How should contingency shrink as a build moves from design to construction?

Your distinction between genuine unknowns and owner-requested upgrades is the important one. I’d take it further by separating the reserve into three accounts rather than simply reducing one percentage:

- **Design-development allowance:** covers incomplete drawings, allowances, coordination gaps, and scope clarification. It should reduce as selections are finalized, bids are reconciled, and the construction documents become reliable.
- **Construction unknowns reserve:** covers concealed conditions such as soil, drainage, utilities, and excavation. This should remain available through the relevant work, with releases based on inspections or completed site-risk milestones.
- **Owner-change budget:** covers elective changes, upgrades, or late selections. It shouldn’t be described as contingency, and using it should require an explicit cost and schedule approval.

Each account needs its own approval rule and tracking. For example, a bid omission might be charged to design development if the documents were incomplete, but an added lighting package after contract award belongs to the owner-change budget. Moving unused money between accounts should require a documented decision, not happen automatically when the overall percentage is reduced.

That structure makes a smaller construction reserve defensible without creating pressure to consume it before the project is actually through the site-risk work.

I’d add a fourth account for contract risk: uncovered escalation, allowance overruns, exclusions, and qualifications that survive bid reconciliation.

This isn’t the same as design-development uncertainty. A cabinet allowance may be clearly stated in the contract but still fall below the required specification. Likewise, a subcontractor’s “by others” exclusion can create a real cost even when the construction documents were complete.

Before reducing this reserve, I’d check:

- Which prices are fixed, and for how long?
- Are escalation clauses, material substitutions, and procurement delays addressed?
- Do allowances include labor, tax, delivery, and installation?
- Have exclusions and scope gaps been assigned in writing?
- What notice deadlines apply to price or schedule claims?

The reserve can shrink as procurement commitments are made, but it shouldn’t automatically disappear just because the plans are finalized. If the contract genuinely transfers a risk to the builder, release that portion. If it only appears covered until an allowance or exclusion is tested, keep it separate from the construction-unknowns reserve.

The proposed release gate should be tied to the cost-to-complete forecast, not just to milestones such as “bids received” or “construction documents issued.” A project can reach those milestones while still carrying meaningful allowance and procurement exposure.

For each review, reconcile the contingency ledger against:

- Committed costs and accepted change orders
- The latest forecast for remaining work
- Allowances still open, including their labor, tax, delivery, and installation components
- Identified but unpriced risks, such as exclusions, escalation clauses, or unresolved site conditions

Only the portion no longer represented in that forecast should be released. For example, finalizing the drawings may reduce design-development uncertainty, but it doesn’t justify releasing money for a lighting allowance that hasn’t been selected or a material price that isn’t locked. This also creates a useful audit trail: the reserve shrinks because a specific risk was priced, transferred, or eliminated, rather than because the project passed an arbitrary phase date.

Carol’s cost-to-complete test is much stronger than releasing reserve merely because bids were received or construction documents issued. I’d add one audit control: every release should identify the specific risk-register entry that was closed, along with the evidence supporting that decision.

That evidence might be a reconciled subcontractor quote, a written clarification assigning an exclusion, a signed allowance selection, or site verification showing that the relevant excavation or drainage condition has been resolved. If a lighting allowance is still open, for example, the design-development reserve may be reduced, but the amount tied to that allowance shouldn’t be released until the selected fixtures, installation, tax, and delivery are reflected in the forecast. This makes the release traceable and helps prevent the same risk from quietly reappearing later as a change order.

An open lighting allowance can create schedule exposure even if the eventual fixture cost stays within budget. A late selection may affect availability, delivery timing, or installation sequencing.

Keep that exposure visible separately from the allowance amount in the cost-to-complete forecast. It does not automatically require a contingency draw, but it should not be released just because the expected fixture price fits the allowance. Once the selection, lead time, and installation requirements are documented, any remaining schedule-related exposure can be priced or closed.

The reserve doesn’t have to fall by a standard percentage at each phase. I’d estimate the plausible cost of each unresolved risk, weighted by how likely it is, then add those exposures to set the dollar reserve. As bids firm up or site conditions are verified, reduce only the exposure those facts actually eliminate. That way the percentage can change as the budget changes, but it isn’t driving the decision.

Would a range for likelihood and cost be more useful than a single estimate for uncertain items, such as concealed site conditions?