Design release is treated as a schedule milestone. It should be treated as a readiness gate.
The distinction matters because a drawing set is a record of decisions. Where a decision has not been made, the design team makes an assumption in order to keep drawing. That assumption propagates. It gets coordinated against, priced, procured, and eventually built — and when the real decision arrives, the cost of reversing it has moved from a meeting to a change order.
Starting design early does not compress the program. It moves the cost of indecision downstream, where it is paid at construction rates instead of design rates.
The decisions that must close first
The list is shorter than most owners expect.
Program. What the asset must do, at what capacity, with what redundancy, and what it must be capable of becoming later. Expansion optionality is a design constraint, not an afterthought.
Delivery model. Whether the work is procured as a single contract or divided; whether design responsibility sits with the owner or the contractor; how risk is allocated. The contracting model determines what the drawings are for, and drawings produced for the wrong model have to be redone.
Utility strategy. Power, water, and discharge — capacity, route, and the permitting pathway for each. These constrain the site layout before they constrain anything else.
Redundancy and resilience. The concurrent maintainability standard, the fault tolerance target, and what the business will actually pay for. Redundancy decided after the utility design is fixed is bought at the worst available price.
Budget basis. Not the number, but what the number was built from: which scope, at what maturity, with what contingency, on which escalation assumption.
How an unresolved decision propagates
A single open question does not stay a single open question.
Suppose redundancy is unresolved. The electrical designer assumes a configuration in order to size the distribution. That sizing determines the switchgear footprint, which determines the electrical room dimensions, which determine the building grid, which the structural engineer then designs to. The mechanical designer coordinates plant space against that grid. Procurement issues the long-lead inquiry against that switchgear specification.
By the time the redundancy decision is actually made, four disciplines and one purchase order depend on the assumption. The decision is no longer a decision. It is a change.
The cost of a decision is not what it costs to make. It is what it costs to unmake everything built on the assumption that stood in for it.
What it costs at each stage
The multiplier is well understood and consistently ignored.
Reopened at concept, a decision costs meeting time. Reopened at sixty percent design, it costs redesign fees across every affected discipline plus schedule. Reopened at tender, it costs a re-price and usually a program extension. Reopened in the field, it costs the redesign, the removal of work already installed, the replacement at change-order rates, the delay, and — where the contractor is already behind — the leverage that comes with a legitimate claim.
The same decision, at four stages, differing by orders of magnitude.
A gate an owner can enforce
The gate has to be specific enough to fail.
Before design is released beyond concept, the owner should be able to produce a signed program brief; a selected delivery model with the risk allocation written down; validated utility capacity with a route and a permitting pathway; a redundancy standard the business has approved and priced; and a budget with its basis of estimate attached.
Where any of those is missing, design can still proceed — but it should proceed knowing which assumption it stands on, who owns closing it, and by when. An assumption that is named, owned, and dated is a manageable risk. An assumption that has quietly become a drawing is not.
