How do I calculate first-pass yield without hiding rework?

Calculate first-pass yield by dividing units that meet the defined process requirements on their first pass, without repair or repeat processing, by the unique units entering that process. Record rework and scrap separately. A repaired unit can become good finished output without becoming a first-pass success.

The practical question is how much work was needed to produce the acceptable result. If the dashboard reports only what eventually passed, the hours spent fixing earlier failures can disappear behind an attractive percentage.

Define the process and the unit before counting

Specify the process boundary, counted unit, quality requirements and measurement cutoff. A physical item, a production attempt, a defect observation and an inspection record are different things. Give each unit a reference that connects its first result with later work.

ASQ's quality glossary defines first-pass yield around units meeting requirements without scrap, repeat processing, retesting or offline repair. Its calculation subtracts defective units from units entering the process and divides by the entry count.

Keep the standard available to the people doing and checking the work. If one shift uses a different specification, its percentage may answer a different question. The standards guide addresses making the expected result understandable; this article concerns counting what actually happened against that defined result.

Set the cutoff so the outcome is interpretable. Units still awaiting their required check should not be quietly marked passed. Keep their status visible and obtain the missing evidence rather than treating uncertainty as either success or employee failure.

An original batch of 120 units

Consider a wholly fictional workshop producing 120 identical wooden display blocks. Each has an approved dimension and finish specification, and every unit receives the defined first check. These are supplied complete observations, not results from an actual workshop test.

Of the 120 unique units, 108 pass the first check. Twelve fail. Ten of the twelve can be repaired and each passes one required repeat check after repair. Two are unusable, cannot be repaired or sold and have no recovery value in this case.

At the final cutoff, all 120 units have resolved outcomes: 108 passed first, ten became conforming after repair and two were scrapped. There is no unfinished work, uninspected unit, missing item or customer return in the main example.

These outcome groups reconcile to 120. Do not start the calculation from the number that eventually looked good and then try to infer how many initially failed.

First-pass yield and final yield answer different questions

First-pass yield is 90 percent: 108 first-pass successes divided by 120 unique units entering the process.

Final conforming output is 118 units: 108 plus the ten recovered through repair. As a share of the 120 starts, the final conforming yield is about 98.33 percent.

Both percentages can be correct. Reporting 98.33 percent as first-pass yield would hide the ten units that needed additional work. Reporting only 90 percent without the repair outcome would fail to explain that ten more units ultimately became usable.

Under this case's definitions, ten of 120 units entered rework, or about 8.33 percent. Two of 120 were scrapped, or about 1.67 percent. Those groups, together with the 90 percent first-pass group, cover the complete batch here.

Do not assume every real report will partition the same way. It may include units still under review, repeated repairs or a different process boundary. Read the actual statuses and definitions before adding percentages.

Fifteen defects do not mean fifteen bad units

The fictional first-check log contains fifteen defect observations across the twelve failed units. Nine units have one recorded defect each, and three have two each: nine plus six observations gives fifteen.

There are still only twelve distinct failed units. Subtracting fifteen from 120 would wrongly produce 105 first-pass successes and an 87.5 percent yield. That substitutes defect observations for defective units.

Preserve the observations because they can guide investigation, but connect them to their unit references. A unit with two problems may require a different repair from a unit with one, even though each counts once in the first-pass denominator.

Likewise, a high defect count in one category does not by itself establish its cause. The example supplies neither a defective supplier lot nor an employee error finding. Investigate the conditions and records before assigning responsibility.

Repeat checks do not create more starts

The original 120 first checks plus ten repeat checks create 130 inspection records. They concern 120 unique original units, not 130 new units entering the process.

There are 118 passing inspection results: 108 first passes and ten later passes. Dividing 118 by 130 gives about 90.77 percent. That is the share of these inspection records showing a pass; it is neither the defined 90 percent first-pass yield nor the 98.33 percent final conforming yield.

Label the metric you actually calculated. A system that counts every visit as another item can make repairs look like production volume, even though no new physical unit was created.

Keep the first-check result and the final disposition in different fields. Overwriting fail with pass after repair destroys the distinction. Add the repair and repeat-check evidence while preserving how the item first performed.

Put the repair hours into paid capacity

The workshop has 48 paid staff hours available for this defined batch and its necessary supporting work. Every hour costs a complete $30, so payroll is $1,440 whether the last available hour is used or not. No unpaid owner work or displaced customer commitment fills the plan.

Initial production takes fifteen minutes per unit, or thirty hours for 120. First checking takes three minutes per unit, totaling six hours. All setup, recordkeeping and other necessary routine work take another eight hours.

Before rework, those tasks use 44 hours, leaving four paid hours. Repairing and rechecking each of the ten recoverable units takes eighteen minutes in total, or 0.3 hour. The ten repairs therefore use another three hours.

Total work is 47 hours against 48 available, leaving one hour. The ten usable units recovered through repair are real output, but they did not arrive without consuming capacity.

An extra task cannot be slipped into an unlimited gap merely because regular payroll is already funded. Establish its actual time, materials and effect on the other work. In this main batch, only the stated one hour remains unassigned.

Follow the money without paying the same labor twice

Initial materials cost a complete $6 for each of 120 units, totaling $720. Rework uses another $2 of materials per recovered unit, adding $20. Total material spending is $740.

Another $200 covers all remaining nonlabor batch charges, including workspace, other consumables and disposal of the two scrapped units. All recordkeeping labor is already included in the paid hours. There are no other receipts, payments or resources in this finite batch case.

Total spending is $2,380: $1,440 payroll, $740 materials and $200 other charges. The workshop has $2,500 available specifically for the batch after protecting every other obligation. It can pay all $2,380 before any customer receipt and retains $120.

The three rework hours represent $90 of paid capacity. That $90 is already inside the $1,440 payroll. Adding it again would incorrectly increase spending to $2,470.

It is still useful to identify $110 of rework resource use: $90 labor and $20 additional materials. Just label it as an allocation of resources, with the labor already funded, rather than another complete payment on top of the batch total.

ASQ's cost-of-quality explanation includes scrap and rework among internal failure costs when problems are found before customer delivery. The workshop's figures are an original finite resource example, not a complete cost-of-quality program.

Keep spending per good unit in its place

Divide the batch's $2,380 spending by its 118 conforming units and the result is about $20.17 per conforming unit. That includes the supplied paid capacity, materials and other batch charges.

This is a spending ratio for the stated case, not an accounting inventory valuation, selling price or company profit. No sales, customer collections or future use of the units are supplied.

Dividing by 120 instead would give about $19.83 per unit started, a different measure that includes the two scrapped starts in its denominator. Neither ratio turns those unusable units into saleable output.

Also avoid assuming that eliminating repair would save the whole $110 immediately. The regular 48-hour payroll is committed in this case. Releasing three hours could create usable capacity if there is suitable work for it, while avoiding the $20 material purchase would change the specified cash spending. Neither benefit establishes an actual future order.

Use the result to investigate a specific problem

Begin with the failed unit records, defect descriptions, process conditions and repair outcomes. Identify what information is needed to distinguish a material problem, unclear specification, process variation or another cause. A percentage alone does not choose the remedy.

Keep containment and investigation separate. A questionable unit needs the appropriate disposition before it is released, even while its cause remains open. Passing the repair check does not prove the underlying failure has been prevented in the next batch.

If the standard or inspection method changes, preserve that version and explain the affected comparison. A more favorable percentage after a different test does not by itself establish better production.

Retain the original observations when evidence is corrected. Record the unit, old classification, reason for the change and authorized reviewer. A dashboard should be reproducible from its records, rather than depend on whoever last edited the total.