The $600K Leak in Every $20M GC

Take a general contractor doing $20 million a year. Profitable on paper. Winning enough work to stay busy. The owner would tell you things are good — maybe not great, but good.

Now run the numbers on what's leaking out of the operation every year through three system failures that nobody tracks because they've been normalized. Stale estimates. Missed change orders. Superintendent burnout from admin overload.

Add them up. You're looking at $600,000. That's not a rounding error. That's a system failure with a line item.

Leak #1: Stale Estimating Data

Most GCs in the $15–25M range estimate using some combination of historical costs, subcontractor quotes from the last similar job, and gut feel. The problem isn't that the estimator is bad — it's that the data is old.

Material costs shift 5–15% quarter over quarter in volatile categories. Labor rates in South Florida have moved 8–12% year over year for the past three years. If your estimate is built on numbers from two projects ago, you're not estimating — you're guessing with a spreadsheet.

The result: you either lose the bid because you're 8–12% high, or you win it because you're 8–12% low and eat the difference on the back end. Neither outcome is a system working. One costs you revenue. The other costs you margin.

On a $20M book of business, estimating drift accounts for roughly $200K in lost opportunity or eroded margin annually. That's conservative — it assumes you're only off on a third of your bids and the variance averages 3%. In practice, it's usually worse.

Leak #2: Late Change Order Capture

This one is quieter, and it's the most expensive.

A superintendent sees a field condition that triggers a change order — unforeseen site work, a spec conflict, an owner-requested modification. He knows it's a change. He makes a mental note. Then he gets pulled into a coordination issue, takes three calls from subs, and by the end of the day the note never gets written up.

Three weeks later, someone in the office asks about the cost variance on that phase. Now the super is reconstructing the change from memory, the documentation is thin, and the GC either submits a weak CO that gets negotiated down or doesn't submit one at all.

The industry average for change order capture rate among GCs under $50M is roughly 60–70%. That means 30–40% of legitimate changes go unrecovered. On a $20M GC running projects with typical 8–12% change order volume, that's $1.6–2.4M in total change orders, with $500–950K going uncaptured.

Call it $250K in net margin erosion after you account for the ones that get partially recovered. That's money the GC earned in the field and lost in the office.

Leak #3: Superintendent Running Admin

The superintendent is the most expensive person on the jobsite — not by salary, but by impact. When the super is running the field, the project moves. When the super is doing paperwork, the project drifts.

Most GCs in this range don't have enough project engineers or field admins to offload daily logs, RFI tracking, submittal follow-ups, and schedule updates. So the super does it. An hour here, two hours there. By Friday, 30–40% of the super's week went to admin.

The cost isn't just the inefficiency. It's the burnout. Superintendent turnover in commercial construction runs 18–25% annually. Replacing a superintendent costs $80–120K when you factor in recruiting, onboarding, project disruption, and the learning curve on active jobs.

A $20M GC running four to six active projects with two to three supers is replacing one every 18–24 months. That's $150K in annualized turnover cost — driven primarily by a role that was never designed to sit at a desk.

The Total

System Failure	Annual Cost
Stale estimating data	~$200K
Late change order capture	~$250K
Superintendent admin overload	~$150K
Total	~$600K

Three percent of revenue. Gone. Not to bad luck or market conditions — to systems that don't exist.

What It Looks Like When Systems Are in Place

Flip each one.

Estimating pulls from live cost data — actual buyout numbers from the last 90 days, updated sub rates, material indices that refresh quarterly at minimum. The estimator still applies judgment, but the baseline is current. You bid tighter and win at margins you actually want.

Change orders get captured at the point of recognition. The super logs the field condition on his phone — photo, two sentences, tagged to the cost code. The system routes it to the PM for pricing. By the time the owner asks about the variance, the CO is already documented and submitted. Recovery rate goes from 65% to 90%+.

The superintendent runs the field. Daily logs auto-populate from scheduled entries and field inputs. RFI tracking lives in a shared system the PM monitors. The super checks status in two minutes instead of building spreadsheets for two hours. Turnover drops because the role becomes what it was supposed to be — field leadership, not data entry.

The Point

Nobody wakes up and decides to lose $600K. It happens incrementally, across every project, every quarter, through gaps that individually seem minor. A stale line item here. An undocumented change there. A superintendent who quit because the job stopped being about building.

The fix isn't more people. It's not more software licenses. It's a system design question — how does information move through your operation, and where does it stop?

That's the diagnostic. The $600K is just what happens when nobody runs it.