Wait, what? A factory dashboard says First-Pass Yield = 95%. At the end of the day, every product that leaves the factory has passed final inspection. One student says, “That is impossible. If only 95% were good, how can 100% of the shipped products be acceptable?”
The surprise disappears when we notice that first-pass yield is about a route through a process, not simply the final condition of everything that eventually leaves it. Some units may fail the first pass, be repaired or reworked, and later pass. A report can therefore show 95% first-pass yield while the final accepted output has a different percentage.
Quick answer
No. A first-pass yield of 95% does not automatically mean that 95% of all finished or shipped products are good. It means that, under the stated process definition, 95% of the units counted in the denominator met the stated requirements without needing rework, rerun, retest, repair or another defined detour before passing. To interpret the number correctly, you must identify the process boundary, denominator, pass rule and treatment of rework.
Owned learner job — and the boundary around it
This Reality Lab owns one real-world evidence-transfer job: evaluating a manufacturing or quality dashboard that reports first-pass yield, especially when the percentage is presented as if it were final product quality, total yield or a guarantee about every unit.
It does not own industrial engineering, lean manufacturing, process design, inspection theory or statistical process control. It also does not replace the PSLE Science owners for variables, fair comparison, percentages, measurement or conclusion writing. Here we apply those skills to a report that compresses a whole process into one percentage.
The factory story hidden inside one number
Imagine an original fictional workshop making 100 water-bottle caps. At the first inspection, 95 meet the stated requirement. Five do not. The five are sent to a repair station. Four can be corrected and later pass; one is scrapped.
| Stage | Units | What happened? |
|---|---|---|
| Entered process | 100 | All units begin the defined route. |
| Passed first time | 95 | No rework or repeat needed. |
| Failed first pass | 5 | Sent for correction or scrap decision. |
| Recovered by rework | 4 | Later meet requirement. |
| Scrapped | 1 | Does not become accepted output. |
| Final accepted output | 99 | 95 first-pass units + 4 reworked units. |
For this simplified case, first-pass yield is 95/100 = 95%. Final accepted yield is 99/100 = 99%. If the one scrapped unit is replaced by another new unit before shipping, the shipment might contain 100 acceptable products. Those are three different statements about the same production day.
Three questions before touching the percentage
When you see “FPY = 95%,” do not begin with multiplication. Begin with the process.
- What entered the denominator? All units entering one step? A batch? Only inspected units?
- What counts as passing? One dimensional check? A complete set of quality requirements? A test threshold?
- What does “first pass” exclude? Rework, repair, rerun, retest, adjustment, or some combination defined by the organisation?
Only after those are clear does the percentage become scientifically interpretable.
Observed, claimed, inferred
| Layer | Example |
|---|---|
| Observed / recorded | 1,000 units entered a defined process; 950 met the quality rule without rework. |
| Calculated / reported | First-pass yield = 950/1,000 = 95%. |
| Reasonable claim | 95% of those units met the stated requirement on the first pass under that process definition. |
| Unjustified leap | Exactly 5% of shipped products are defective; 95% of customers will receive good products; every production stage has 95% yield; or the process is 95% efficient. |
A number needs a process boundary
A production route can contain several steps: moulding, trimming, measurement, assembly, leak testing and packing. “95% FPY” might refer to only the leak-test station. It might instead refer to an entire route. Those are not interchangeable.
Suppose 100 units move through three stages. Stage A has 98% first-pass yield, Stage B 97%, and Stage C 99%. If the events behave according to a simple independent teaching model, the expected proportion that gets through all three stages without rework is approximately 0.98 × 0.97 × 0.99 = 0.941, or 94.1%. The point is not the advanced arithmetic. The point is that a local percentage cannot automatically be promoted into a whole-route percentage.
Representation trap: the giant green 95%
Picture a dashboard with a large green box:
QUALITY: 95%
Below it, in tiny text, the metric is “first-pass yield at Station 4.” The large label encourages a broad interpretation; the small label supplies the actual scope. Scientific reading reverses that visual priority. The small definition may matter more than the large number.
Ask what the dashboard chose to foreground, what it compressed, and which process details are needed to interpret the claim.
Comparison trap: Factory A versus Factory B
A product brochure says Factory A has 97% FPY while Factory B has 94%, therefore A makes better products. That may be true, but the two figures alone are insufficient.
- Do both factories define a unit the same way?
- Do they use the same pass criteria?
- Are they measuring one step or the complete route?
- Is rework classified consistently?
- Are the products equally difficult to make?
- Are the time periods comparable?
- Did either factory change inspection rules during the period?
A fair scientific comparison needs matched meanings, not merely matched percent signs.
Why rework matters even when the final product is acceptable
A unit that fails first pass but is successfully repaired may eventually satisfy the specification. For the final product, that recovery matters. For the process, the first-pass failure also matters because extra time, material, labour, testing and equipment may have been consumed.
That is why FPY can reveal a hidden process burden that final inspection alone does not show. A factory could ship acceptable products while spending substantial effort correcting them before shipment. The scientific communication job is to preserve both truths instead of replacing one with the other.
What evidence strengthens an FPY claim?
- A clear process boundary.
- A clear denominator: units entering the process in a stated time period.
- A clear pass criterion.
- A clear rule for rework, retest and rerun.
- Consistent definitions across periods or sites being compared.
- Enough observations to avoid overreacting to one tiny batch.
- Separate reporting of scrap, rework and final accepted output where those matter.
- An explanation of any method or inspection change that could alter the apparent trend.
What weakens it?
- “95% quality” with no metric definition.
- Comparing FPY from one stage with final yield from another factory.
- Changing the inspection threshold and then claiming the process improved.
- Removing difficult units from the denominator without disclosure.
- Calling reworked units “first-pass” units.
- Using one unusually good shift to represent the whole year.
- Treating the percentage as a guarantee for the next individual unit.
Worked case 1: same final output, different process evidence
Line P starts 200 units. 190 pass first time; 10 are reworked and all later pass. Line Q starts 200 units. 198 pass first time; 2 are reworked and later pass. Both eventually produce 200 accepted units.
Final accepted yield alone cannot distinguish the hidden process burden: both appear to reach 100%. FPY does: P has 95% first-pass yield; Q has 99%. That does not prove Q is superior in every possible way, but it is relevant evidence about first-pass process performance under the stated definitions.
Worked case 2: better FPY, worse final accepted count
Line R starts 100 units and 98 pass first time. Two are scrapped. Line S starts 120 units and 114 pass first time, while six are reworked and all later pass. R has higher FPY: 98%. S has 95%. Yet S ends with 120 accepted units while R ends with 98.
The lesson is not that FPY is bad. It is that different metrics answer different questions. FPY answers a first-pass process question. Final accepted count answers an output question. Neither should silently impersonate the other.
Worked case 3: the inspection rule changed
In January, a surface mark longer than 2 mm counts as a first-pass failure. In February, the rule changes and only marks longer than 4 mm fail. FPY rises from 92% to 97%.
Can we conclude the manufacturing process improved? Not from the percentages alone. The measurement rule changed. Some or all of the apparent improvement may come from reclassification rather than a physical change in the process. Before comparing the two months, restore a common definition or explicitly account for the change.
A PSLE-style transfer case
An original table shows the production of two fictional teams:
| Team | Units entering test | Pass first time | Pass after rework | Scrapped |
|---|---|---|---|---|
| A | 100 | 92 | 7 | 1 |
| B | 100 | 96 | 2 | 2 |
A poster says, “Team B produces more good products because its first-pass yield is higher.” Evaluate the claim.
Team B does have higher first-pass yield: 96% versus 92%. But final accepted output is 99 for A and 98 for B. The poster has used a first-pass metric to make a final-output claim. A stronger conclusion is that B had better first-pass performance under the stated test, while A had slightly more accepted units after rework. More information would be needed to judge overall process quality or cost.
Tempting reasoning that fails
| Tempting statement | Evidence problem |
|---|---|
| “95% FPY means 5% of customers get defective products.” | First-pass failures may be reworked, rejected or never shipped. |
| “The final shipment is 100% accepted, so FPY does not matter.” | Final acceptance can hide rework and process burden. |
| “Factory A has a bigger FPY number, so everything about A is better.” | One metric cannot own cost, safety, reliability, speed and final quality. |
| “FPY rose, so the physical process improved.” | The definition, test or product mix may have changed. |
How far can the conclusion travel?
A well-supported FPY statement can travel to the defined process, product family, period and pass rule. It cannot automatically travel to every future batch, every stage, every factory, every customer outcome or every quality dimension. The broader the conclusion, the more bridges of evidence you need.
This is exactly the discipline PSLE Science asks for when a learner evaluates whether available evidence is sufficient for a conclusion: do not throw away useful evidence, but do not make it carry more than it measured.
Delayed independent return
Come back later and answer without rereading: a line starts 500 units. 475 pass without rework. Twenty are repaired and later accepted; five are scrapped. What is the first-pass yield? How many units are finally accepted? Why are those two answers not competing answers?
The answers are 95% FPY and 495 accepted units. They describe different stages of the evidence chain.
Explained practice
Practice 1. “FPY rose from 90% to 96%, so defects in shipped products fell by exactly 6 percentage points.” Not established. We need final inspection, rework and shipment evidence.
Practice 2. “A 99% FPY guarantees the next item will pass.” No. A batch or process percentage is not a certainty statement about one future unit.
Practice 3. “Two factories both report 97% FPY, therefore their processes are equally good.” Not yet. Match process boundary, pass criteria, product type and time period first.
Route to existing owners
- Reality Lab Vol.121 | “The Process Is In Control” — Does That Mean Every Product Meets the Specification? — for control versus specification.
- Reality Lab Vol.239 | “AQL 1.0” — Does That Guarantee At Most 1% Defects? — for lot acceptance versus unit-level guarantees.
- How to Evaluate a PSLE Science Experiment and Improve the Method — for the underlying method-evaluation job.
- How Industrial Engineering Works | Master Edition — broader concept ownership for production flow, yield and rework.
Parent and tutor teaching guide
Use a simple paper-folding task with ten sheets. Define one measurable pass rule, such as whether the fold lands inside a marked band. Count how many pass on the first attempt. Allow correction for failed sheets and count how many become acceptable after correction. The learner now has two legitimate quantities: first-pass success and final accepted output.
Then change the rule deliberately. Widen the acceptable band and ask why FPY may rise even if folding skill does not. This makes the hidden measurement-definition problem visible without requiring industrial jargon.
Finish by showing only “95%” and asking, “95% of what, at which stage, under what pass rule?” The learner should become uncomfortable with a naked percentage. That discomfort is productive scientific scepticism.
Why this belongs in the 2026 PSLE Science frame
The current 2026 PSLE Science assessment objectives require learners to interpret and analyse information, evaluate observations, information and methods, and communicate explanations and reasoning. The 2023 Primary Science syllabus advocates healthy scepticism, objectivity and integrity in handling and communicating data. A quality dashboard is a real-world scientific communication object in which those habits matter.
No examiner rule says you must write “first-pass yield” in a PSLE answer. The value lies in transfer: learn to ask what a percentage counts, what it excludes and whether the conclusion matches the process that produced it.
Authoritative sources and further reading
- Singapore Examinations and Assessment Board — 2026 PSLE Science syllabus.
- Ministry of Education Singapore — 2023 Primary Science Teaching and Learning Syllabus.
- American Society for Quality — Quality Glossary, First-Pass Yield.
- NIST Manufacturing Extension Partnership — manufacturing case material discussing first-pass yield as an operational quality signal.
Quiet return
A green 95% can look like an answer. In Science, it is often the beginning of the question. What entered? What passed? What was allowed to happen next? Once you can reconstruct that hidden process, the percentage becomes evidence instead of decoration.
