Almost every contractor carries a contingency in their estimate. Almost none of them can tell you, mid-job, how much of it is left. That gap — between the number you put on paper and the number you actually track — is where profit quietly disappears. A contingency is supposed to absorb the unknowns so your margin survives. Instead, it usually becomes an invisible slush fund that gets nibbled to zero by unbudgeted costs nobody flagged.

This is a discipline problem, not a math problem. Sizing a contingency is easy. Protecting it takes a system. Here is how experienced contractors treat contingency as a controlled reserve instead of a hope.

What a Contingency Actually Covers (and What It Doesn't)

A construction contingency is money set aside for costs you know will happen but can't price precisely yet. Hidden rot behind a wall. A footing that hits rock. A material substitution when the spec'd product goes on backorder. These are known unknowns — risks inherent to the work.

What a contingency is not: it is not a budget for scope the owner adds later. That is a change order, and it should be billed separately at full markup. The single fastest way to destroy a contingency is to let it silently swallow owner-driven scope changes that should have generated new revenue.

If you're spending contingency on work the owner asked for after signing, you're financing their project with your profit. Write the change order instead.

Draw a hard line between the two. Contingency absorbs your estimating risk. Change orders capture their scope additions. Blur that line and you'll finish the job wondering where the money went.

How to Size It: Percentages by Risk, Not Habit

Most contractors slap a flat 10% on everything out of habit. That's lazy in both directions — too much on a job you know cold, too little on a gut renovation full of surprises. Size contingency to the actual risk profile of the work:

  • New construction, complete plans — 3–5%. You control most variables and the drawings are dialed in.
  • Renovation with known conditions — 7–10%. You've opened walls or have a solid inspection, but old buildings hide things.
  • Gut renovation or historic work — 12–20%. Every wall is a question mark and code upgrades stack up fast.
  • Design-build or incomplete drawings — 10–15%. Scope will shift as details resolve; price the ambiguity.

The percentage should ride on the hard cost base — labor, materials, subs, equipment — not on the total contract including your overhead and profit. Applying contingency to your own markup inflates the number and hides how thin your real buffer is.

On cost-plus and GMP jobs, contingency belongs to a specific party depending on the contract language. On a GMP, spell out in writing who keeps unused contingency at closeout. That one clause can be worth tens of thousands, and it's the first thing owners try to claw back.

Track Every Draw Against It in Real Time

Here is the part almost everyone skips. A contingency you don't track is already gone. The moment you approve an unplanned cost, you need to log it against the reserve — date, amount, reason, and running balance. Treat it like a checkbook, not a savings account you never open.

The mechanics matter. Build a simple contingency ledger with four columns: what happened, how much it cost, which cost code it hit, and what's left. Update it the day the cost is committed, not the week you reconcile. A contractor who checks their contingency balance every Friday makes different decisions than one who discovers the reserve is empty at final billing.

You can't protect a number you never look at. A contingency reviewed weekly behaves; one reviewed at closeout is already spent.

This is exactly the kind of running tally that field-first tools handle well. Apps like TrestleBook let you log a contingency draw from the jobsite the moment a surprise cost hits, so the balance is current before you've even left the site — no waiting until you're back at the office and the detail has gone fuzzy. The same job-costing discipline that keeps freelancers and small operators solvent applies here; if you also do independent or side work, Stintly brings that track-every-dollar habit to self-employment and small-business finance.

Ready to put this into practice? Download TrestleBook Free — it’s free and works offline.

Require Approval Before Contingency Gets Touched

On a solo operation you're both the requester and the approver, but the discipline still holds: no contingency draw without a written reason. On a team, make it a rule that no one spends contingency without sign-off from whoever owns the budget. Field crews are optimists — they'll dip into the reserve to solve a problem fast rather than flag it and price it properly.

A lightweight approval gate does two things. It forces someone to ask whether this cost is genuinely a contingency item or actually a change order in disguise. And it creates a paper trail, so when the reserve is 60% gone by the halfway point, you can see exactly which decisions drained it.

  • Written trigger — a one-line note explaining why the reserve is being tapped.
  • Cost-code assignment — charge the draw to the trade it belongs to, so your job-cost history stays honest for the next bid.
  • Change-order check — before approving, confirm this isn't owner-requested scope that should be billed.

The change-order check is the money-saver. Contractors routinely burn contingency on work they could have billed. Every draw should survive one question: would the owner pay for this if I put it in front of them? If yes, it's a change order, not a contingency item.

Watch the Burn Rate, Not Just the Balance

A balance tells you where you are. A burn rate tells you where you're headed. If you've spent 70% of your contingency at the 40% completion mark, you don't have a spending problem — you have an estimating problem that's about to eat your profit unless you act now.

Compare contingency spent against percent complete every time you run a billing cycle. If the two are roughly aligned — 40% spent at 40% done — you sized it right and you're managing it well. If contingency is outrunning progress, stop and diagnose. Usually it's one of three things:

  • Underbid scope — a trade was priced too lean and every task is running over.
  • Scope creep disguised as risk — owner changes are hitting the reserve instead of change orders.
  • Poor documentation upstream — you didn't catch conditions during estimating that a better site visit would have flagged.

Catch it at 40% and you have room to tighten up, renegotiate, or write the change orders you missed. Catch it at closeout and all you can do is explain the loss. This early-warning habit is the same one that separates profitable operators from break-even ones — landlords managing rental turnovers run the identical playbook, and tools like KeyLoft exist because tracking spend against a fixed budget is universal to anyone managing property and repair costs.

Roll Unused Contingency Into Documented Profit

When a job finishes with contingency left over, that money is profit — but only if the contract says so and only if you can prove you managed it. On fixed-price and lump-sum work, unused contingency is simply yours; it was baked into your price. On cost-plus and GMP, the leftover often goes back to the owner or gets split, depending on what you negotiated.

Either way, document the closeout. A clean contingency ledger showing every draw, every reason, and the ending balance does three things: it justifies your final billing, it defends you if the owner disputes charges, and it becomes estimating gold for the next job. The costs that hit your contingency this time are the line items you underbid — feed them back into your next estimate and your buffer gets more accurate every project.

Your contingency ledger is next year's estimating data. The surprises that drained this job are the line items to price properly on the next one.

This is where a running digital record pays off long after the job closes. When your draws are logged with cost codes as you go, closeout is a five-minute export, not a weekend of reconstructing what happened from memory and crumpled receipts. TrestleBook keeps that history attached to the job so the data survives past final payment and informs the next bid.

A Simple Contingency Discipline You Can Start Today

You don't need software or a system overhaul to fix this. You need a habit. Here's the minimum viable discipline that will protect margin on your very next job:

  1. Size it to risk — pick a percentage that matches the actual uncertainty, not your default 10%.
  2. Set the base honestly — apply it to hard costs, not to your marked-up total.
  3. Log every draw the day it happens — date, amount, cost code, reason, running balance.
  4. Screen each draw — is this really contingency, or a change order I should be billing?
  5. Compare burn to progress weekly — if contingency outruns completion, diagnose immediately.
  6. Reconcile at closeout — capture what you underbid and feed it into the next estimate.

The contractors who stay profitable aren't the ones who never hit surprises — every job has surprises. They're the ones who reserved for those surprises deliberately, watched the reserve like a hawk, and refused to let owner scope quietly drain a fund meant for their own risk. Treat your contingency as a managed account with a balance you check, an approval you require, and a closeout you document. Do that, and the number on your estimate stops being wishful thinking and starts being the margin protection it was always supposed to be.