The Dashboard Said 78% OEE. The P&L Said Something Else.
A shop owner installs a monitoring box on the CNC mill, watches the OEE number climb into the high 70s over a few weeks, and feels like the floor is finally running well. Then the quarter closes, and the accountant says margin is down two points from last year. Same equipment, same operator, same "good" OEE number — and somehow the shop made less money.
This isn't a contradiction. It's a sign that OEE and job costing are answering two different questions, and the shop has only been asking one of them.
OEE (Overall Equipment Effectiveness) asks: how well did this machine run during the time it was scheduled to run? Job costing asks: did this specific job, at the price we quoted, cover its labor, its burden, and leave something behind? A machine can post excellent OEE running a job that loses money, and a machine with mediocre OEE can still be attached to a job that's comfortably profitable. Confusing the two — or worse, only tracking one — is why so many small shops feel busy and broke at the same time.
This article works through what each metric actually measures, where they diverge, and why a job shop under 50 people almost always needs to get job costing right before OEE becomes worth the overhead.
What OEE Actually Measures
OEE is a manufacturing-floor metric built from three multiplied components: availability (was the machine running when it was supposed to be), performance (was it running at its rated speed), and quality (did the output pass). Multiply the three together and you get a single percentage that describes equipment effectiveness over a measured window of time.
OEE is machine-centric by design. It doesn't know or care which customer the part belongs to, what the part was quoted at, or what burden rate applies to that work center. It's a production-engineering metric, born on high-volume, repetitive lines where the same part runs through the same machine for hours or days at a stretch, and small deviations in speed or uptime compound into large output losses.
That's exactly the environment OEE was built for — and exactly the environment most small job shops don't run in.
What Job Costing Actually Measures
Job costing is customer- and job-centric. It asks a different question at a different level: for this job, at this quoted price, what did it actually cost to run — labor hours logged against each operation, machine/burden rate for the work centers touched, any scrap or rework time charged back to the operation that caused it — and how does that actual cost compare to what was quoted?
Where OEE aggregates across time, job costing aggregates across a job. It doesn't care whether the machine ran at 100% of rated speed for eight hours; it cares whether the three hours actually logged against Job 4471's second operation matched the 2.5 hours that was quoted, and if not, why.
For a shop that quotes a different part almost every week — short runs, one-offs, prototype-to-production handoffs — job costing is the metric that maps directly onto how the business makes or loses money. A job profitability analysis built job-by-job tells an owner exactly where the quote was wrong, where the floor ran slow, and where scrap ate the margin — three very different problems that a single OEE percentage collapses into one number.
Why the Two Numbers Can Disagree
Here's a worked example, illustrative for a representative shop: a work center posts 85% OEE for the week — strong by most standards. But the jobs that ran through it that week were quoted using a standard time that was set two years ago, before a fixture change added ten minutes of setup per job. Every job comes in "on time" relative to the machine's own rated capability, and "over" relative to what was quoted. OEE looks great. Job costing shows a margin problem on every job that touched that work center.
Run it the other way: a work center posts a mediocre 60% OEE — lots of short setups, small lot sizes, a changeover every ninety minutes — but the jobs running through it are quoted at a rate that already assumes short runs and frequent changeovers. Actual-vs-quoted comes in right on target, job after job. The equipment "looks bad" on paper and is, in fact, exactly as profitable as planned.
Neither number is lying. They're measuring different things, and a shop that only watches one will misdiagnose the other's problem. A high OEE with eroding margin usually means the quote is stale, not that the floor is slow. A low OEE with healthy margin usually means the work is inherently short-run and setup-heavy, not that the operator is underperforming.
Which Metric a Small Shop Needs First
For a 5–50 person job shop running mixed part numbers with paper travelers, a spreadsheet, and QuickBooks, job costing comes first — not because OEE isn't a real metric, but because job costing is the one that answers the question the owner is actually asking at quarter-end: which jobs made money, and why didn't the others?
OEE requires machine-level monitoring infrastructure and enough repetition in the part mix to make a rolling average meaningful. Job costing requires something simpler and more foundational: a routing per part number, a way to log actual time against each operation, and a comparison of that actual time to what was quoted. That's the data a shop needs before it can make sense of anything higher up the stack — including OEE, if and when the shop's volume and part-mix consistency ever justify it.
Put another way: OEE is a refinement question for a mature, high-repetition floor. Job costing is a survival question for a shop that quotes from feel and finds out at year-end whether it guessed right. Most job shops in the 5–50 employee range are answering the survival question, whether they realize it or not.
Building the Data Foundation Either Metric Needs
Both metrics ultimately draw from the same raw material: accurate, operation-level time data tied to a routing. WorkTickets is built around that foundation specifically for job costing — a routing/operation-sequence builder per part number, a traveler that carries that routing onto the floor (printed with a barcode/QR or viewed digitally on a tablet), one-tap clock-in/out with downtime reason codes for setup, run, waiting, and rework, and an actual-vs-quoted comparison per operation and per job the moment time is logged. Scrap and rework get tied to the operation that caused them, not buried in a general variance line.
That's a deliberately narrower scope than a full MES platform that also chases OEE, capacity scheduling, and equipment monitoring — WorkTickets is a standalone, SaaS-native execution-and-costing layer that sits below full ERP, not a replacement for shop-floor monitoring hardware. But it's exactly the layer that a shop needs in place before an OEE conversation is worth having at all: you can't calculate meaningful equipment effectiveness on top of data nobody trusts, and you can't fix quoting drift with a machine-monitoring dashboard.
If OEE is eventually the right next question for a shop's floor, it will be a much shorter conversation once actual-vs-quoted, downtime reason codes, and scrap-by-operation are already being captured daily instead of reconstructed at quarter-end.
Where to Go From Here
For a broader look at which numbers actually deserve a spot on a shop's dashboard, see machine shop KPIs worth tracking, work center utilization tracking, and machine downtime analysis — each digs into one piece of the picture this article compares at a high level. The job profitability analysis piece and the execution-and-costing guide for small job shops go deeper on the job-costing side specifically.
If downtime reason codes are the piece missing from your current setup, the Downtime & Reason-Code Analysis Workbook walks through setting up a clean reason-code taxonomy — setup, run, waiting, rework — so whichever metric you track next, the underlying data holds up.
Want more like this — practical breakdowns of the metrics that actually explain shop-floor margin, without the vendor pitch? Subscribe to the WorkTickets newsletter and we'll send the next one directly.
