FIXING AN OVERHEAD RATE YOU HAVE ALREADY BID.
Fixing an overhead rate is four moves, and only the last one touches a bid. Recode the chart of accounts into the eight overhead categories the CONTROL Book uses. Take direct job expense back out of overhead, which is the nine cost types belonging to a specific job. Settle the borderline roles in writing, because a project manager dedicated to a few jobs at a time is a job cost while one who floats across seven or more is overhead, and owner compensation belongs in overhead at market rate. Then recalculate on a rolling twelve month average, load it into the estimate, and deal with the backlog you already sold at the old rate. Why the rate was low before anybody miscalculated it's a separate argument, and this page starts after you've accepted it.
The part nobody warns you about is the backlog. A corrected rate applies to work you haven't bid yet, and jobs already sold at the old rate keep running at the old rate until they close. So in the first month after the correction the reports read worse before they read better: the recode moves cost between categories, the new rate loads more overhead into every fresh estimate, and the old backlog still carries the shortfall it was priced with. Companies that abandon the repair usually do it in that window, reading old work as evidence the new number is wrong. Knowing the window exists is most of getting through it.
WHAT IT MEANS.
An overhead rate repair is the recoding and recalculation sequence that moves a company off an inherited bid rate and onto a rate computed every month from its own eight overhead categories.
The rate is total overhead divided by revenue, so a repair has two halves that get confused with each other. The numerator is a coding problem, meaning which costs are in overhead at all. The denominator is a forecasting problem, meaning which revenue you divide by. Fix one, skip the other, and you get a new number that's wrong in a new direction.
The starting point for most subcontractors is the ten percent overhead and five percent profit carried in a bid by convention. The CONTROL Book treats that pair as an assumption to be replaced rather than a convention to be respected, because a bid built on an overhead rate half the size of the real one loses money on the day it's won. What replaces it's your own number, recomputed monthly.
FOUR WAYS A REPAIR COMES UNDONE.
The rate gets recalculated and the coding doesn't
Recomputing the same categories gives you a tidier answer that's still the wrong one. If supervision, shop time, small tools and direct job expense are sitting wherever they first got posted, the numerator is built from categories that don't hold what their labels claim. The recode comes first and the arithmetic comes second, and doing those two in the other order wastes the month.
Only the numerator gets fixed
Overhead as a share of revenue moves every month, because the costs in the numerator hold still and the revenue underneath them doesn't. A rate computed in a busy year is too low for a slow one. Repairing the cost categories and then dividing by last year's revenue leaves half the defect in place, and the half left behind is the one that moves with the market.
The corrected rate reaches the estimate and nothing else
A rate loaded into the estimating template and left out of the monthly reporting drifts back inside a year, because nothing checks it. The two have to move together: the same number in the bid and in the report, recomputed on a rolling twelve month average at every close, so a change in either one is visible in the other.
The backlog gets ignored
Work already sold at the old rate runs at the old rate until it closes, and for a subcontractor carrying several months of backlog that's several months of jobs which can't absorb the corrected overhead. Owners who read the first months after a correction as evidence it failed are reading old work. The repair plan has to say what happens to that backlog, or the reporting will look like the fix made things worse.
FOUR MOVES, IN THIS ORDER.
The eight are office requirements, software subscriptions, administrative expenses, employee benefits and development, owned equipment idle time and normal maintenance, insurance, non-direct job employees, and everything else. On a repair the work is reclassification rather than construction, so the last twelve months of history has to move into the new categories too, or the rolling average compares this month's coding against last month's. Non-direct job employees is where most of the movement happens, because it holds the estimating team, the safety manager, in-house accounting, business development, the C-suite and the owner.
Superintendents, non-working foremen, project managers, assistant project managers, safety, job trailers, storage containers, permits, and legal are required to execute a specific job and show no physical progress on it. Leaving those nine cost types in overhead inflates the pool and understates what the work costs. Moving them out lowers the rate and raises job cost in the same close, which is one correction reading as two opposite changes, and it's the step that makes an owner think something broke.
A project manager dedicated to a few jobs at a time is a job cost. One who floats across seven or more, or who runs company-wide functions, is overhead. Owner compensation belongs in overhead at market rate for the work being done and not at whatever the draws happen to total. Write both decisions down per person and date them, because a rate can't be compared month to month if the coding rule changes with whoever posts the entry.
The corrected rate goes on a rolling twelve month average, recomputed at close and loaded into the estimate, and it appears in the monthly report beside the jobs it priced. Work sold at the old rate gets tracked as its own group until it closes, so the report can show corrected bids and old backlog apart from each other instead of averaging them into one margin. Thirteen months of history in the report is what lets this month be compared against the same month a year ago while both groups are still running.
Not one site-wide figure. A painting contractor and a tunnel contractor don't run the same cost structure, so the comparison is your own trade at your own revenue band, which the per-trade overhead benchmark pages carry. The percentages that circulate as construction overhead targets are whole-industry averages, and treating an average as a target is the error those benchmark pages exist to correct.
THE OUTPUTS, NAMED.
FLAT MONTHLY FEE. NO SURPRISES.
Three tiers, priced by your trailing twelve month revenue. Which one you're in depends on how much of the work you want off your desk. No hourly billing, no payroll, and no add-ons.
| Last 12 months revenue | Monthly fee |
|---|---|
| Up to $1M | $1,900 to $2,900 |
| $1M to $3.5M | $2,600 to $3,900 |
| $3.5M to $6.5M | $3,800 to $5,700 |
| $6.5M to $9.5M | $5,100 to $7,100 |
| $9.5M to $12.5M | $6,100 to $8,500 |
| $12.5M to $15.5M | $7,400 to $11,000 |
| $15.5M to $18.5M | $9,400 to $13,500 |
| $18.5M+ | Quoted individually |
Range reflects the three tiers below. Which one you're in depends on how much of the work you want off your desk. No payroll. No hidden line items.
You stop guessing.
You get the CFO work. Job costing built against the way you estimate, a 13 week cash forecast, monthly WIP, and a meeting every month that ends in decisions rather than a report.
Your bookkeeper keeps doing the books.
You stop touching the books.
Everything in Core, and we run the bookkeeping and the controllership as well. Nobody in your office is answering coding questions or chasing a reconciliation at month end.
We do the books. No payroll.
Every job shows its margin while it's still running.
Everything in Executive, plus the job costing and WIP platform set up, loaded with your cost codes, and managed for you every month. You never have to learn it.
We do the job costing.
FREQUENTLY ASKED.
