An assembly that passes inspection after rework counts as a good part in many plants. In the monthly figures it looks exactly like an assembly that was built right the first time. That is the problem: the labor, the equipment time and the second inspection it consumed never show up in the output. The rework rate makes this hidden effort visible. If you calculate it cleanly and interpret it correctly, you find capacity that currently disappears into correction loops.
- The rework rate is the share of units that are brought back into a condition that meets requirements through rework after a detected defect, relative to all units produced.
- The formula is: rework rate (%) = reworked units ÷ units produced × 100.
- There is no reliable public benchmark for a normal rework rate across industries. Sigma level, first pass yield and your own trend provide the frame of reference.
- The rework rate falls permanently when defects are recorded where they occur, causes are analyzed systematically and processes are safeguarded through process data, error proofing and digital worker guidance.
Calculate
Count every unit that was reworked at least once, separately for each process step and defect type.
Interpret
Always evaluate rework rate, scrap rate and first pass yield together.
Reduce
Start with the most frequent defect type and prove the effect with data.
Rework is part of daily life in many production plants. A screw joint does not reach its torque, a connector is not fully seated, a dimension is just outside tolerance. The part goes to a rework station, is corrected, inspected again and moves on. At the end of the day the quantity is right.
Cost and capacity are not. Richard E. Crandall and Oliver Julien, both professors at Appalachian State University, explain in the magazine of the Institute of Industrial and Systems Engineers (IISE) that many companies do not know their actual quality costs because they are hard to measure. They explicitly count rework among internal failure costs. Without a clearly defined metric, this cost item stays invisible.
This article covers the formula with a worked example, how the rework rate relates to scrap rate and first pass yield, sourced benchmarks for interpretation and seven measures that manufacturers use to reduce the rework rate permanently. It is written for those responsible for quality and production in automotive, mechanical engineering and medical technology.
01 · Basics
What is the rework rate?
The rework rate is a quality metric that shows what share of the units produced had to be reworked after a detected defect before meeting the requirements. It does not measure how many parts are good in the end, but how many needed a second attempt to get there.
The terminology comes from ISO 9000:2015, the standard on fundamentals and vocabulary for quality management systems. Clause 3.12.8 defines rework as action on a nonconforming product or service to make it conform to the requirements. The European Union Agency for Railways (ERA) lists this definition in its terminology collection. Clause 3.12.9 separates repair from it: repair makes a nonconforming product acceptable for its intended use without it having to meet all requirements.
This distinction matters for the metric. If you mix repairs into the figure, you count parts that meet the specification again together with parts that are merely usable. In the automotive industry, IATF 16949 therefore treats both cases in separate clauses: 8.7.1.4 covers reworked product and 8.7.1.5 covers repaired product.
| Term | What happens to the part | Meets requirements afterwards | Record in |
|---|---|---|---|
| Rework | The defect is corrected | Yes | Rework rate |
| Repair | The part is made acceptable for its intended use | Not necessarily, often only with a concession | Separate metric, not in the rework rate |
| Scrap | The part can no longer be used for its intended purpose | No | Scrap rate |
02 · Formula
How do you calculate the rework rate?
The basic formula is simple. What matters is that numerator and denominator share the same frame of reference: the same process step, the same period, the same product family.
Rework rate (%) = number of reworked units ÷ number of units produced × 100
Count every unit that was reworked at least once. If a part has to be reworked twice, it still counts as one unit. Record the number of rework operations as well, because it shows how often the first correction is not enough.
Three variants of the metric
| Variant | Formula | Best used for |
|---|---|---|
| By units | reworked units ÷ units produced × 100 | Comparing lines, shifts and defect types on the shop floor |
| By time | rework hours ÷ productive manufacturing hours × 100 | Capacity planning and staffing |
| By cost | rework cost ÷ manufacturing cost × 100 | Budget, business case and cost of poor quality |
The unit based figure is the common control metric on the line. The time and cost variants translate the same information into the language of production planning and controlling. For a sound decision you need at least the unit based figure and the cost based figure.
Worked example: calculating the rework rate
1
Input data
An assembly line builds 20,000 assemblies per month. 600 of them go to the rework station after an inspection failure, another 120 are scrapped.
2
Rework rate
600 ÷ 20,000 × 100 = 3.0 %. The scrap rate is 120 ÷ 20,000 × 100 = 0.6 %.
3
First pass yield
(20,000 less 720) ÷ 20,000 × 100 = 96.4 %. That many assemblies were finished on the first pass without correction.
The final yield in this example is 99.4 %, because the 600 reworked assemblies count as good parts in the end. The first pass yield of 96.4 % shows how many assemblies passed the process correctly the first time. The difference of three percentage points is exactly the share of rework. The figures are a worked example, not industry values.
03 · Distinction
How are rework rate, scrap rate and first pass yield related?
The three metrics describe the same process from different angles. The scrap rate shows material loss, the rework rate shows correction effort, and the first pass yield (FPY) shows the share of units that pass a process step correctly the first time.
The knowledge portal iSixSigma, a widely used reference for Six Sigma methods, describes first pass yield as a metric for the level of rework: units that pass through a step again after rework are ignored in the calculation. FPY is therefore the direct counterpart of the rework rate. Where scrap occurs, the following applies to a single process step: FPY = 100 % less rework rate and scrap rate.
Across several process steps, the first pass yields multiply into the rolled throughput yield (RTY), the yield across the entire chain. iSixSigma shows how quickly this value drops:
77.38 %
RTY of a five step chain when each step reaches an FPY of 95 %
Source: iSixSigma
59.87 %
RTY of a ten step chain at 95 % FPY per step
Source: iSixSigma
A defect rate of 5 % per station sounds low. Across a chain of ten stations it means that roughly four out of ten units are flagged at least once on their way. A line can therefore show good values at every station and still tie up a large share of its capacity in rework.
Key takeaway
A low rework rate at every single station does not guarantee a good flow across the whole line. Alongside the figure per station, always track the rolled throughput yield of the entire chain.
04 · Benchmarks
What is a good rework rate? Benchmarks for interpretation
An honest answer first: there is no reliable, publicly available benchmark for a normal rework rate across industries. The rate depends on product complexity, number of process steps, inspection depth and definition. Even the question of whether a second tightening attempt at the same station counts as rework changes the result. Crandall and Julien also point out that the cost of poor quality varies strongly with product complexity.
Three frames of reference provide solid orientation: the sigma level from Six Sigma, your own trend and the cost impact.
Orientation via the sigma level
Six Sigma translates defect rates into a sigma level. The sigma table from iSixSigma assigns defects per million opportunities (DPMO) and the share of defect free results to each level. If you consider one defect opportunity per unit, the defect share can be read approximately as the sum of rework rate and scrap rate.
| Sigma level | Defects per million opportunities | Defect free results | Corresponds to a defect share of about |
|---|---|---|---|
| 3 sigma | 66,807 | 93.319 % | 6.7 % |
| 4 sigma | 6,210 | 99.379 % | 0.62 % |
| 5 sigma | 233 | 99.9767 % | 0.023 % |
| 6 sigma | 3.4 | 99.9997 % | 0.00034 % |
In the worked example above, the defect share is 3.6 %. That corresponds to a level between three and four sigma. The table is a yardstick for interpretation, not an industry average. It shows how far a process is from a target level.
Your own trend as a benchmark
The most meaningful comparison is with your own process. Establish a baseline over several months, separated by line, product family and defect type. The target then follows from cost and customer requirements, for example halving the most frequent defect type within a year. This turns an abstract figure into a verifiable goal.
Cost impact as the limit
In the end, economics decide. A process with a 2 % rework rate can be more expensive than one with 4 % if every correction requires disassembly. That is why the unit based figure should always be accompanied by the cost based one.
05 · Cost
What does rework really cost?
In the cost of quality model, rework belongs to internal failure costs. In their IISE article, Crandall and Julien include failure analysis, sorting inspection, reinspection and retesting and avoidable process losses alongside scrap and rework. The order of magnitude shows why accurate measurement pays off:
around 15 %
average cost of poor quality in manufacturing companies, relative to sales
Source: Crandall and Julien, IISE
5 to 35 %
range of these costs depending on product complexity
Source: Crandall and Julien, IISE
60 to 70 %
share of failure costs in quality costs at companies with less developed quality programs
Source: A. V. Feigenbaum, cited by IISE
Factor 10
increase in correction cost when a defect is detected only after the stage where it was created
Source: rule of ten, Business Wissen
According to the German business portal Business Wissen, the rule of ten states that the cost of correcting a defect rises by a factor of ten if the defect is detected only after the stage in which it was created. For rework this means: a defect caught directly at the station costs one hand movement. The same defect at the end of the line costs disassembly, reinspection and lead time.
What belongs in the cost of rework
- labor at the rework station, including troubleshooting
- spare parts and consumables for the correction
- reinspection and release of the unit
- equipment time blocked during the correction
- transport, buffer space and additional work in progress
- extra shifts, overtime and planning effort
- documentation required by IATF 16949 for reworked product
An overview of inspection methods and their costs is available in the article on quality control in manufacturing.
Prevent operator errors before they become rework
The free checklist “How digital is your line?” asks seven questions, one for each principle of safe worker guidance. It shows where operating errors can be prevented at the workstation instead of being corrected at the rework station.
06 · Measurement
How do you measure the rework rate so that you can control it?
A figure pulled from the ERP system once a month describes the past. The metric only becomes controllable when it is recorded where the defect is detected and linked to the process that caused it.
Six mandatory data points per rework case
| Data field | Why it is needed |
|---|---|
| Serial number or batch | Links the case to process data and traceability |
| Process step and station | Separates where the defect is created from where it is detected |
| Defect type from a defect catalog | Enables Pareto analysis instead of free text |
| Time, shift and tool | Reveals patterns by time, staff and equipment |
| Rework step and inspection result | Proves that the correction was effective |
| Effort in minutes | Provides the basis for the time and cost variants |
In the automotive industry, part of this is mandatory. IATF 16949 clause 8.7.1.4 requires a documented process for rework confirmation and documented information on the disposition of reworked product, including quantity, disposition, disposition date and traceability information. The list of mandatory documents from Advisera, a provider of standards knowledge and documentation templates, includes both records.
What a robust data model looks like is described in the article on production traceability.
Separate the point of creation from the point of detection
A common measurement error is to assign rework to the station where it is detected, usually final inspection. The cause, however, often lies further up the line. Record both locations. Only then can you see how far a defect travels through the line before it is noticed and where a measure has the greatest effect.
07 · Causes
What drives rework in manufacturing?
Causes can be structured with the Ishikawa diagram, which groups influencing factors into the categories man, machine, method, material, measurement and environment. In assembly and joining, typical drivers appear in each category:
Man
Operating errors in high variant assembly, outdated work instructions, onboarding of new employees.
Machine
Tool wear, screwdrivers past their inspection interval, gradual process drift.
Method
Joining parameters without proof of process capability, unclear inspection sequence.
Material
Fluctuating component quality from suppliers, unfavorable tolerance stacks.
Measurement
Gauges with excessive measurement uncertainty, false rejects that trigger rework on good parts.
Environment
Temperature, contamination, cycle time pressure and disruptions in material flow.
Errors that originate with people are often a sign of a process that allows wrong actions. For screw joints, the torque and angle curves frequently provide the decisive clue to the cause.
If process data are stored per serial number, the team can compare the data of reworked parts with those of good parts in the workshop instead of speculating about causes. If the defective parts cluster at one screwdriver, in one shift or after a tool change, the cause is narrowed down. If a suspected cause cannot be proven in the data, it is probably not the right one.
08 · Measures
Which seven measures reduce the rework rate?
The following seven measures build on each other. The first two create the data basis, the next four prevent defects, and the last one safeguards the rework itself.
Record defect types where they occur
A defect catalog with fixed codes replaces free text. This turns the figure into a Pareto analysis that shows which few defect types account for the largest share.
Analyze causes systematically
With the Ishikawa diagram, 5 Why and the 8D method you identify the root cause instead of correcting the symptom. IATF 16949 clause 10.2.3 requires a documented problem solving process.
Prevent defects by design
Poka Yoke, meaning error proofing through fixtures, sensors or software interlocks, prevents a defect from occurring in the first place. IATF 16949 clause 10.2.4 requires a documented process for the use of such methods.
Prove process capability
Indices such as Cpk show whether a process reliably stays within tolerance. A Cpk at or above the target level is the prerequisite for low rework at a joining step.
Guide employees digitally
Digital worker guidance shows every work step in the correct order, releases tools only when the prerequisites are met and documents the result. How this reduces operating errors is shown in the article on reducing errors with digital worker guidance.
Monitor process data in real time
Capturing torque, angle, force and travel per component and checking them against limits detects deviations at the station instead of at final inspection. The article on real time monitoring in manufacturing explains the setup and costs.
Standardize the rework itself
Every rework needs an approved instruction and a reinspection. Otherwise a correction creates new defects and the figure rises somewhere else.
Preventing defects at the source is also a core principle of lean production, where defects count as one of the classic types of waste.
Six steps to a lower rework rate
| Step | Content | Result |
|---|---|---|
| 1. Standardize recording | Define a defect catalog, record every case with serial number, station and defect type | Reliable data basis |
| 2. Determine the baseline | Measure rework rate and first pass yield per station over several weeks | Reference value for proving the effect |
| 3. Set priorities | Pareto analysis by defect type and station, evaluate effort per defect type | The defect types with the highest effort |
| 4. Analyze causes | Ishikawa and 5 Why in the team, cross check with the process data of the affected parts | Proven cause per defect type |
| 5. Implement measures | Prevent defects (Poka Yoke, worker guidance) or detect them early (SPC, real time monitoring) | Measure effective in the process |
| 6. Secure the effect | Compare the metric with the baseline, update FMEA, inspection plan and work instruction | Proven reduction and new standard |
Key takeaway
The most effective measure is the one that eliminates the most frequent defect where it is created. Measures at final inspection only speed up the correction.
09 · Practice
Which mistakes slow down projects to reduce the rework rate?
Five typical mistakes from practice
- The rework rate is collected without a fixed definition, so lines and plants cannot be compared.
- Corrections directly at the station, such as a second tightening attempt, are not counted and remain invisible as hidden rework.
- Targets are set for the overall figure instead of individual defect types.
- Measures start at final inspection instead of at the point of creation.
- The effect of a measure is not proven with data.
A simple test: can you say for the most frequent defect type on your line at which station, with which tool and in which shift it occurs? If not, the first lever lies in data capture, not in a new measure.
Build the basis for less rework
The checklist “How digital is your line?” helps you check whether steps are performed in the right order, tools are released only under the right conditions and every result is documented.
10 · Solution
How does the Manufacturing OS help reduce the rework rate?
CSP Intelligence is a software company for the manufacturing industry based in Bavaria, Germany. Its Manufacturing OS is a software platform for quality assurance in manufacturing whose modules can be used individually or together. Three modules are relevant for reducing rework:
IPM
Process data management captures and monitors quality relevant process data, for example from screwing and joining processes, centrally and per component. Every rework case can then be assigned to the process value and the station where it originated. More about IPM
PG
Digital worker guidance leads employees step by step through assembly, releases tools selectively and documents every step. This reduces operating errors that would otherwise end up as rework. More about PG
QST
Quality assurance and tool testing documents the testing of screwdrivers. A worn or incorrectly set tool therefore does not go unnoticed as a cause of rework. More about QST
Together, the modules deliver exactly the data that make the metric controllable: defect type, station, tool and process value per component. Which types of software are suitable for reducing scrap and rework is compared in the article on software for reducing scrap and rework.
FAQ
Frequently asked questions
How do you calculate the rework rate?+
The rework rate is the number of reworked units divided by the number of units produced, multiplied by 100. If 20,000 assemblies are built in a month and 600 of them are reworked, the rework rate is 3.0 %. Each unit counts once, even if it was reworked several times.
What is the difference between rework rate and scrap rate?+
The rework rate covers units that meet the requirements again after a correction. The scrap rate covers units that can no longer be used for their intended purpose. Together they make up the share that does not pass the process correctly the first time.
What is a normal rework rate?+
There is no reliable public benchmark for a normal rework rate across industries, because it depends heavily on product complexity, number of process steps and definition. Sigma level, first pass yield and the trend of your own process are suitable for interpretation. According to the iSixSigma sigma table, a defect share of about 0.62 % corresponds to a level of four sigma.
What is the difference between rework and repair?+
According to ISO 9000:2015, rework makes a nonconforming product conform to the requirements again. Repair only makes it acceptable for its intended use without it having to meet all requirements. Repaired parts therefore do not belong in the rework rate.
How are first pass yield and rework rate related?+
First pass yield is the share of units that pass a process step the first time without rework. For a single process step: first pass yield equals 100 % less rework rate and scrap rate. Across several steps, the values multiply into the rolled throughput yield.
What costs does rework cause?+
Rework causes internal failure costs: labor, material, reinspection, blocked equipment time, logistics and documentation. According to Crandall and Julien in the IISE magazine, the cost of poor quality in manufacturing companies averages around 15 % of sales, with a range of 5 to 35 % depending on product complexity.
Does rework have to be documented under IATF 16949?+
Yes. IATF 16949 clause 8.7.1.4 requires a documented process for rework confirmation and documented information on the disposition of reworked product, including quantity, disposition, disposition date and traceability information. Repaired product is covered separately in clause 8.7.1.5.
Why does the rework rate rise after recording is introduced?+
The measured rework rate can rise at first because corrections that used to be made directly at the station without being recorded now become visible. The higher value is the realistic starting point against which later measures can be measured.
How can you reduce the rework rate?+
The rework rate falls permanently with seven measures: record defect types where they occur, analyze causes systematically, prevent defects with Poka Yoke, prove process capability, guide employees digitally, monitor process data in real time and standardize the rework itself.
Sources
ISO 9000:2015, Quality management systems, Fundamentals and vocabulary, clauses 3.12.8 and 3.12.9; definition of rework in the terminology collection of the European Union Agency for Railways (ERA)
IATF 16949:2016, clauses 8.7.1.4, 8.7.1.5, 10.2.3 and 10.2.4; list of mandatory documents, Advisera 16949Academy
Crandall, R. E.; Julien, O.: Measuring the cost of quality, Institute of Industrial and Systems Engineers (IISE)
iSixSigma: first pass yield, throughput yield and rolled throughput yield
iSixSigma: Sigma Performance Table
Business Wissen: Fehlerkosten berechnen und senken (mit Beispiel), rule of ten for failure costs (in German)