Why the job that looked profitable at 38% closed at 11%

Revenue is visible. Cost is not. That is the whole problem.

Every inspection firm knows what a job invoiced. Almost none know what it cost. The invoice is one document produced by one process. The cost is fifteen fragments produced by payroll, accounts payable, a fuel card, a purchase order to a subcontractor, and a Level III's Tuesday evening. Only some of those fragments carry a job number, so the gross margin the accounting system reports is revenue minus the costs that happened to be coded — a different number from the one the owner needs to make decisions with.

The gap is neither small nor random. It concentrates in two components, report preparation and travel-linked cost, and both of them scale with exactly the work that looks most attractive on a revenue report: complex multi-method scopes and out-of-town turnarounds. The result is a firm that pursues more of the work that is quietly losing money, because every report it runs says that work is its highest-revenue segment.

Fixing this does not require replacing the accounting system. It requires the job to become the unit of record, with every cost a job causes attaching to it at the moment it is incurred rather than being reconstructed at month end. The inspection ERP holds the operational half; the ledger keeps the financial half. Neither pretends to be the other, and the boundary between them is a design decision rather than an accident.

The loaded rate: what a field hour actually costs you

A technician's hourly wage is the smallest part of what an hour of their time costs. On top of it sit employer payroll taxes, workers' compensation — priced high for industrial inspection — health benefits, paid time off, PPE, vehicle, phone, and the non-billable hours the firm pays for: travel between sites, safety orientations, training and recertification. Costing a field hour at base wage understates it by a margin wide enough to turn a reported profit into a real loss on a thin job.

Build the loaded rate from your own numbers rather than a rule of thumb. Take twelve months of total employment cost for the technician group — wages, taxes, insurance, benefits, allowances, training and certification cost — and divide by the billable hours actually delivered in that period, not by 2,080. The denominator is where most firms flatter themselves, because dividing by a full year of hours silently assumes a utilisation rate that no inspection firm achieves.

Premium hours need their own treatment. Federal law pays overtime past forty hours in a workweek. California pays one and one-half times the regular rate for hours past eight in a workday and up to twelve, double time past twelve, time and a half for the first eight hours on the seventh consecutive day and double time beyond that. Turnaround schedules generate those hours by design, so a twelve-hour-day scope costed at flat rate is wrong before the crew mobilises.

Per diem is a daily cost, not a travel expense

Per diem is coded to travel, approved by whoever approves expenses, and never seen again. It is in fact one of the largest direct costs on any out-of-town job, incurred every single day the technician is away, whether or not that day was billable. GSA's FY2026 standard CONUS rate is $110 for lodging plus $68 for meals and incidental expenses, effective 1 October 2025 through 30 September 2026; non-standard localities are set higher, and Gulf Coast turnaround markets during outage season price accommodation well above the federal figure.

The distortion this creates is specific and directional. A five-day local job and a five-day out-of-town job at the same day rate look identical on a revenue report and differ by the entire travel and subsistence stack in reality. When per diem sits in an overhead pool, that difference is spread evenly across all jobs, which subsidises travel work with margin generated by local work and makes the local work look worse than it actually is.

Allocate per diem to the job, per day, at the point the travel is scheduled rather than when the expense claim lands. Two things follow immediately. Bidding on out-of-town work starts from the real cost floor rather than from a day rate. And the comparison between clients becomes honest, because the client who calls you only for distant emergency work stops being credited with margin that local clients generated.

Mobilisation and mileage on regional work

Mobilisation is frequently invoiced as a line item and almost never costed as one. The real cost is drive hours at the loaded rate, vehicle running cost, and — where the drive is long enough to require it — an overnight before the shift starts. Firms that bill a flat mobilisation charge across a wide service radius are cross-subsidising the far edge of that radius with the near edge, and the further they let the radius grow, the worse the trade becomes.

For vehicle cost, the IRS business standard mileage rate is the accessible published reference: 72.5 cents per mile for 1 January through 30 June 2026, rising to 76 cents per mile for 1 July through 31 December 2026. It is a tax and reimbursement rate rather than a costing rate, but it is published, defensible, and far closer to true cost than a fuel receipt, which excludes depreciation, tyres, insurance and maintenance entirely.

The scheduling side of the same decision — which qualified technician is closest, and what that choice costs — is covered in crew dispatch. Costing and dispatch read the same data from opposite ends: dispatch needs the cost consequence before the assignment is made, costing needs the assignment record afterwards. Holding both in one system removes the reconstruction step, and mileage stops being an argument at month end.

Consumables, equipment and the cost of staying qualified

Consumables are expensed at purchase and disappear into overhead: couplant, penetrant and developer, film and chemistry, batteries, wedges, transducers, magnetic particle media. Each is caused by a specific method on a specific job. Radiographic and penetrant scopes consume materially more than a UT thickness survey, so pooling consumables into overhead makes RT and PT look cheaper than they are and makes UT carry cost it never incurred.

The cost of remaining qualified belongs to the methods that require it, not to the company at large. Radiography carries a stack of recurring obligations under 10 CFR Part 34: survey instruments calibrated at intervals not to exceed six months under 34.25(b)(1), refresher safety training for each radiographer at intervals not to exceed twelve months under 34.43(d), and observation of each radiographer's job performance during an actual radiographic operation at intervals not to exceed six months under 34.43(e).

Those are real, recurring, method-attributable costs. Allocated to the RT line, they change the answer to "should we keep doing radiography in house or subcontract it". Left in overhead, that question cannot be asked at all, because the cost of holding the capability is invisible against the revenue it produces. The same logic applies to instrument depreciation and calibration on every other method you run.

Report preparation: the cost nobody meters

This is the largest silent cost in the industry. Field work ends and report work begins: data reduction, drafting, sketching, photographs, technique documentation, Level III technical review, quality review, client comments, revision, re-issue. On a complex multi-method scope, that work consumes a meaningful fraction of the field hours it documents. Almost none of it reaches a timesheet with a job number on it, because the people doing it are salaried and their time is treated as overhead.

The consequence is systematic rather than random. Report burden does not scale with revenue — it scales with scope complexity, component count, number of findings, and how much a particular client revises. Two jobs at identical revenue can differ several-fold in report hours. Costing that lands report time in overhead reports both jobs at the same margin and gives the sales team no signal that one client's work is structurally more expensive to deliver.

Metering it costs almost nothing. Report hours are captured the same way field hours are — against the same job number, by the same people, on the same timesheet. What changes is the analysis: revision cycles become a visible cost per client, and the argument for fixing report quality upstream turns into arithmetic instead of opinion. That argument is made in full on what makes an NDT report defensible.

Subcontracted methods and pass-through margin

Subcontracting is where negative margin hides best, because the transaction looks clean. A purchase order goes out, an invoice comes back, the client is billed at a markup, and the arithmetic appears to work. What that arithmetic omits is the coordination: scheduling the sub, verifying and filing their certification records, receiving and reviewing their report, integrating it into your deliverable, and answering for it when the client queries a finding.

That coordination is your Level III's time and your project manager's time, and it is rarely charged to the job. On a thin markup it exceeds the margin outright. Firms discover this as a pattern long before the accounting shows why — the same method, the same subcontractor, consistently the job nobody in the office wants to run.

Cost the sub's invoice to the job, and cost the internal hours the sub generates to the same job. The markup that clears both is the real minimum. Where a method is subcontracted repeatedly at a loss, the decision to build the capability in house finally has a number behind it — including the recurring qualification cost that the method will carry once you own it.

Reading profitability by client, by job type and by technician

With costs attached to jobs, three cuts become available that no revenue report can produce. By client: which accounts pay for the delivery they demand. By job type: which method mixes and scope sizes clear the loaded rate. By crew: which combinations complete scopes inside estimate and which consistently overrun, which is a training and estimating signal rather than a disciplinary one.

The client cut usually produces the most uncomfortable finding. Firms routinely discover that their largest account by revenue sits near the bottom by margin, because volume was won on rate concessions and the delivery burden — revision cycles, short-notice callouts, distant sites, extra reporting formats — was never priced. The response is not necessarily to resign the account. It is to reprice the specific behaviours that cost money, which requires knowing which ones they are.

The job-type cut drives quoting. When the historic cost of a five-day, two-technician multi-method turnaround is known with report hours and per diem included, the next quote for that shape of work starts from evidence instead of instinct. Firms quoting from a day rate and a guess are running an uncontrolled experiment on every bid, and only the accounting system ever learns the result.

Where the ledger stops and the job record starts

The accounting system's job is the general ledger, payroll, accounts payable and receivable, tax and the financial statements. It is good at those, and it is not built to be a job-costing engine for a field services firm, because it never sees the operational facts — who was dispatched, for how long, with what instrument, drawing how many per diem nights, generating how many hours of report preparation and review.

The workable architecture keeps both systems and gives each one the job it is good at. Operational cost is captured where it happens, in the job record: dispatch, timesheets, per diem, mileage, consumables, subcontractor purchase orders and report hours. Financial postings synchronise to the ledger. The mechanics of that synchronisation are set out on the QuickBooks integration page, and invoicing draws from the same job record, so what you bill and what you costed reconcile by construction rather than by effort.

Atlantis builds this on Odoo, configured for inspection work rather than generic professional services — affordable, accessible and fully customisable to your method mix and cost structure. Ask for a demo run against one of your own recent turnaround jobs. Re-costing a job you have already closed and invoiced is the fastest way to find out whether the leak described on this page is yours, or request a quote scoped to your crew size.

Request a consultation