Contingency Drawdown: Are You Burning It Faster Than You Are Building?

Contingency is not a reserve sitting in a drawer. It is a curve that should flatten as risk retires. The useful question on a capital project is not how much contingency is left, it is whether the burn rate is outpacing percent complete. A project 30 percent built that has consumed 60 percent of contingency is already over budget; it just has not told anyone yet. Most owners cannot answer that question because contingency use is recorded as a budget line transfer rather than tracked as a drawdown against progress.

Why contingency balance alone tells you nothing

A contingency line reading "$1.8M remaining" is a number without a denominator. It is meaningful only against two other figures: what the project has actually completed, and what risk remains unretired in the scope ahead.

The failure mode is specific. The project manager approves a $140,000 unforeseen conditions change, and the cost report shows a transfer from contingency into the sitework line. The next month a $90,000 design clarification moves into the MEP line. Three months later the contingency line has dropped by $600,000, and the trade lines have all grown. Every individual move was approved and defensible. Nobody looked at the cumulative rate.

By the time the pattern is visible in a quarterly report, the remaining scope is the scope that carries the most risk, and the money that was supposed to absorb it has gone to problems that surfaced early. This is one of the most common paths to a budget overrun that nobody saw coming, and it belongs to the same family of failures covered in how enterprise capital programs go over budget.

How much contingency a construction project should carry

Contingency should be derived, not borrowed. The percentage is an output of a risk exercise, not an input to the budget.

Practical ranges most owners work from, before adjustment:

Project condition	Typical owner contingency	Why
New build, clean site, 100% CDs at GMP	3–5% of construction cost	Documents complete, few unknowns in the ground
New build priced at design development	5–8%	Drawings still maturing between DD and CD
Renovation, occupied building	10–15%	Existing conditions, phasing, after-hours work
Adaptive reuse, historic or unsurveyed structure	15–20%+	Structural and hazmat discovery is near certain
Fast-track, overlapping design and construction	8–12%	Scope defined after trades are bought

Adjust from there against the specific risk register. A project with a known utility conflict, a long-lead switchgear order and an unresolved AHJ interpretation carries more quantified exposure than the percentage suggests. Price those three items, add them, and compare the total to the contingency. If the named risks already exceed the reserve, the number is wrong regardless of what the benchmark says.

Keep design development allowance and escalation out of contingency. Design allowance covers documents maturing. Escalation covers market movement on unbought scope. Contingency covers events. Blending the three produces a single number that cannot be forecast, because you cannot tell which pressure is consuming it.

Owner contingency vs GC contingency

These are different instruments with different owners, different triggers and different accounting. Confusing them is how an owner ends up paying twice for the same risk.

	GC contingency	Owner contingency
Sits in	The contract sum / GMP	The owner's project budget, outside the contract
Covers	Contractor coordination gaps, minor rework, buyout variance, estimating error	Scope changes, design changes, unforeseen conditions, owner-directed additions
Who draws	GC, usually with notice to the owner	Owner or owner's representative
Effect on contract price	None	Increases contract sum via change order
At closeout	Often shared per the savings clause	Returns to the owner or funds scope adds

Two provisions decide whether GC contingency actually protects the owner. First, the definition of qualifying use: if the contract permits draws for "costs not otherwise recoverable," the contractor can absorb scope that should have been a change order, or conversely charge the owner for what should have been contractor risk. Name the categories explicitly. Second, the reporting obligation: require a monthly log of GC contingency draws with description, amount and date, delivered with the pay application. Without it, the owner learns the GMP contingency is exhausted at the moment a change order arrives that should have been covered by it.

Owner contingency should never be shown to the contractor as a line item in the project budget. It is not a target.

Building a contingency drawdown schedule

The drawdown schedule is a planned curve: how much contingency should remain available at each milestone. It is set at buyout and does not move.

A workable curve for a conventional build:

• Buyout complete: 100% available
• Foundations and underground complete: 80–85% remaining
• Structure topped out: 55–65% remaining
• Building dry-in / envelope closed: 30–40% remaining
• MEP rough-in complete and inspected: 20–25% remaining
• Substantial completion: 10–15% remaining
• Final completion and closeout: release

The shape matters more than the exact figures. The curve is front-loaded in release terms because geotechnical, existing-conditions and structural risk retires early. Once the building is dry and rough-in is inspected, the population of events that can surprise you shrinks sharply, and so should the reserve you are holding against them.

Set the curve against your own project's risk register rather than copying this one. A hospital renovation with occupied-floor phasing will retire risk far later than a warehouse shell. An owner who releases contingency on the warehouse curve while building the hospital will be holding nothing when the phasing conflicts start.

Then plot actual against planned every month. Two lines, one chart. When the actual line crosses below the planned line, you have a forecast problem to solve while the remaining scope is still unbought, which is the only moment the problem is cheap.

Who approves the use of contingency

Write the authority matrix before construction starts, and put it in the project execution plan rather than in someone's inbox.

A defensible structure:

Draw amount	Authority	Required documentation
Under $25,000	Project manager	Cost backup, category coded, logged within 5 days
$25,000–$150,000	Owner's representative or program director	Backup, schedule impact statement, category
$150,000–$500,000	VP of development / asset owner	Above, plus revised cost-to-complete forecast
Above $500,000, or any pro forma impact	Investment or executive committee	Above, plus written risk memo and funding source

Thresholds scale with project size; the tiering is the point. The second rule matters as much as the first: every draw is coded to a category. Unforeseen conditions. Design error or omission. Owner-directed scope. Regulatory or AHJ. Market or escalation. Coordination between trades.

Category coding is what converts a contingency log into intelligence. Four months in, if 70 percent of draws are coded "design error or omission," the problem is not the contingency, it is the document set, and the remedy is a coordination review before the next trade package goes out. Without coding, every draw looks like bad luck. The approval discipline that applies here is the same discipline that governs change order management, and the two processes should share a single approval record.

Tracking contingency on a capital project

The tracking requirement is not complicated, but it is rarely met, because contingency lives in the cost report while the events that consume it live in the change order log, the RFI log and the field.

What a working contingency register holds, per draw:

• Date of authorization and authorizing party
• Amount, and the running balance after the draw
• Category code
• Linked source document: the change order, the RFI response, the field directive, the ASI
• Percent complete at the date of the draw
• Schedule impact, if any

That sixth field, percent complete at the date of the draw, is the one almost nobody captures, and it is the one that makes the burn curve possible. Without it you have a list of expenditures. With it you have a rate.

The same register answers the question a lender or an investment committee actually asks, which is not "how much is left" but "what did the last $2 million buy, who approved it, and what does that tell us about the remaining $4 million." An auditable chain from draw to source document to approver is the difference between a defensible funding request and a line item someone has to reconstruct from email. That record-keeping discipline is covered more broadly in audit trails in construction administration.

Contingency tracking fails structurally when cost, commitments, change orders and progress sit in separate systems, because the comparison that matters requires all four at once. Platforms like INGENIOUS.BUILD hold the budget, the commitments, the change orders and the schedule position in one record, so the burn curve is a view rather than a monthly reconstruction in a spreadsheet.

Bring the curve to every OAC meeting. Two lines, planned against actual, with the month's draws listed by category underneath. It changes the conversation from reporting to forecasting, which is the only version of the conversation worth having. If you want a sharper set of prompts for those meetings, start with the questions every owner should ask at an OAC.