Key takeaways
Short answer: AQL sampling inspects a statistically determined sample from a lot and accepts or rejects the whole lot based on the number of defects found. 100% inspection checks every unit. It feels safer to check everything, but manual 100% inspection is surprisingly unreliable (inspectors miss defects), slow and expensive. The right answer depends on defect cost, volume, and whether the process is actually in control. See also quality by design vs quality by inspection.
AQL sampling uses statistics to inspect a defined sample size from a lot and accept or reject the entire lot based on how many defects appear against an acceptance number. It trades certainty about every unit for efficiency, and it works because, for a process in control, the sample reliably represents the lot.
100% inspection checks every single unit. It feels like the safe choice — surely checking everything catches everything — but that intuition is misleading for manual inspection, which is slow, expensive, and statistically known to miss a meaningful share of defects through fatigue and monotony.
A plant ships a critical part and decides "we cannot risk it — inspect 100%." An inspector checks every unit by eye, but studies of repetitive visual inspection suggest humans catch only around 80% of defects when checking everything, so one in five escapes anyway — and the line slows to the inspector's pace. An automated inline gauge checking 100% would be reliable, but a manual 100% check is both expensive and leaky. For a stable process, AQL sampling plus fixing the root cause would catch more, cheaper. The 100% manual check gave false confidence at high cost.
Manual 100% inspection suffers from fatigue, monotony and the expectation that most parts are good, so inspectors miss defects — often 10-20%. It also slows throughput and adds cost. Automated 100% inspection avoids the human reliability problem, but manual 100% checking is frequently both the most expensive and least reliable option.
Reaching for 100% inspection is often a symptom: the process is producing defects and inspection is the band-aid. The durable answer is quality-at-source — mistake-proofing and process control that stop the defect being made, so neither sampling nor 100% checking has much to catch. Inspection sorts; it does not improve.
1. Choosing manual 100% for "safety." It is slow, costly and still misses defects.
2. AQL sampling an out-of-control process. Sampling assumes the lot is homogeneous; an unstable process violates that.
3. Inspecting instead of fixing. More checking does not improve a defect-prone process.
4. Ignoring inspector reliability. Human 100% checks are far from 100% effective.
Inspection strategy interacts with the OEE Quality rate and throughput. Heavy manual inspection slows the line (Performance) while still letting defects through (Quality). Moving from inspection to quality-at-source cuts defects and removes the inspection bottleneck — improving both terms at once.
See how Fabrico captures this automatically on your lines — explore OEE for manufacturing or book a demo.
No — manual 100% inspection is slow, costly and still misses 10-20% of defects through fatigue.
For stable, in-control, high-volume processes where the sample reliably represents the lot.
Fatigue, monotony and the expectation that parts are good cause inspectors to overlook defects.
Quality-at-source — mistake-proofing and process control that stop the defect being made.
No — it sorts good from bad; only changing the process improves quality.
Login
Blog
