First pass yield: how to measure and reduce rework
Calculate first pass yield with a worked example, separate scrap from rework, and use defect data to choose and verify your next process improvement.
First pass yield (FPY) is the percentage of units that pass a defined process on the first attempt, without repair, rework or repeat testing. It helps you see losses that a scrap count misses: a part can eventually ship and still have consumed an extra operation, inspection and queue slot.
To reduce manufacturing scrap and rework, start with a stable definition of the process being measured. Then record failures at source, choose a specific cause to investigate and check whether the change improves the same measure over comparable work.
First pass yield formula and worked example
First pass yield = units passing first time ÷ units entering the process × 100.
This follows ASQ’s definition of first pass yield, which excludes units that require additional work or testing to meet the quality criteria. Define the operation or route boundary and observation period before counting.
In a hypothetical completed batch of 200 units, 170 pass first time, 20 pass after rework and 10 are scrapped. The three groups do not overlap.
| Measure | Calculation | Result |
|---|---|---|
| First pass yield | 170 ÷ 200 × 100 | 85% |
| Units needing rework | 20 ÷ 200 × 100 | 10% |
| Scrap rate | 10 ÷ 200 × 100 | 5% |
| Final accepted yield | 190 ÷ 200 × 100 | 95% |
The 95% final accepted yield does not mean the process achieved 95% FPY. Twenty units needed another attempt. Counting them as first-time passes would hide the work you are trying to remove.
If some units are still waiting for a result, show them as pending rather than silently treating them as good or bad. If one part has three defect observations, distinguish the number of defective units from the number of defects. Otherwise the same batch can appear to contain more failures than parts.
Capture enough context to investigate
A useful defect record identifies the job or batch, operation, quantity affected and observed failure. Record the time, relevant machine or fixture and material identity where they help distinguish causes. Attach a measurement or photograph when it preserves evidence that will otherwise disappear.
Use a short, maintained list of defect categories for analysis and allow a note for detail. “Other” needs periodic review: if half the records end up there, the categories no longer describe the process. Avoid asking the operator to guess a root cause. “Bore oversized” is an observation; “tool wear” is a hypothesis to investigate.
Bulk’s Quality supports linked quality records, including non-conformance reports and 8D reports. That provides a place to carry the issue and its follow-up. The usefulness of any yield calculation still depends on how production counts, rework and dispositions are defined and recorded.
Use a Pareto to choose a problem
A Pareto chart ranks categories by a chosen measure. Use defect count when you need to understand frequency, and a defined loss-cost measure when you need to prioritise economic impact. Keep those views separate if they answer different questions.
Do not assume your data must follow an exact 80/20 split. Nor should the largest bar automatically outrank a less frequent defect with a serious customer or safety consequence. Review severity alongside frequency and cost.
Before choosing a fix, split the leading category by relevant conditions. Does the problem cluster around a fixture, part family, supplier lot or changeover? Compare rates as well as counts: the busiest machine may have the most defects while also processing far more units.
Separate containment from corrective action
Containment protects the next operation or customer while the cause is being investigated. Sorting a batch may be necessary, but it does not establish why the defect occurred or prevent another batch from failing.
For recurring or significant problems, ASQ’s 8D method gives a structured progression through problem definition, containment, cause analysis, corrective action and prevention. Use evidence to test the suspected cause before changing the process. A completed form is not evidence that the cause was correct.
In a hypothetical machining investigation, inspection finds a dimensional issue concentrated after a fixture change. The team checks the setup and measurement method, tests a proposed correction on controlled work, and compares results under similar conditions. It records what changed and which work the change applies to.
Verify the improvement without moving the goalposts
Keep the same FPY boundary, acceptance criteria and counting rules before and after the change. Compare similar product mix and record the number of units observed. A higher percentage over a small or easier batch is a reason to keep watching, not a complete proof of improvement.
Look at rework hours and scrap cost alongside FPY. If more inspection is finding defects earlier, the recorded failure rate may initially rise even though the process is becoming better understood. Explain that change rather than suppressing the records.
Bring the leading issue, its owner and the next evidence check to the daily SQDCP meeting. Start with one recurring loss that the team can investigate, then use the next comparable run to decide whether the action worked.
