How do you calculate first pass yield?

FPY = units that complete the step correctly on the first attempt ÷ units that entered the step. Anything reworked, repaired, retested, or scrapped counts against it, even if the unit is later recovered and shipped. Calculate it per process step, from records captured where the work happens.

What is the difference between first pass yield and final yield?

Final yield counts all good units out, including those fixed by rework, so it measures what shipped. First pass yield counts only units right the first time, so it measures how well the process works. The gap between them is the hidden factory, the rework, retest, and repair loop that consumes capacity while producing nothing new.

What is rolled throughput yield (RTY)?

RTY is the probability a unit passes every step of a multi-step process defect-free the first time, calculated by multiplying each step's FPY. In a hypothetical four-station example, 97%, 95%, 98%, 96%, RTY is 0.97 × 0.95 × 0.98 × 0.96 = 86.7%, even though every station individually clears 95%.

What is a good first pass yield percentage?

It depends on the process, the product, and how strictly "first pass" is defined, so benchmark a line against its own history rather than published figures. A strictly defined 88% that is trending up tells you more than a soft 99%. The more useful universal test: is your FPY meaningfully below your final yield? If not, rework probably isn't being logged.

Does rework count against first pass yield?

Yes, by definition. Any rework, repair, adjustment-after-failure, or retest makes the unit a first-pass failure even if it ultimately ships as good product. That is the entire point of the metric: it refuses to let recovery work launder process failures out of the number.

Is first pass yield the same as the quality factor in OEE?

They share the same logic, first-time good units over total units, and should be built from the same records. OEE's Quality factor is typically computed per machine or line for the OEE roll-up, while FPY is tracked per process step and compounds into rolled throughput yield. If the two disagree, one counting rule is wrong and worth auditing.

First Pass Yield: The Quality Metric That Doesn't Lie

Why FPY beats final yield and scrap rate, how rolled throughput yield compounds across stations, and how to measure both without softening the definitions.

First pass yield (FPY) is the percentage of units that complete a process step correctly the first time, with no rework, repair, or scrap. FPY = units good on the first attempt ÷ units entering the step. Unlike final yield, it counts rework as a failure, which is why it doesn't lie.

Plenty of plants report 99% yield while running a full-time rework crew. The trick is in the definition: if a unit fails, gets reworked, and eventually ships, most yield numbers count it as good. First pass yield refuses the laundering. It asks one blunt question, did the process do it right the first time?, and its answer is usually the most honest quality number in the building. This post covers FPY versus its softer cousins, the rolled throughput yield calculation with a worked example, and how to put FPY to work.

What is the difference between FPY, final yield, and scrap rate?

They differ in what they forgive. Final yield forgives rework; scrap rate forgives everything that isn't thrown away; FPY forgives nothing:

Final yield = good units out ÷ units in, after all rework and repair. A unit that failed twice and was fixed twice counts as good. Useful for material planning; useless for judging the process.
Scrap rate = units discarded ÷ units in. It sees only the irrecoverable losses. A process can hold scrap near zero while reworking a third of everything it touches.
First pass yield = units right the first time ÷ units in. Rework, repair, retest, and scrap all count against it. It measures the process, not the recovery operation.

The gap between final yield and FPY is the hidden factory the inspection, rework, retest loop that consumes labor, floor space, and schedule while producing nothing new. That gap is a major component of cost of quality which the American Society for Quality notes commonly runs 15–20% of sales revenue, up to 40% at poor performers. A plant reporting 99% final yield and 84% FPY is paying for the difference every day; the yield report just declines to mention it.

The rework loop, hypothetical numbers. Final yield counts the 9 reworked units as good; FPY does not. The 9-point gap is the hidden factory.

What is rolled throughput yield?

Rolled throughput yield (RTY) is the probability that a unit passes every step of a multi-step process right the first time: multiply the FPY of each step. RTY = FPY₁ × FPY₂ × … × FPYₙ. Like OEE's three factors the yields compound, and the compounding is brutal at scale.

A worked example with hypothetical numbers. A line has four stations, each with a respectable FPY:

Station	FPY	Running RTY
1 · Forming	97%	97.0%
2 · Assembly	95%	92.2%
3 · Finishing	98%	90.3%
4 · Pack-out inspection	96%	86.7%

Every station clears 95%, yet only 86.7% of units sail through untouched, about one unit in eight needs intervention somewhere. Stretch the same math to a 10-station process at 97% each and RTY drops to about 74%. This is why multi-step operations with "great" station-level yields still drown in rework, and why RTY, not any single station's FPY, is the number that describes what the customer's order actually experiences.

RTY compounds station FPYs: 0.97 × 0.95 × 0.98 × 0.96 = 86.7%. Hypothetical numbers, honest arithmetic.

How do you start measuring first pass yield?

Define "first pass" at each step, count entries and first-time passes, and resist every temptation to soften the definition. The sequence:

Map the steps that can fail. Every station with an inspection, test, or the possibility of rework gets its own FPY. Lumping steps together hides which one bleeds.
Define "good, first time" per step, in writing. Any rework, repair, retest, re-run, or adjustment-after-failure counts as a first-pass failure, even a ten-second touch-up. Fuzzy definitions are how FPY drifts back into final yield.
Count units in and first-pass units out. From test records, inspection logs, or operator tallies at the station. Rework must be logged where it happens, not reconstructed at month-end.
Attach defect codes to every failure. An FPY number without reasons is a scoreboard without a game plan; a short code list feeding defect tracking turns it into a Pareto.
Compute FPY per step and RTY for the line, weekly at minimum. Post both where the crew can see them.
Work the worst step's top defect code with root-cause tools, SPC on the variable that drives it, and poka-yoke where the failure mode is human error. Then re-measure.

Where should you measure FPY first?

Start where the failures are expensive, not where the counting is easy. Three placements pay fastest:

The step just before value concentrates. A defect caught before final assembly costs a component; the same defect caught after costs the assembly. Measuring FPY at the gate in front of your most expensive operation protects the most dollars per detection.
The step everyone already suspects. Every plant has a station with a permanent rework bench beside it. That bench is a confession; give it a number.
The final step, as the customer's proxy. Pack-out or final-test FPY approximates what escapes toward the customer, and pairs naturally with complaint and return data on the scorecard.

One caution: do not launch FPY as a people metric. The first honest measurement usually looks worse than anyone expected, and if the number is used to grade operators, the logging will quietly improve instead of the process. FPY measures the process, design, materials, settings, standards, and the fastest way to kill it is to make it a scoreboard for blame.

How does FPY connect to OEE and downtime?

FPY is the Quality factor of OEE wearing its own uniform. OEE's Quality term counts first-pass good units over total units for exactly the reason FPY does: rework is a loss even when the unit is saved. The two metrics should be built from the same records, if OEE Quality and station FPY disagree, one of the counting rules is wrong, and finding which one is a useful afternoon. (The OEE calculator makes the Quality factor's arithmetic explicit.)

The rework loop also feeds back into the time ledger. Reworked units consume capacity twice, rework stations starve or block their neighbors, and defect investigations stop lines, losses that land in the six big losses as quality losses and in the downtime log as quality stops. A falling FPY is frequently the earliest warning that downtime and schedule misses are coming, which is why it belongs on the daily board, not just the monthly quality review.

One practical note on data: FPY is only as honest as rework logging, and paper rework logs are where honesty goes to die, a touch-up at the station takes ten seconds, and writing it down takes thirty. Plants that digitize capture at the station, the way Harmony turns paper checks and quality logs into live, searchable records feeding root-cause analysis (see the platform), find their real FPY within weeks. It is usually lower than the reported one, and knowing that is the beginning of fixing it.