The shipment is complete, but the repair area handled units that the final output count no longer reveals. First pass yield (FPY) measures the share of units that meet the declared process requirements on their first attempt without rework or other excluded recovery. FPY = first-pass good units / units entering the process × 100%. Keep the original result even when repair later makes a unit shippable.
ASQ’s quality glossary excludes units that are scrapped, rerun, retested, returned, or diverted to offline repair from first-pass success. Your measurement needs a matching entry population and an accepted first-pass disposition for those units. A final count of good output does not preserve that history on its own.
Define the process and what counts as a first pass
Choose the operation or end-to-end process you need to assess. A unit may pass final inspection on its first attempt while having required repair upstream. Station FPY and end-to-end first-pass success therefore need different boundaries. Put the process name, entry event, and acceptance criteria on the report so a local improvement does not get reported as an improvement across the whole route.
Name the first-attempt finish event and the events that prevent first-pass success. Inspection records should distinguish an accepted initial result from a later repair acceptance. If a test itself is invalid, follow the approved quality rule for invalid tests and retain the evidence. Deleting the failed record because a retest passed would leave no basis for explaining the reported yield.
Use a stable unit or lot reference to connect entry, inspection, and disposition. Serial numbers support unit tracking; lot quantities need enough detail to reconcile splits and partial outcomes. A lot-level accepted status can conceal failed units inside the lot if the quantity history is missing. Preserve quantities at the point the outcome is established.
Reconcile first-pass success with final good output
Assume a hypothetical closed cohort of 500 units entering one operation. All first-attempt outcomes are known: 440 pass without additional work, 40 need rework, and 20 are scrapped. Later, 35 of the reworked units are accepted and five are scrapped. The example has no pending inspection, transfers, or missing units, and uses units rather than defect counts.
| Outcome | Units | Treatment in this example |
|---|---|---|
| Good on first attempt | 440 | Included in first-pass numerator |
| Sent to rework | 40 | Excluded from first-pass numerator |
| Scrapped on first attempt | 20 | Excluded from first-pass numerator |
| Accepted after rework | 35 | Included in final good output only |
| Scrapped after rework | 5 | Included in final scrap only |
First pass yield is 440 / 500 × 100 = 88%. Final yield under this example’s definition is (440 + 35) / 500 × 100 = 95%. Final good output is 475 units and final scrap is 25, reconciling to the original 500. The 35 recovered units improve final yield while remaining outside the first-pass numerator.
Do not add the 40 rework entries to the original 500 entries when calculating this cohort’s FPY. Those entries represent additional processing attempts on units already counted. A workload report can count the extra activity, but its denominator answers a different question. Keep physical-unit counts and processing-event counts in separate fields.
Follow entry cohorts across reporting cutoffs
The formula is easiest to interpret when every unit in the entry cohort has a known first-pass outcome. A calendar report can instead contain entries from today, completions from yesterday, and repairs from earlier lots. Dividing those unmatched totals may create a yield that no physical group of units experienced.
Choose an entry cohort, such as units entering during an accepted shift window, and follow its outcomes until the declared cutoff. If some remain in process or on hold, show their quantity beside the provisional result. Do not count them as scrap or success without a defined disposition. A cohort with unresolved outcomes cannot support the same completed-process conclusion as the closed example.
Record when a result became known as well as when the unit entered. Late inspection can change the accepted view of an earlier cohort. Preserve the previous report version and the new evidence so a revised percentage is explainable. A reporting correction should not erase the source record that originally supported the published figure.
Explore the evidence behind a quality calculation
Start with an inspection export and a unit or lot reference you can access, then examine how accepted definitions connect the records.
Combine counts without averaging unlike percentages
Suppose two hypothetical closed runs use the same process and acceptance rules. One has 90 first-pass good units from 100 entries; the other has 475 from 500. Combined FPY is (90 + 475) / (100 + 500) × 100 = 94.17%, rounded for display. Averaging their 90% and 95% rates gives 92.5%, which assigns equal weight to different entry quantities.
Keep product mix visible when comparing periods. If one product needs a different test or has a different routing, a combined percentage may move because the mix changed. Compare like products and processes before attributing the change to an improvement action. Retain the aggregate for its declared purpose, with a path to the product and operation contributions.
Count defective units once in the FPY numerator decision, even when one unit has several defects. Defect counts can support investigation of failure modes but are not interchangeable with failed-unit counts. A unit with two defects remains one unit that failed its first pass. Keep defect records linked without multiplying the number of units affected.
Distinguish route success from equipment effectiveness
Minitab’s rolled throughput yield reference describes multiplying step yields across a process. For a route with several operations, preserve the connection between units and use compatible populations and pass definitions in that sequential model. Rework loops, branches, and changing cohorts need explicit treatment before a product of percentages becomes an end-to-end claim.
A hypothetical route with sequential stage FPYs of 90% and 95% produces an 85.5% product under that model: 0.90 × 0.95. Treat it as a modeled route measure, not a direct count of unique units with no intervention unless the records establish that population. When unit histories are available, reconcile the route result against those histories and explain any difference.
Vorne’s OEE calculation reference defines its quality factor with first-pass good output and combines quality with availability and performance. FPY alone does not show lost running time or speed. The overall equipment effectiveness formula covers those factors and their matching time and count inputs.
Use failed-unit evidence to choose the next review
Segment the failed units by the operation and accepted reason before assigning an action. A count of first-pass failures identifies the affected work; it does not prove whether tooling, incoming material, the method, or a recording problem caused it. Keep unknown reasons visible so a complete-looking chart does not imply that every cause has been established.
Link material and production identifiers when the review depends on which inputs reached the failed units. The lot traceability workflow addresses those links. A matching lot identifier establishes a relationship to investigate; it does not establish that the material caused the failure.
Keep the quality result with the relevant acceptance rule and responsible owner. If the rule changes, mark the effective period and assess whether the earlier baseline remains comparable. The quality management software comparison addresses the broader record and workflow needs around controlled quality processes.
Permute: governed first-pass calculations with unit evidence
Permute sells the context layer between business data and the AI tools people already use. We connect approved production and inspection records from sources such as NetSuite and Excel, reconcile unit or lot references, and preserve the evidence behind the calculated view. Your quality owner supplies the entry, first-pass, and disposition rules.
We make those accepted definitions reusable in reports and applications, with current permissions controlling access. A quality manager can ask which recorded outcomes changed a cohort’s FPY and inspect the contributing rows. Claude, ChatGPT, or Copilot can use that prepared context while the original acceptance and repair histories remain available.
Keep acceptance and release authority with quality owners
We do not determine whether a product meets its specification, approve repair instructions, or release held material. This reporting workflow reads connected records and executes declared calculations (including the approved treatment of invalid tests). Quality decisions and production release remain in the owning process; missing inspection outcomes stay unresolved.
First pass yield measures success before rework by keeping first-attempt good units aligned with the units that entered. The hypothetical cohort has 88% FPY and 95% final yield because repair recovers some output without changing its initial result. Cohorts, unit identities, and accepted dispositions preserve that distinction across reporting periods. Use the failed-unit evidence to investigate the work rather than treating the percentage as a diagnosis.
Review a first-pass report’s source evidence
Bring entry counts, inspection outcomes, and repair dispositions for one process. We can review the definitions and connections needed for a repeatable quality view.
Questions about first pass yield
What should the report show when no units entered?
Show no applicable yield for that cohort, with an entry count of zero. A zero denominator cannot produce a valid percentage. Distinguish that state from units entering without known inspection outcomes, which leaves a provisional cohort.
Can a passing retest turn a unit into first-pass good output?
Preserve the first attempt and apply the approved policy for the retest. ASQ’s definition excludes retested units from first-pass success; an invalid initial test needs its own documented treatment. Final acceptance and first-pass classification should remain distinguishable.
What is a good target for FPY?
Set the target for the declared process, product mix, and acceptance rules using the quality owner’s requirements and a comparable baseline. A cross-process percentage without those conditions cannot establish an appropriate target. Review both the rate and the unresolved or excluded populations before making a performance judgment.