Contingency: build it from a list of risks, not from a habit percentage

The math, worked on a real budget

A $340,000 whole-house renovation on a 1958 property. Rather than applying 15% to the total, price the actual risks:

• Knob-and-tube or ungrounded wiring beyond the planned scope: $14,000 if found, 60% likely — $8,400.
• Cast iron drain line failure under the slab: $11,000 if found, 35% likely — $3,850.
• Structural repair at the rear wall where settlement is visible: $22,000 if confirmed, 45% likely — $9,900.
• Asbestos in flooring mastic or duct insulation: $6,500 abatement, 50% likely — $3,250.
• Rot at the sill plate behind the failed siding: $9,000, 40% likely — $3,600.

Total expected value: $29,000, which on $340,000 is 8.5%. But expected value is not the same as sufficiency — if three of the five occur at the higher end, the exposure is closer to $47,000. A defensible contingency here is $34,000 to $40,000, or 10% to 12%, with the reasoning documented.

The difference between presenting "we carry 15% contingency" and presenting that list is the difference between an owner who suspects padding and an owner who understands the building they bought.

Who owns the contingency, and how it gets spent

This is where most contracts are silent and most disputes begin. Three structures exist, and each behaves differently:

• Contractor-held, undisclosed. Common in fixed-price work. You carry it inside the number, you absorb the risk, and anything unspent is yours. Simple, and it works — provided the price is high enough to be defensible if the owner shops it.
• Owner-held, disclosed. The budget shows a contingency line the owner controls. Spending it requires their approval. This is standard in cost-plus and construction management arrangements, and it makes the owner an active participant in risk decisions.
• Shared, with a split of savings. Unspent contingency is divided by an agreed formula. It aligns incentives well and is common in larger negotiated work — but the formula must be written before the project starts, not discovered at the end.

Tracking it so it does not disappear

Contingency that is not tracked does not get spent on contingencies — it gets absorbed into overruns nobody logged. Four practices prevent that:

• A contingency ledger, separate from the cost report. Every draw dated, described, with the reason and the approver.
• A rule that scope changes never come from contingency. If the owner adds a bathroom, that is a change order at full price. Letting scope creep into contingency is how builders finish a job with no reserve and no additional revenue.
• Monthly review of remaining contingency against remaining risk. Having 60% of contingency left at 30% complete is fine; having 20% left at 30% complete is a signal that requires a conversation now rather than at closeout.
• Formal release at milestones. Once the walls are open and the structure is verified, part of the contingency is no longer needed. Releasing it explicitly is honest, and on owner-held contingency it builds enormous trust.

Twelve months, measured

A residential builder doing $4.2 million a year across eight to twelve projects, historically carrying a flat 10% and finishing most jobs with margin erosion nobody could explain.

• Months 1–2: introduced a risk register per project — five named risks, cost, probability — replacing the flat percentage.
• Month 3: contract updated with contingency ownership, approval and remainder language.
• Month 5: contingency ledger started, with scope changes explicitly excluded.
• Month 8: milestone releases introduced on cost-plus jobs, returning unused contingency to owners at framing inspection.
• Month 12: average gross margin per project up 4.1 points, with no price increase.

The margin recovery did not come from carrying more contingency. It came from the rule that scope changes cannot draw on it — the review found that roughly two-thirds of historical contingency spend had actually been unbilled scope additions the owner had requested and the builder had absorbed to avoid a difficult conversation.

The second effect was reputational. The milestone releases — returning money the owner had already committed — generated more referrals in eight months than the company's advertising had in two years.

Explaining contingency to an owner

Owners do not object to contingency; they object to not understanding it. The framing that works has three parts:

• Name the specific risks. "This house was built in 1958 and the panel is original — here is what we do if the wiring is what we expect." Specificity reads as competence.
• State clearly that unspent money is theirs, when the structure is owner-held. It converts contingency from a suspected markup into an insurance policy they control.
• Explain the alternative honestly. A budget with no contingency is not a cheaper project; it is the same project with the surprises arriving as change orders during construction, which is the most expensive and most stressful way to encounter them.

Escalation, schedule and the risks contingency should not carry

Contingency works for conditions you might discover on a specific building. It works badly for two other categories of risk that builders routinely dump into it, and separating them protects the number.

Material escalation. Prices moving between bid and purchase is not an unknown condition — it is a market risk, and it has its own instruments. A price validity period on the proposal, an escalation clause tied to a published index, early purchase and storage of volatile items, or allowances written against specific commodities all handle it more precisely than contingency does. Absorbing escalation into contingency means a lumber move consumes the reserve that was meant for the rot behind the siding, and then both problems are unfunded.

Schedule risk. Weather delays, inspection backlogs and subcontractor availability cost money, but they are managed with float in the schedule and with contract language on excusable delay — not with a cash reserve. A project that runs six weeks long has a general conditions problem, and general conditions should be priced as a duration-based line so that extra time carries its own cost visibly.

There is a third category worth naming because it is the most damaging: estimating error. A takeoff that missed 400 square feet of drywall is not a contingency event. Charging it to contingency hides the error, which means the same error repeats on the next three jobs. Logging it separately as a variance is uncomfortable and is the only way estimating improves.

Five mistakes that make contingency useless

• Applying one percentage to every project. Risk varies enormously by building age and scope definition.
• Letting scope changes draw on it. The single largest cause of margin loss on renovation work.
• Not defining ownership in the contract. Guarantees an argument at closeout.
• Treating it as profit. The first genuine surprise then eats real margin.
• Never releasing it. Holding contingency to the final invoice on owner-held money damages trust for no gain.

Five numbers to manage contingency by

• Contingency as a percentage, by project type. Should vary; if it does not, it is not being calculated.
• Contingency consumed versus percent complete. The early warning indicator.
• Percentage of contingency spent on true unknowns. Anything else is scope leakage.
• Change orders issued per project. A low number on a renovation usually means scope is being absorbed.
• Contingency returned to owners. On owner-held work, this is a marketing asset as much as an accounting one.

Contingency is the line that decides whether a renovation finishes at the margin you priced or somewhere well below it. Build it from a named list of risks rather than a habit percentage, write down who owns it and what may draw on it, keep scope changes out of it entirely, and release what is not needed as the unknowns resolve. Done that way it stops being a cushion you quietly lose and becomes the mechanism that lets you price honest work on old buildings without gambling the job.

Frequently asked questions

How much contingency should a construction project carry?

It should follow the risk rather than a habit percentage. New construction on a clean site with complete drawings typically warrants 3% to 5%, since few unknowns exist beyond weather and material pricing. Renovation of a modern, non-structural building runs 8% to 12% — you will find something behind the drywall, but the building is documented. Renovation of a pre-1980 building justifies 15% to 20%, because of hazardous materials, undocumented alterations, structure that does not match the drawings and systems at end of life. Renovation with structural modification or unknown foundations warrants 20% or more, with the honest caveat that even that may not be enough. And design-build with incomplete drawings should size contingency to the percentage of design completion rather than to the construction type. The more valuable habit is to build the number from a list: name the five things most likely to go wrong on this specific job, estimate each cost, assign a probability and sum — that figure is defensible in a conversation with an owner, which a round 10% never is.

What is the difference between contingency and an allowance?

An allowance is a placeholder for a scope item whose selection has not yet been made — the customer will choose tile, so you carry $9 a square foot. Contingency covers conditions nobody chose and nobody wanted, such as rot behind siding or a failed drain line under the slab. There is a third category people confuse with both: a scope gap, which is something you simply did not price. The test that separates them is straightforward — if it is a decision the owner will make, it is an allowance; if it is a condition you might discover, it is contingency; and if you did not price it, it is a scope gap, which no contingency should be covering. It is also worth saying that contingency is not profit: treating it as hidden margin means the first real surprise consumes your actual margin instead.

Who should hold the construction contingency?

Three structures exist and each behaves differently, but the important thing is that the contract says which one applies. Contractor-held and undisclosed is common in fixed-price work: you carry it inside the number, absorb the risk, and keep anything unspent — simple, and it works provided the price is defensible if the owner shops it. Owner-held and disclosed shows a contingency line the owner controls, with their approval required to spend it, which is standard in cost-plus and construction management and makes the owner an active participant in risk decisions. Shared, with a split of savings, aligns incentives well and is common in larger negotiated work, but the formula has to be written before the project starts rather than discovered at the end. Adding one paragraph to the contract stating who holds it, what triggers its use, who approves a draw and what happens to the remainder prevents more closeout conflict than any other clause.

How do you keep contingency from disappearing into scope changes?

With an explicit rule that scope changes never draw on contingency — if the owner adds a bathroom, that is a change order at full price. This single rule is where most of the money is: in one builder doing $4.2 million a year, a review found roughly two-thirds of historical contingency spend had actually been unbilled scope additions the owner requested and the builder absorbed to avoid a difficult conversation. Alongside it, three practices help: a contingency ledger separate from the cost report, with every draw dated, described and approved; a monthly review of remaining contingency against remaining risk, since 20% left at 30% complete is a conversation to have now rather than at closeout; and formal release at milestones, because once the walls are open and the structure is verified, part of the reserve is no longer needed. Introducing those changes recovered 4.1 points of gross margin with no price increase.

How do you explain contingency to a client?

Owners do not object to contingency; they object to not understanding it, and the framing that works has three parts. Name the specific risks — 'this house was built in 1958 and the panel is original, here is what we do if the wiring is what we expect' — because specificity reads as competence while a round percentage reads as padding. State clearly that unspent money is theirs when the structure is owner-held, which converts contingency from a suspected markup into an insurance policy they control. And explain the alternative honestly: a budget with no contingency is not a cheaper project, it is the same project with the surprises arriving as change orders during construction, which is the most expensive and most stressful way to encounter them. One builder found that returning unused contingency at milestone releases generated more referrals in eight months than two years of advertising.