First-pass yield KPI: formula and production example

The first pass yield KPI is the percentage of units that pass a defined operation without rework. Divide first-pass good units by the units entering that operation, and retain each unit's initial result even when a later repair succeeds. The measure shows how much production is right the first time.

The formula and the boundary

First-pass yield = Units passing without rework / Units entering the defined process × 100

For a fully inspected cohort: Entering units = First-pass good units + Units requiring rework + Immediate scrap units

Define the process before calculating the ratio. An inspection after machining measures the machining operation; an inspection after final assembly covers a different boundary. Neither is automatically the yield of the whole factory.

Define rework consistently too. A unit requiring corrective processing to meet its original specification has failed the first pass. Routine steps in the approved production route do not become rework simply because the unit passes through several stations.

First-pass yield is a KPI when it guides a quality or production decision. It should point to a review of defect causes, fixtures, material or work instructions, rather than become a percentage with no action owner. NIST's process improvement guidance describes quality improvement in terms of identifying defect causes and reducing variation.

The export rows you need

Use a production and inspection export with one original entering record per physical unit and operation, or a reconciled lot summary covering the same population. For a lot summary, include these fields:

Retain the first result separately from the latest status. The O*NET inspector profile includes recording inspection data and marking acceptance or rejection status. A file containing only today's accepted stock cannot reconstruct which units originally failed.

Use closed cohorts for the simple identity below. If units are awaiting inspection or disposition, show them separately and label the yield provisional. Do not silently call an unresolved unit a pass, a failure or a zero.

Worked example: two completed production lots

An illustrative Indiana component manufacturer compares two lots at the same operation. Every entering unit has its first inspection result. There are no pending dispositions, and each unit belongs to exactly one initial outcome category.

Percentages in this example use three decimal places so the reconciliation remains visible. Calculate from unrounded counts and round only the displayed result.

If entering units are in column B, first-pass good units in C, rework in D and scrap in E, the lot yield in F2 is:

=IF(B2=0,"",C2/B2)

Format the result as a percentage and fill down for the other lot. A blank here means there is no entering population to measure; it does not mean poor yield. The combined yield is:

=IF(SUM(B2:B3)=0,"",SUM(C2:C3)/SUM(B2:B3))

Do not average the two percentages. The simple average is (95.000% + 90.000%) / 2 = 92.500%, which gives the smaller lot too much weight. The entering-unit-weighted result is (500 × 95.000% + 300 × 90.000%) / 800 = 93.125%, exactly matching the ratio of totals.

The checks that prove the result

First, reconcile the initial disposition of each lot:

In the spreadsheet, the count control for each row is:

=B2-SUM(C2:E2)

It must return zero for a complete, mutually exclusive disposition set. A zero does not prove that the source records are unique, so also check repeated unit-operation identifiers or overlapping lot extracts. Reconcile the entering count with the production log independently of the inspection summary.

Then separate the failure burden. Rework required is 44 / 800 = 5.500% of entering units; immediate scrap is 11 / 800 = 1.375%. Their sum is the same 6.875% first-pass failure rate. This split helps production distinguish potentially recoverable work from immediate material loss.

The broader automotive manufacturing KPI example puts that quality result beside recorded downtime and unit spending. Keep those measures distinct: the yield alone does not reveal how long a repair took or what it cost.

Final yield answers a different question

Suppose 40 of the 44 rework-required units are recovered and the remaining 4 are scrapped. First-pass good units remain 745; successful repair cannot change the history of the first attempt.

Final yield is 785 / 800 = 98.125%, while first-pass yield stays 93.125%. That difference reflects recovered units. It does not mean the earlier first-pass report was wrong.

A multi-operation rolled throughput measure raises another question: which units completed the entire route without rework? Do not multiply unrelated monthly percentages and assume the result describes the same population. Route changes, skips and different product mixes require a separate review.

Change one input: fewer rework-required units

Now consider a separate variation on the initial inspection example. Hold L2's entering units and immediate scrap fixed, but reduce its rework-required units from 24 to 12. Its first-pass good count is the residual: 300 − 12 − 6 = 282. L1 remains unchanged.

The changed L2 check is 282 + 12 + 6 = 300. The combined check is 757 + 32 + 11 = 800. The improvement is 12 / 800 × 100 = 1.500 percentage points. This variation represents fewer units needing rework on their first attempt, not a decision to relabel previously repaired units.

Where first-pass yield goes wrong

• Counting rework visits as new entrants. Count original units once at the defined operation.
• Overwriting initial failures with final passes. Preserve both events so repair recovery does not inflate first-pass quality.
• Changing inspection scope. A new test or different sampling method changes what the percentage measures; label the break in comparability.
• Mixing entry and completion periods. Compare the same production cohort, particularly when inspection or rework crosses the reporting cutoff.
• Punishing appropriate stops. The jidoka production principle includes stopping at abnormalities to prevent defects. Review the cause before treating a stop as poor operator performance.

For individual reviews, the machine operator KPI guide explains how product mix and equipment availability change the interpretation of shift results.

Calculate first-pass yield from your own file in Covirage

Upload a production-quality export or Google Sheet with entering units, initial dispositions and rework outcomes. Covirage maps the columns, and its tools compute and check the figures; the external AI model explains the comparisons with every figure traced to its rows. Create a free Covirage account, set up your business and bring a sample of your own data.

See Covirage for industrial manufacturers for the wider production reporting context.