budget upward.
A governance design. Who decides what, in what increments, on whose recommendation, and what escalates to the board itself. If the answer arrives after approval, it will be designed by the delivery organization for the delivery organization's convenience.
The utility and permitting position, validated. For power-intensive projects this is the item most often asserted and least often owned: the interconnection position, the water supply and discharge path, and the permitting critical path, each with a named owner and evidence rather than an encouraging letter.
The questions that test it
A package can be assembled to pass inspection, so the test is not whether the documents exist but whether they survive three questions.
Who owns each assumption? Every load-bearing assumption in the package — the energization date, the escalation rate, the productivity basis — should have a named owner who can defend it. An assumption owned by nobody is a risk assigned to the board.
What would have to be true? For the budget to hold, for the date to hold, for the delivery model to perform — the conditions should be statable in plain language. If management cannot state them, the package is a narrative rather than a plan.
What did the authors of this package have an incentive to believe? Not an accusation; a structural observation. An approval package is built by people whose projects proceed when packages are approved. The board's job is not to distrust them but to know which parts of the story only an outsider would test.
The board's approval is the last decision on the project made entirely on the board's terms. Every decision after it is made partly on the project's.
Conditions worth attaching
Approval does not have to be binary, and the conditions attached to it are the board's way of keeping some leverage after the vote.
Staged release is the strongest. Approve the program; release capital in tranches against defined gates — design maturity reached, orders placed, permits determined — so that the board's next decision arrives while it still has one.
Baseline discipline is the cheapest. The approved cost and schedule baseline changes only with board consent, and reporting always shows the original baseline alongside the current one, on the same page. A project on its fourth baseline is not on plan, and the board should be structurally unable to forget that.
An independent reporting line is the most resented and the most valuable. Some account of the project — assurance reviews, an owner-side representative, periodic independent assessment — should reach the board without passing through the delivery organization it reports on. A board whose only information is the project's self-assessment has approved its own blindfold.
What to refuse
Three things in an approval deck justify sending it back unvoted.
A single-number budget with no basis of estimate. An in-service date that cannot be traced to the interconnection, the permits, and the order book. And the phrase "to be finalized after approval" attached to the delivery structure, the governance, or the contingency policy — because after approval, the board is no longer the customer for those documents.
None of these refusals delays a well-prepared project by more than the time it takes to produce what should have existed anyway. What they delay reliably is the other kind of project — and that delay, at the approval gate, is the cheapest this board will ever buy it.
