A plant quality dashboard usually fails the same way: it fills up with indicators nobody is accountable for. Twenty neat charts, a monthly review, and when something goes wrong on the line nobody finds out until the customer calls.
The quality KPIs that actually run a manufacturing operation are few, they are calculated from data the plant already produces, and each one has a person’s name next to it. You do not need an MES or a six-month project to start: you need to decide what you measure, with which formula, and how often you look at it.
Below are the ten that get used on the floor, with the formula, a typical reference range and a sensible review frequency. None of them is a literal ISO 9001 requirement: the standard asks you to measure process performance and analyse the results (clause 9.1), but it lets you choose which ones.
Table of Contents
Why plant KPIs are not the same as your QMS indicators
A documented management system measures things like audit programme completion or corrective action progress. Those are fine, but they are management indicators: they tell you whether the system is being run, not whether the product is coming out right.
Manufacturing quality KPIs measure the product and the process. They answer three concrete questions: how much am I throwing away, how much am I redoing, and how much is reaching the customer. A system can have every audit up to date and still ship defective parts.
The practical consequence is that you need both families, on separate dashboards. The plant one gets reviewed per shift or weekly with production. The QMS one goes monthly to management. Mixing them into one sheet is exactly what makes nobody look at either.
The 10 quality KPIs in manufacturing, with formulas
The reference values below are typical industry ranges, not limits set by the standard. Your correct range comes from your own history: measure for three months, take your average and set the target from there.
| KPI | Formula | Typical reference | Frequency |
|---|---|---|---|
| Scrap rate | (Units scrapped ÷ units produced) × 100 | 1% to 5% depending on process | Daily or per shift |
| Rework rate | (Units reworked ÷ units produced) × 100 | Under 3% | Daily |
| First pass yield (FPY) | (Good units with no rework ÷ units started) × 100 | Above 95% | Daily |
| Defects per million (PPM) | (Defective units ÷ units produced) × 1,000,000 | Sector-dependent; automotive demands very low figures | Weekly |
| Supplier PPM | (Parts rejected at receiving ÷ parts received) × 1,000,000 | Set contractually per supplier | Monthly |
| On-time delivery (OTD) | (Deliveries on date ÷ total deliveries) × 100 | Above 95% | Weekly |
| Cost of poor quality (COPQ) | (Internal + external failure + appraisal cost) ÷ sales × 100 | Between 2% and 10% of sales | Monthly |
| Complaints per million units | (Complaints received ÷ units shipped) × 1,000,000 | A sustained downward trend | Monthly |
| Nonconformity closure time | Average days from opening to verified closure | Under 30 days | Monthly |
| Audit programme completion | (Audits performed ÷ audits planned) × 100 | 100% | Quarterly |
Scrap, rework and FPY: the three you cannot skip
If you could only track three, these are the ones. Scrap tells you how much material went in the bin, rework tells you how much labour you paid for twice, and FPY rolls both into one number management understands without a briefing.
Watch one common trap: if you count reworked units as good production, your FPY comes out inflated. First pass yield counts only what came out right without being touched again. That is precisely the number that exposes the hidden cost of a line that looks stable.
PPM and supplier PPM: the language your industrial customers speak
Parts per million is the unit manufacturing customers use, especially in automotive, electronics and medical devices. If a customer asks you for a quarterly quality report, it will come back in PPM.
Supplier PPM is the same metric pointed back up the chain. It is what turns supplier evaluation into something objective: instead of an opinion about who answers the phone fastest, you get a figure per supplier per period.
COPQ: the indicator that unlocks budget
Cost of poor quality adds up what your failures cost. Internal failures (scrap, rework, downtime), external failures (returns, warranty, expedited freight, customer credits) and appraisal (inspection, testing, lab work).
It is the only KPI on this list expressed in money, which is why it moves decisions outside the quality department. Presented as a percentage of sales, the conversation stops being about procedures and starts being about margin.
Worked examples with real numbers
First pass yield. A line starts 4,200 parts in a week. 3,940 come out right the first time; 180 are reworked and end up conforming, 80 are scrapped. FPY is 3,940 ÷ 4,200 × 100 = 93.8%. Counting the reworked units as good would give you 98.1% and hide the 180 parts you paid for twice.
Defects per million. 86,000 parts produced in the month, 12 rejected by the customer. PPM is 12 ÷ 86,000 × 1,000,000 = 140 PPM. That sounds small until you compare it with the limit in your contract.
Cost of poor quality. Monthly sales: 1,200,000. Scrap 14,000, rework 9,500, expedited freight for replacements 6,200, inspection 11,000. Total 40,700. COPQ is 40,700 ÷ 1,200,000 × 100 = 3.4% of sales. Taking one point off is worth 12,000 a month, and that is the sentence that unlocks budget.
Nonconformity closure time. 18 nonconformities closed in the quarter. If the days add up to 684, the average is 684 ÷ 18 = 38 days. The average lies here: also look at how many went past 60 days, because those are the ones the auditor will ask for.
How to build the dashboard in 6 steps
- Pick six indicators at most to start. Scrap, rework, FPY, OTD, complaints and COPQ cover about 80% of what a small manufacturer needs to see.
- Define where each figure comes from. Which record, who captures it and at what point in the shift. If the data depends on someone remembering, the indicator will break in month two.
- Set the target from your own history. Measure for three months, take the average and set the target one step above it. Copying another company’s benchmark produces targets nobody believes.
- Assign one owner per indicator. A person, not a department. “Production” does not answer in a meeting; the line supervisor does.
- Set the review frequency and keep it. Daily for floor metrics, weekly for delivery, monthly for the money ones. An indicator reviewed when there is time is not reviewed.
- Write down what happens when it goes off target. This is the step almost nobody takes and the one that separates a dashboard from wall decoration: if scrap goes over 5% two weeks running, a nonconformity is opened and root cause analysis starts.
Mistakes that make a plant dashboard useless
Tracking twenty things. With twenty indicators nobody looks at any of them. Six measured properly beat twenty measured halfway, every time.
Calculating from different sources. If production counts parts one way and quality another, the indicator becomes an argument instead of a decision. One source per figure.
Reviewing results with no action. A meeting where numbers are read out loud and no dated task comes out of it is a meeting you can cancel at no cost.
Leaving everything in scattered spreadsheets. It works for three months. After that nobody knows which version is the good one, one person owns the file, and when the audit arrives there is no way to show how last quarter’s number was calculated.
How this connects to ISO 9001
Clause 9.1 asks you to determine what needs to be measured, by which methods, when to measure it and when to analyse it. Nothing more, nothing less. It carries no mandatory list of indicators, so the auditor will not ask you about FPY: they will ask why you chose the ones you chose and what you did with the results.
That second part is where it usually falls apart. Having the data is easy; showing that you analysed it and that something changed because of it is what gets audited. That is why these indicators feed straight into management review: they go in as a required input and come out as decisions with resources attached.
The full map of what the chapter asks for is in ISO 9001 clause 9. And if what you need is the general catalogue of quality indicators rather than just the plant ones, it is in the quality KPIs guide with 10 formulas.
From spreadsheets to a dashboard that updates itself
There is a point where the problem stops being what to measure and becomes who updates the file. When the dashboard depends on somebody consolidating four sheets every Monday, the measurement collapses on its own during the first busy week.
That is where quality management software changes the equation: data goes in once where it happens, the indicator calculates itself, and the nonconformity triggered by an off-target indicator stays linked to the figure that caused it. If you run a manufacturing operation, you can see what that looks like in QualityWeb 360 for manufacturing.
Frequently asked questions about quality KPIs in manufacturing
How many production KPIs should a small manufacturer have?
Between four and eight to start. The right number is the one your team can review in full in a thirty-minute meeting. If there is not enough time, you have too many indicators.
Does ISO 9001 require specific indicators?
No. Clause 9.1 asks you to determine what to measure, how, when, and to analyse the results, but it sets no list. The criterion for choosing them is that they support decisions about your processes.
What is the difference between scrap and rework?
Scrap is thrown away, rework is recovered. Both cost money, but differently: scrap costs you the full material, rework costs you labour and line capacity. That is why they are tracked separately.
How often should manufacturing quality KPIs be reviewed?
Floor metrics daily or per shift. Delivery and service metrics weekly. Anything expressed in money, monthly. And all of them reach management review at least once a year.
Can COPQ be measured without a detailed cost system?
Yes, with a simple version: valued scrap plus rework hours plus extraordinary delivery costs. It is not exact, but the trend is right, and the trend is what you use to decide.
What if an indicator has been on target for months and never moves?
Two possibilities: the target was set too loose, or the indicator has done its job. Either way it needs adjusting. An indicator that is always green is not telling you anything.
