Job costing for commercial roofing

What job costing is, and why roofers lose money without it

Job costing is the practice of tracking every dollar a project consumes (labor, material, equipment, subcontractors) against that specific job, and comparing it to what you budgeted and what you have billed. Done well, it answers the only question that matters mid-project: are we still going to make money on this?

Commercial roofing is unusually punishing when you get this wrong. Margins are thin, so a few points of slippage erases the profit on a job. Labor is the most variable cost you have. Weather, crew productivity, and rework can swing field hours dramatically from the estimate. Material prices move between bid and buyout. And projects run for months, which means an overrun you don’t catch in week three is baked in by the time you see it in the P&L. Contractors who "find out at the end" aren’t running job costing; they’re running a post-mortem.

Real job costing is a live, running comparison (budget versus committed versus actual versus forecast) updated from source documents as the job moves. That’s what lets you act while you still can.

Cost codes and cost types: the backbone

You can’t cost a job you can’t categorize. The backbone of job costing is a consistent structure of cost codes and cost types, and every document you generate gets coded to them.

Cost codes are the buckets of work: tear-off, membrane, flashing and metal, insulation, and so on, often numbered in a standard scheme. They tell you where money is going within a job. Cost types classify the nature of each dollar: commonly Labor, Materials, Equipment, Subcontractor, and general requirements. They tell you what kind of cost it is.

The discipline is simple and non-negotiable: every purchase order, every bill, every timecard, every material issue carries a cost code. When a foreman logs hours, they’re coded to a cost code and department. When a PO is cut, its lines are coded. If a document isn’t coded, it can’t land in the right bucket, and an uncoded cost is a hole in your job cost you’ll only notice when the totals don’t reconcile. A good system forces the code at entry rather than letting costs pile up in a miscellaneous pile.

Committed vs. actual: the gap most contractors miss

Here is the single most important distinction in construction job costing, and the one most spreadsheets get wrong: the difference between committed cost and actual cost.

Committed cost is money you’ve promised to spend but haven’t paid yet. The moment you issue a purchase order or sign a subcontract, that money is spoken for. The obligation exists even though no invoice has arrived. Actual cost is money that has actually hit the job: bills you’ve received and burdened labor you’ve incurred.

The trap is watching only actuals. Early in a job, actual costs look reassuringly low, but you may have already committed 80% of a cost code through open POs and subcontracts that simply haven’t been billed yet. If you budget against actuals alone, you’ll happily approve more spending on a code that’s already fully committed, and the overrun only surfaces when the bills land. The open commitment (committed minus what’s been billed against it) is the exposure you have to watch. A proper job cost view shows all of it: original and revised budget, committed to date, actual to date, and the unbilled commitment still coming.

Labor burden: the true cost of an hour

Field labor is your biggest lever and your biggest risk, and it’s where contractors most often understate cost. The wage on the paycheck is not what an hour of labor costs you. The true cost is the burdened rate: base wage plus everything that rides on top of it.

That burden typically includes the employer share of payroll taxes, workers’ compensation insurance (which is expensive in roofing), general liability, and any benefits: health, retirement, training. Add it up and the burden can be a substantial percentage on top of base pay.

A worked example. Say a roofer’s base wage is $30/hour. Add employer payroll taxes, workers’ comp, liability, and benefits, and the loaded burden comes to, say, roughly 40% on top. The burdened rate is about $42/hour, the number your job cost should use. If your estimate and your job cost both used $30, every labor hour is understated by $12, and a job that "looks" on-budget is quietly bleeding. Always cost labor at the burdened rate, and make sure the rate you estimate with matches the rate you cost with.

Why the ledger should be source-document-only

There’s a temptation to "help" a job cost report along with estimates: proration, percentage guesses, an accrual to smooth things out. Resist it. The value of a job cost ledger is that it’s true: every number traces to a real purchase order, a real bill, a real timecard, a real material issue. When you mix estimates into actuals, you lose the one thing the ledger is for, the ability to reconcile to the penny and trust what you’re looking at.

Keep the actuals ledger source-document-only, and put projections where they belong: in a clearly-labeled forecast view alongside it, never blended in. This is exactly how SitewideOps’ job costing is built (a source-document cost ledger by cost code and cost type, with a separate, labeled cost-to-complete forecast) so the numbers your controller signs off on always tie out.

How job costing feeds the WIP schedule

Accurate job cost feeds more than the project manager; it’s the raw material for your work-in-progress schedule. WIP compares cost incurred to total estimated cost to figure percent complete, then compares earned revenue to what you’ve actually billed to reveal over- and under-billing. If your job cost is wrong, your WIP is wrong, and your controller and bonding company are reading a fiction.

That’s why the two are inseparable: clean, source-document job costing by cost code rolls straight into a defensible WIP. For the other half of the picture, see our guide on how to build a WIP schedule.

Labor productivity: the signal that arrives before the cost

Job cost tells you what you’ve spent. But it’s a trailing number on labor (your most volatile cost) because the dollars don’t land until payroll is run and the bill is posted. By then the week is over. There’s an earlier, sharper signal available: whether your crews are installing at the man-hour rate you budgeted. That productivity is visible the day the work is done, and it’s what a WIN log (a labor-productivity, earned-value log) is built to surface.

It works in units and hours, not just dollars. For each scope of work you set a budget in man-hours per unit: the hours it should take to install a square of membrane, a linear foot of flashing, a detail. Crews then log, day by day, how many units they put in place and how many hours it took. From those two numbers the log earns hours against the units actually installed and compares them to the hours actually burned, ahead of the labor budget or behind it, per crew, per activity, updated daily.

The output is a set of leading indicators no dollar report gives you this early:

• Win/loss hours: earned hours versus actual hours, so you can see a crew banking or bleeding time against the budget.
• An efficiency index: a single number that says whether a crew is beating (>1.0) or trailing (<1.0) the man-hour budget for the work they’ve completed.
• Percent complete by installed units: real physical progress, not a percentage someone estimated.
• A projected finish: where the crew lands if the current run rate holds.

A membrane crew trending 12% over its man-hour budget shows up this week (while you can still reslot the crew, fix the staging, or get ahead of a change order), not next month, when the labor cost finally posts to the ledger and the margin is already gone.

In SitewideOps this is the WIN Log, and it’s fed by the same daily field reports that hit job cost, so labor productivity and labor cost come out of one entry, not two. It sits in front of your job cost as the leading indicator: the WIN Log warns you a cost code is going to run hot; the job-cost ledger confirms it in dollars a few weeks later.

Common mistakes

The failures are predictable, which is good news. You can design them out:

• No cost codes (or too few). Costing a whole job to one bucket tells you nothing. So does inventing a new code per document. Pick a consistent, meaningful structure and hold to it.
• Ignoring commitments. Watching only actuals hides exposure you’ve already taken on. Track open POs and subcontracts as committed cost.
• Un-burdened labor. Costing hours at base wage understates your single largest cost. Always use the burdened rate.
• Stale budgets. Comparing against the original estimate after change orders have moved the scope. Cost against the revised budget.
• Coding to the wrong job. Material or hours landing on the wrong job number corrupts two jobs at once. Make the job and code unambiguous at entry.
• Mixing estimates into actuals. The moment the ledger contains guesses, it stops being reconcilable. Keep projections in a separate, labeled forecast.