Free field guide · Updated May 2026

Four numbers most plants run on.

A short field guide for plant managers and ops directors. The math behind OEE, first-pass yield, on-time-in-full, and cost per good unit — the common ways each gets fudged in reporting, and a worksheet that auto-computes the honest version against what your plant reports today.

Read time
≈ 25 min
Worksheets
4 interactive
Sign-up
None

Free. No email gate, no nurture sequence, no follow-up. If after reading you'd like to talk it through with a manufacturing engineer, we can schedule a 30-minute call — but only if you ask.

Before you start
Intro

You probably already have these numbers. They may be quietly wrong.

Every manufacturer we've worked with reports a version of these four metrics. Many of them, on close inspection, have been rounding the corners on at least one — usually by inheritance: the way a report was set up six years ago, by someone who no longer works there, using assumptions nobody documented.

This guide doesn't introduce new metrics. It re-introduces four numbers you already use, with the specific errors and exclusions that often creep into them. Each section gives you the formula, the common ways the number gets fudged, an example of what that looks like in practice, and an interactive worksheet you can run with your own numbers — right here on this page.

If your reported numbers come out close to the strict numbers, your reporting is honest and you can stop reading. If there's a gap — and there usually is — the gap is the most useful number in this guide.

How to use the worksheets

Pick one line, one product, or one customer — last shift, last week, or last month. Type your numbers into the white cells. The shaded cells auto-compute the strict version. Compare to what your reports show today.

Nothing is sent anywhere. The worksheets run entirely in your browser; close the tab and the numbers are gone.

01 The hero

Overall Equipment Effectiveness is the most-cited and most-fudged number in shop-floor reporting. The math is simple. The inputs are where the trouble lives.

OEE — and the three traps inside it.

Formula
OEE = Availability × Performance × Quality

The math is honest. The inputs aren't.

OEE rolls three independent measurements into one number. Each input has its own definition, and each definition tends to get quietly stretched somewhere in your plant's history of reporting. Three places to look:

Trap 01
Availability

Excluding planned downtime from "available time."

The strict definition uses scheduled run time as the denominator. A common shortcut also strips out planned changeovers, breaks, and meeting time — making the equipment look more available than it is. Typical inflation: 8–15 points of OEE.

Trap 02
Performance

Using average cycle time instead of the nameplate ideal.

Performance is supposed to compare actual rate to the equipment's ideal cycle. A common shortcut compares to the average rate the line happens to run — building historical underperformance directly into the target.

Trap 03
Quality

Counting reworked units as "good."

Strict quality counts only first-pass good units. The shortcut counts anything that eventually ships — folding rework cost invisibly into OEE and disconnecting the metric from process control.

What this looks like in practice

Plant reports 82% OEE. On audit: availability denominator excludes an hour of daily changeover (trap 01), performance is benchmarked to a 60-second cycle when the press's nameplate ideal is 48 seconds (trap 02), and rework loops are counted as good (trap 03).

Strict OEE: 64%. The 18-point gap isn't a measurement error — it's a strategic decision somebody made at some point, and almost certainly never re-examined.

Worksheet 01

Recompute your OEE the strict way.

Pick one line, one shift. Fill in the strict-definition inputs in the white cells — the shaded cells compute themselves. Compare to what your reports show today.

A · Availability
Scheduled run time, and what was actually running.
Scheduled run time (minutes)
All downtime, including planned (minutes)
Actual run time (auto)
Availability = actual ÷ scheduled
B · Performance
Compare to the nameplate ideal, not the historical average.
Total units produced this shift
Nameplate ideal cycle time (seconds)
Theoretical max units (auto)
Performance = actual ÷ theoretical max
C · Quality
Only first-pass good units count. Rework, scrap, re-inspect are out.
Units made first-pass-right
Total units made (including rework)
Quality = first-pass ÷ total made
D · The honest number
Multiply the three. Then compare to what gets reported.
OEE strict = A × B × C
OEE as currently reported (%)
The gap (reported − strict)
Fill the cells above. If the gap turns out to be more than 5 points, one of the three traps is hiding in your reporting.
02 The truth-teller

First-pass yield captures the units that almost shipped before someone caught them. It's the single best leading indicator of whether your process is actually under control.

First-pass yield — behind the scrap line.

Formula
FPY = Units made right first time ÷ Units started

Scrap is downstream. FPY is upstream.

By the time a unit hits the scrap line, the cost is mostly paid: labor, material, energy, and the slot on the line. First-pass yield tells you what fraction of units cleared the first quality gate without intervention. It catches process drift before it becomes scrap — and it is also the metric most often quietly inflated.

Fudge 01

Counting reworked units as first-pass.

If a unit fails inspection, gets reworked, and then ships, strict FPY says it's not first-pass. The shortcut counts anything that ultimately leaves the plant — burying rework cost.

Fudge 02

Excluding "minor" rework from the count.

"It was just a quick polish." Plants that don't define rework explicitly tend to let small fixes go uncounted — turning FPY into a measure of major rework only, which is much less useful.

Fudge 03

Measuring only at the last station.

If FPY is computed at final inspection, every catch upstream is invisible. A line with five quality gates and four interventions per part can show 100% FPY at the last gate.

Fudge 04

Excluding units that never reach the line.

Material that fails incoming inspection. Setup parts scrapped during changeover. Calibration runs. These are sometimes excluded from "units started" — understating real cost of input variation.

What this looks like in practice

Plant reports 98.4% FPY at final inspection. On audit: rework loops at stations 3 and 5 aren't counted (fudge 01), polish-line repairs are excluded as "minor" (fudge 02), and FPY is measured only at the final gate (fudge 03).

Strict FPY computed at the first quality gate, counting every intervention: 86.1%. The 12-point gap is the size of the process-control problem you don't currently see.

Worksheet 02

Find your strict FPY.

Pick a single product or line. Choose your first quality gate — not the last. Count every intervention, including minor ones.

A · The denominator
Units that started down the line. Include setup and calibration.
Units started (include setup, calibration)
Units scrapped before the first gate
B · Cleared first-pass
Made it through the first quality gate with zero intervention.
Units that passed first gate untouched
Units that required rework (any size)
C · The honest number
First-pass clean ÷ units started. Then compare to what reports say.
FPY strict = cleared first-pass ÷ units started
FPY as currently reported (%)
The gap (reported − strict)
Fill the cells above. A gap above 3 points usually means rework loops aren't being captured anywhere — that's the next thing to map.
03 The customer's view

On-time-in-full is what your customer is actually scoring you on. The difference between "on-time" and "on-time-in-full" is where partial shipments quietly erode customer relationships you thought were healthy.

On-time-in-full — the customer's number.

Formula
OTIF = Orders shipped complete AND on date ÷ Total orders

Two words matter: "and" and "delivered."

OTIF is the customer's metric. They score you on whether each order arrived complete and on the date promised. Many plants report a version that has been quietly relaxed in one of four ways — and the gap between your number and theirs explains a lot of difficult calls from key accounts.

Fudge 01

Splitting orders into multiple shipments.

An order for 1,000 units split into three shipments — 600 on time, 300 a day late, 100 a week late — can be reported as three separate "fulfillments," two of which are on-time. From the customer's side, the order was late and incomplete.

Fudge 02

Using ship date instead of delivery date.

OTIF is what your customer experiences. If your contract is on delivery date but your metric is on ship date, you can ship "on time" and still be late on what matters. Carrier delays, customs, and weekend pickups are silent OTIF killers.

Fudge 03

Counting partial shipments as full.

A 1,000-unit order ships 940 on time and 60 the following week. Strict OTIF counts that order as not on-time-in-full. The shortcut counts it as a 94% fill rate — which technically isn't wrong, but isn't OTIF either.

Fudge 04

Adjusting promised dates after the fact.

When the original promise was Tuesday and the order shipped Friday, some systems update the "promised date" to Friday so the order shows on-time. This destroys both the metric and the institutional memory of what was actually promised.

What this looks like in practice

Plant reports 96% OTIF. Customer's vendor scorecard for the same period: 78%. The gap is split shipments (fudge 01), promised-date updates (fudge 04), and ship-date measurement on a contract that's on delivery date (fudge 02).

The 18-point gap is the size of the trust gap. Customers usually don't share scorecards unless something has already gone wrong — by the time you see the score, you may already be losing share.

Worksheet 03

Recompute OTIF from your customer's side.

Pick one major customer, last quarter. Strict OTIF treats each customer-order as one unit: in-full and on the originally promised delivery date.

A · The denominator
Customer orders. Not shipments — orders.
Total customer orders last quarter
Of those, orders split into multiple shipments
B · On-time
Delivered to the customer by the original promised date.
Orders delivered on or before original promised date
Orders where promised date was updated mid-stream
C · In-full AND on-time
100% of units, in one delivery, on the original date.
Orders both 100% complete in one delivery AND on time
Orders with backorder, split, or partial shipments
D · The honest number
On-time AND in-full ÷ total orders.
OTIF strict = on-time AND in-full ÷ total orders
OTIF as currently reported (%)
The gap (reported − strict)
Fill the cells above. If you can get a copy of your customer's vendor scorecard, compare strict-OTIF, your-OTIF, and their-OTIF. The three-way gap shows where the broken assumptions live.
04 The cross-check

Cost per good unit sorts the efficiency improvements that paid off from the ones that quietly didn't. When OEE goes up but CPGU also goes up, something's wrong with the OEE number — not with accounting.

Cost per good unit — the cross-check.

Formula
CPGU = Total cost of production ÷ First-pass good units

The denominator is the punchline. The numerator is the trick.

Most plants track cost per unit. The strict version is cost per good unit — first-pass good only — and it gets compromised in two places: by counting too many units in the denominator, and by leaving too many costs out of the numerator. CPGU is also the only one of the four metrics that has to talk to finance, which is where the most common errors enter.

Fudge 01

Cost per unit shipped, not cost per unit good.

If reworked units count in the denominator, the cost of rework is invisible in the per-unit number. Cost per shipped unit can look flat while cost per good unit is climbing — and margin is leaking.

Fudge 02

Excluding rework labor from "cost of production."

Rework hours sometimes get coded to a separate cost center — quality, maintenance, "indirect" — so they don't show up in standard cost. The hours were real and the wages were paid. They belong in the numerator.

Fudge 03

Excluding scrap material from "cost of production."

Scrap value is sometimes netted against scrap revenue and reported as a small line item. The full material cost of scrapped units belongs in production cost — what you paid for material that didn't become a sellable unit.

Fudge 04

Using standard cost instead of actual.

Standard cost is what the system thinks each unit costs based on routings and rates set last year. Actual includes variance, overtime, expedited material, and the changeover that ran three hours long. Standard is the cost you wish you had.

What this looks like in practice

Plant reports $4.18 cost per unit, flat year over year. On audit: rework labor coded to maintenance (fudge 02), scrap material netted against scrap proceeds (fudge 03), denominator is units shipped not first-pass good (fudge 01), and the whole calculation runs at standard cost (fudge 04).

Strict CPGU last year: $4.18. Strict CPGU this year: $4.61. Margin lost: ~10%.

Worksheet 04

Strict cost per good unit.

Pick one product family, last month. Use actual costs, not standard. Include rework labor and full scrap material.

A · Production cost (numerator)
Everything you spent to make units of this product last month.
Direct labor (actual hours × rates, $)
Rework labor — wherever coded ($)
Direct material consumed, incl. scrap ($)
Overhead allocated to this product ($)
Total cost of production
B · Good units (denominator)
First-pass good only. No reworked units.
First-pass good units last month
CPGU strict = total cost ÷ first-pass good
Cost per unit as currently reported ($)
The gap (strict − reported)
Fill the cells above. Repeat monthly for one product. The trend of strict CPGU is the best test of whether continuous-improvement work is actually generating savings — or just moving them between cost centers.
Putting it together
Cross-check

The four numbers, read as a system.

Each metric, in isolation, can be moved by gaming a single input. Read together, they catch each other. The point of running all four is the cross-check — the place where two numbers contradict each other is the most useful information in your reporting.

If OEE goes up but CPGU also goes up…

  • The OEE gain is paper. Look for rework loops counted as good (Trap 03).
  • Or the availability denominator changed quietly (Trap 01).
  • Or someone re-baselined performance to a worse ideal (Trap 02).

If FPY is high but CPGU is climbing…

  • You're catching defects upstream of where FPY is measured. Move the measurement gate to the first quality check.
  • Or scrap material is being netted against scrap revenue, hiding cost.
  • Or actual-vs-standard variance has widened without anyone reporting it.

If your OTIF is fine but your customer's isn't…

  • You're measuring ship date; they're measuring delivery date.
  • Split shipments count as multiple fulfillments on your side, one missed order on theirs.
  • Promised dates were updated mid-order without flagging the change.

If three improved but CPGU is flat…

  • Either operational gains are real and going to overhead (look at allocation).
  • Or operational gains were measured wrong — and CPGU is the only one telling the truth.
  • Run the strict versions on the worst-performing month. The gaps will rank-order what to investigate.
A short note
How to use this

You don't need new software to start this.

The four worksheets above can be done in a single afternoon with whatever reports you already pull. The point isn't to install a new dashboard — it's to do the strict calculation by hand, find the gap, and figure out where the gap came from.

What software is for — and what FactoryView is for, specifically — is making the strict calculation automatic, in real time, with the actual data flowing in from your equipment and your ERP. The afternoon-with-a-worksheet exercise is the diagnostic. Software is the cure, but only if you've already diagnosed which trap was hiding in your reporting.

Do the worksheets first. Find the gaps. Then we can talk about whether automating these numbers is worth doing in your plant — and where the highest-leverage place to start would be.

One way to use this with your team

Run all four worksheets on one product or line, last month. Compare strict to reported. Take the biggest gap to your next staff meeting and ask one question: how did this gap get into our reporting?

The answer is almost never a person doing something wrong. It's almost always a system that was set up correctly four years ago for a business that has since changed. Finding that system, and updating it, is the work.

If you found this useful

The guide is the deal. The call is optional.

If after running the worksheets you'd like to talk through the gaps you found with a manufacturing engineer, we're happy to schedule a 30-minute call. We won't follow up otherwise — no nurture sequence, no sales rep "checking in."

FactoryView is shop-floor software for manufacturers. We sit between off-the-shelf MES and pure custom development. Pre-built where it should be — dashboards, scoreboards, downtime, PLC connectivity. Customized where it has to be.