Cost of poor quality (COPQ) is the total money defects cost you: internal failures like scrap, rework, and quality-stop downtime, plus external failures like warranty claims, returns, and lost accounts. Build yours line by line from your ERP, stoppage log, and claims data to get one annual dollar figure, then shrink it by catching defects at the station with 100 percent camera-based computer vision inspection, instead of sampling.
Quality-related costs run nearly 20% of sales revenue for many manufacturers, and as high as 40% of total operations for some, according to ASQ. Very little of that money is spent on purpose.
You can probably quote your scrap rate from memory. But you might not be so sure about your cost of poor quality (COPQ). What did defects actually cost you last year, in dollars, across scrap, rework, downtime, claims, and the orders that never came back?
The majority of plants have never computed it. It's one of the most useful numbers in a quality leader's budget conversation, because it's usually the largest number no one person owns.
What Is COPQ?
COPQ stands for cost of poor quality: the total money an organization loses because work wasn't done right the first time. It's internal failure costs (defects you catch before shipping: scrap, rework, re-inspection, downtime from quality stops) plus external failure costs (defects that reach the customer: warranty claims, returns, recalls, expedited replacements, lost accounts).
It excludes the money you spend on purpose to prevent and find defects (prevention and appraisal costs), though those complete the full cost-of-quality picture.
Quality costs sort into four buckets:
- Prevention: process design, training, supplier qualification, maintenance before things drift.
- Appraisal: inspection, testing, audits. The cost of checking work.
- Internal failure: defects caught before they ship. Scrap, rework, re-inspection, downtime from quality stops.
- External failure: defects that reach a customer. Warranty claims, returns, recalls, expedited replacements, and lost accounts.
COPQ is buckets three and four. The old 1-10-100 rule of thumb still holds: a defect that costs $1 to catch at the station costs roughly $10 at end of line and $100 once it ships. The longer a defect survives, the more it eats.
How to Calculate COPQ
Let's look at a specific example of how COPQ can be calculated: a plant shipping $80 million a year of product. The plant has two shifts, a normal scrap rate; in short, nothing unusual. Its first COPQ pass looks like this:
| Cost line | Bucket | Annual cost |
|---|---|---|
| Scrap (1.8% of production value) | Internal failure | $1,440,000 |
| Rework labor and line time | Internal failure | $410,000 |
| Downtime from quality stops (120 hrs at $8,000/hr) | Internal failure | $960,000 |
| Held product awaiting disposition | Internal failure | $350,000 |
| Warranty claims and returns | External failure | $900,000 |
| Expedited freight and replacements | External failure | $180,000 |
| COPQ total | $4,240,000 | |
| Manual inspection labor (6 inspectors) | Appraisal | $420,000 |
That's $4.24 million of failure cost, 5.3 percent of output, at a plant doing well (so far). Add the $420,000 of inspection labor spent finding those defects and the full picture clears $4.6 million.
There are two things that make this table effective: First, every line traces to a system of record: scrap from ERP write-offs, downtime from the stoppage log, claims from finance. Second, it's conservative (it leaves out the costs nobody can pin down, like the checks that get skipped on night shift).
Copy the table, swap in your numbers, and you have a simple budget justification for inspection automation.
Why Defects Slip Through
Your inspectors are good at their jobs. So why is the external failure line the scariest one in the table?
Because a human inspector checks one unit in fifty, the line runs three shifts, and attention is a finite resource at the last hour of a shift. But your defects don't just show up when it's time to be sampled.
Computer vision can help. Cameras watch every unit, every shift, and a model such as RF-DETR flags the defect while the part is still at the station. Sampling becomes 100 percent inspection, and each bucket in the table moves:
- Internal failure: catching a defect at the station instead of end of line collapses the 1-10-100 ladder. You scrap a component instead of an assembly, and first pass yield climbs.
- External failure: escapes fall when inspection never blinks. Fewer claims, fewer returns, fewer hard phone calls.
- Appraisal: your inspectors stop staring at conveyors and move to disposition, root cause, and the judgment calls a camera can't make. The scrap you were tolerating becomes a defect trend you can act on.
How Roboflow Brings COPQ Down
Machine vision that watches every unit has been around for a while. Rules-based systems are tuned per part, per defect, per lighting condition, and the first process change sends you back to the integrator. This hurts your COPQ.
Roboflow's computer vision is different: models that learn your defects from your images. Upload examples of the flaws your line actually produces, label them with AI-assisted annotation, and fine-tune RF-DETR on your parts. Teams go from first image to a working model in an afternoon, not a quarter.
Three things make this the right tool for the COPQ problem specifically:
- It's one platform from first image to deployed inspection: annotate, train, and connect the model to your line with Workflows, so you're proving value on line one instead of stitching tools together.
- The model keeps getting better. Active learning routes the hard edge cases from production back into training, so every shift makes the system smarter, while a rules-based system gets more brittle with every process change.
- It runs where your line runs: on the cameras you already have, at the edge, or fully on-premise when your images can't leave the building.
And every catch is also a data point. Run 100 percent inspection for a month and you have a Pareto of your defect classes by line, shift, and station, which is exactly the evidence prevention spending needs. Detection pays for itself out of buckets three and four, then hands you the map to bucket one (the cheapest place to spend a quality dollar).
Manufacturers including over half of the Fortune 100 use Roboflow for exactly this kind of work: defect detection, part counting, assembly verification.
Take the Number to Finance
Quality leaders lose budget fights when they argue with adjectives. COPQ lets you argue with a number finance already recognizes, built from systems finance already trusts.
Bring three things: (1) the table, (2) the COPQ total, and (3) a scope small enough to prove fast. One line, one defect class, one quarter. Then let the results argue for line two.
Want to see what 100 percent inspection would look like on your line? Talk to our team, or start building with your own images today. As always, happy building!
Cite this Post
Use the following entry to cite this post in your research:
Erik Kokalj. (Jul 1, 2026). Cost of Poor Quality (COPQ): How to Calculate It and Bring It Down. Roboflow Blog: https://blog.roboflow.com/cost-of-poor-quality/