Updated July 2026 · 7 min read · Adente Vision Engineering Team
Picture a quality manager who wants to justify inspection without a vendor price in hand. The problem is that the cost of a missed defect stays invisible until it explodes: cheap on the line, ruinous in the field. Without a number for that, inspection reads as a cost rather than a saving. This post gives a defensible way to put a number on one escape, using only your own figures.
What does one missed defect actually cost?
The cost of a missed defect is almost never the value of the part itself; it is everything that follows because the flaw was not caught in time. A defective part that escapes inspection carries two kinds of cost. The direct cost is tangible: the scrap, the rework, the replacement part, the labour to tear a good assembly apart to reach the bad component.
What are the indirect costs an escape triggers?
The indirect cost is larger and harder to invoice, which is why it is the one that surprises a plant. A single field failure can pull in a line stoppage while the problem is traced, a returned shipment, warranty administration, replacement freight, sorting of suspect stock, penalties written into a supply contract, and the engineering hours to run a root-cause analysis. In regulated sectors, automotive, medical, aerospace, an escape can force a formal corrective action or a recall, whose cost dwarfs anything on the production floor.
Beyond the ledger sits reputation: a customer who receives a defect once inspects your shipments more closely, and a customer who receives two starts sourcing elsewhere. None of these appear in the price of the part, and all of them start with a single missed defect. A quality manager who counts only the scrapped part undercounts the real number by an order of magnitude or more.
How does the rule of ten multiply the cost of an escape?
The rule of ten, a long-standing rule of thumb in cost-of-quality literature, says the cost of a defect multiplies by roughly ten each time it moves undetected to the next stage. Catch a flaw at the inspection station and you pay to scrap or rework one part. Miss it there and catch it downstream in your own plant, and the cost is around ten times higher, because the defective part now sits inside a larger assembly that must be dismantled. Miss it again and ship it, and catching it in the field costs on the order of a hundred times the station cost.
How much does the cost climb at each stage?
| Where the defect is caught | Relative cost (rule of ten) | What drives the cost |
|---|---|---|
| At the inspection station | 1x | Scrap or rework of the single part, contained on the line |
| Downstream in your own plant | About 10x | Teardown and rework of a larger assembly, line stoppage |
| In the field, at the customer | About 100x | Returns, warranty, recall logistics, lost reputation |
These 1x, 10x and 100x figures are an approximate industry range from cost-of-quality literature, not an Adente measurement and not a price. The multipliers vary by industry, but the direction is not in dispute: the later an escape is found, the more it costs, and the cost climbs steeply. The escalation is often cited to the ASQ cost-of-quality framing of prevention, appraisal and failure costs.
How does a 0.69% false-negative rate change the math?
An escape rate is the frequency term in the cost equation, so lowering it lowers the expected cost of quality directly. Multiply the number of defective parts by the escape rate to get the escapes that reach the next stage, then multiply by the rule-of-ten cost of catching them there. Drop the escape rate and every downstream term shrinks with it.
On a live cap-inspection line, Adente Vision holds a 0.69% false-negative rate at 99.65% F1, inspecting each part in about 30 ms, which means roughly seven defects per thousand slip through rather than the far higher fraction a manual or sampled check would pass. You still put your own cost figures into the model; what changes is the frequency those figures multiply against. Adente Vision is an edge-AI visual inspection unit built by ADENTE Advanced Engineering Technologies, part of the Aden Group, sold through automation system integrators.
Why does 100% inspection beat sampling for escape cost?
Sampling inspects a fraction of parts, so by design it lets the defects in the unsampled remainder through at full rate. A sampling plan that checks one part in fifty can only catch defects in that one part; every defect in the other forty-nine escapes to the next stage at the rule-of-ten cost. Inspecting every part at line speed closes that gap.
The unit checks each part in about 30 ms and keeps pace with lines running 100+ parts per minute, so a defect is caught at the cheap 1x station stage instead of the expensive downstream one. That is the real economic argument for automated inspection, and it is why the payback model is built on your escape rate and the rule of ten, not on the price of the device. To size the two errors behind that rate, see escapes vs false rejects. For the wider method, read the pillar on AI visual inspection, and see the full range of tasks a single unit inspects this way across real applications.