Quality Rate Calculation

The quality rate is the share of a line's output that comes out good on the first attempt. Quality rate = good units made right the first time ÷ total units produced. It is the third factor in OEE multiplied against availability and performance, and it counts every reworked unit as a loss.

That last clause is where most quality-rate numbers quietly go wrong. A line stamps out 1,000 parts, sends 50 to the rework bench, fixes 45 of them, and scraps 5. Ask the floor and they will tell you quality was 99.5%, only five parts lost. OEE says it was 95%, because the 50 parts that failed the first check were not good the first time, no matter how they ended up. This guide walks the quality-rate calculation step by step, shows why reworked units do not count, lists what actually feeds the number, and lays out how to move it.

What is the quality rate in OEE?

The quality rate is the percentage of total production that passes clean on the first attempt, with no rework, repair, or adjustment after a failed check. In OEE it is one of three multiplied factors: OEE = Availability × Performance × Quality. Availability accounts for time the line was stopped, performance for time it ran slow, and quality for the units it made but could not sell as-is.

Because OEE measures quality from a first-pass-yield perspective, the quality factor and first pass yield rest on the same rule: a unit either passed clean the first time or it did not. The difference is scope. FPY is usually tracked per process step and compounds across a line; the OEE quality rate rolls the reject count up to one machine or line so it multiplies cleanly into the OEE score.

How do you calculate the quality rate?

Count the good first-time units, divide by the total units the machine produced, and express the result as a percentage. In words: quality rate = (total count − reject count) ÷ total count, where the reject count includes every unit scrapped and every unit sent to rework.

A worked example with hypothetical numbers. A filling line runs a shift and the counters read like this:

Notice the 50 reworked units sit in the reject column even though they will ship tomorrow. That is deliberate. Move them into the good column and the quality rate climbs to 96.25% and the line looks better than it ran. The entire purpose of the factor is to refuse that move.

Why don't reworked units count as good?

Because rework is a loss even when the unit is saved. A reworked part consumed the line's capacity once to make it wrong, then consumes labor, floor space, and schedule a second time to make it right. Counting it as good erases both the defect and the double cost, which is exactly what a quality metric is supposed to expose. If reworked units were good, a plant could report near-perfect quality while running a full-time rework crew, and plenty do exactly that when the definition slips.

This is why the OEE quality factor and first pass yield share the same logic and why a plant that wants an honest number pairs the quality rate with defect tracking. The gap between what a line ships and what it makes right the first time is the hidden factory, and that gap is a major slice of cost of quality.

The American Society for Quality reports that the cost of poor quality commonly runs 15–20% of sales revenue, reaching as high as 40% at poor performers much of it the inspection, scrap, and rework that a soft quality rate hides. A quality rate that counts reworked units as good is not just an accounting choice; it is the choice to stop measuring a fifth of your revenue.

What counts against the quality rate?

Two of the six big losses are quality losses, and together they define everything that lowers the number:

• Production defects (process rejects). Scrap and rework produced during stable, steady-state running. These are the losses most plants already count.
• Startup rejects (reduced yield). Parts lost during warm-up, after a changeover, and after any stoppage while the process settles back to spec. These are the losses most plants under-count, because they feel like a normal cost of starting and never get logged as rejects.

Both categories belong in the reject count. Leaving startup rejects out is the most common way a quality rate reads high while the scrap bin fills. If your line makes 60 off-spec units every time it restarts and restarts eight times a day, that is 480 units the quality rate should see and usually does not.

One more distinction matters here. A defect that is caught, scrapped, and never leaves the plant is a quality-rate loss but not a customer problem. A defect that escapes to the customer is both, and it is far more expensive per unit. The quality rate does not by itself tell you which defects escaped, so pair it with escape data, returns, complaints, and final-check results, and treat a rising escape rate as a louder alarm than a rising internal reject rate. The two move together often enough that a slipping quality rate is usually the earliest warning that escapes are coming.

What is a good quality rate?

Benchmark a line against its own trend, not against a headline figure. The widely cited world-class OEE target of 85% assumes a quality rate near 99.9%, but that assumes a strict, honest count, and a strict 96% that is climbing tells you more than a soft 99.9% that has quietly reclassified rework as good. Use published targets to size the gap, then track your own line.

Two cautions. First, a quality rate near 100% on a line that obviously reworks parts is not a triumph; it is a counting problem, and the fix is to tighten the definition, not celebrate. Second, quality rate is a factor, not the whole story: pair it with the rest of your manufacturing KPIs and the overall OEE score so a good quality number does not paper over an availability or performance problem.

How do you improve the quality rate?

Improving the quality rate is mostly about seeing the defects clearly, then removing the biggest cause and holding the gain. A dependable sequence:

1. Fix the definition before you chase the number. Write down that rework, repair, retest, and startup rejects all count against quality. A quality rate built on a soft definition improves by relabeling, not by making better parts.
2. Separate startup rejects from production defects. They have different root causes, startup is about process stabilization and changeover, production defects are about steady-state variation, and lumping them hides which problem to work.
3. Attach a defect code to every reject. A quality rate without reasons is a scoreboard without a game plan. A short, disciplined code list feeding defect tracking turns the number into a ranked list of causes.
4. Pareto the codes and work the top one. Most of the loss lives in a few codes. Attack the largest with root-cause analysis and statistical process control on the variable that drives it.
5. Mistake-proof the top failure mode. Where the cause is human error or a settable parameter, design the error out with a poka-yoke or an interlock so the defect cannot recur.
6. Re-measure weekly and hold the gain. Post the quality rate where the crew can see it, watch the worked code shrink, then move to the next one. Improvement that is not re-measured drifts back.

How does the quality rate fit the rest of the plant?

The quality rate is where quality shows up in the time-and-output ledger, so it never stands alone. Reworked units consume capacity twice, which drags on throughput; defect investigations stop lines, which lands in the downtime log; and the whole picture only holds together if the counts are captured cleanly enough to trust. A quality rate assembled from paper tally sheets at the end of a shift is an estimate wearing a decimal point.

That is the practical catch: the quality rate is only as honest as reject logging, and a ten-second touch-up at the station takes thirty seconds to write down, so it usually is not written down. Plants that capture counts and rejects digitally at the point of work, the way Harmony turns paper checks and quality logs into live, searchable records feeding production reporting (see the platform), find their real quality rate within weeks. It is usually a few points lower than the reported one, and knowing that is the start of fixing it. For a worked look at replacing paper logs with real-time capture, see the CLS case study.