What it means
Quality control sits at the end of a process and asks one question: does this meet specification? It covers inspection, sampling, testing and measurement, and its output is a decision to accept, rework or reject.
It is not the same thing as quality assurance, although the two are constantly confused. Quality assurance designs the process so that defects are unlikely, while quality control checks the output to see whether the process actually worked.
Checking every unit is rarely economic, so most operations sample. Statistical sampling takes a defined number of units from each batch and uses the defect rate in that sample to judge the whole batch, accepting a small and quantified risk of getting it wrong.
The financial case rests on where a defect is caught. A fault found on the production line might cost a few dollars to put right, while the same fault found by a customer costs a return, a replacement, a support call and sometimes the relationship itself.
Service businesses run quality control too, just with different tools. Call monitoring, file reviews in professional firms, mystery shopper visits and code review in software all serve the same purpose as a gauge on a factory line.
The records it produces have value of their own. Inspection logs, batch results and rejection reports show which machines, shifts, materials or suppliers generate the most faults, which is exactly the evidence an improvement team needs to fix the cause rather than the symptom.
In practice
Real-world examples
Example 1
A bottling plant weighs one bottle in every 200 coming off the line. When three consecutive samples fall below the labelled volume, the line is stopped and the filler recalibrated, which prevents a batch of 40,000 underfilled bottles from reaching retailers and triggering a recall.
Example 2
An accounting firm reviews one in every ten completed tax returns before filing. The review catches a misapplied allowance on a client return that would have produced a $9,000 assessment plus interest, along with a difficult conversation about professional competence.
Example 3
A software team requires every change to pass automated tests and a second engineer's review before release. In one quarter the process catches 84 defects before deployment, while the 6 that reach production take an average of 5 hours each to fix, against 40 minutes for those caught at review.
Think of it
Quality control is like checking your ingredients and following a recipe step by step before serving a meal to guests, ensuring it tastes great and is safe to eat.
Formula
Calculation
Defect Rate = (Defective Units / Total Units Produced) x 100, and Cost of Poor Quality = Rework Cost + Scrap Cost + Warranty and Return Cost. A components manufacturer produces 25,000 units in a month and inspection identifies 400 defective units, so the defect rate is (400 / 25,000) x 100 = 1.6% and first-pass yield is 98.4%. Of those 400 defects, 300 can be reworked at $18 each, costing 300 x $18 = $5,400, while 100 must be scrapped at a unit cost of $46, costing 100 x $46 = $4,600. Warranty claims on units that escaped inspection in earlier months add $12,000, so the cost of poor quality is $5,400 + $4,600 + $12,000 = $22,000. Against monthly revenue of $2,750,000 that is $22,000 / $2,750,000 = 0.8%.Case study
Seen in the real world
Brightfold Packaging, an illustrative fictional manufacturer of printed cartons, was losing customers to complaints about colour variation. Its inspection process checked only the first and last carton in each run of 5,000, so drift in the middle of a run went undetected until the customer opened the pallet.
The operations team introduced a check every 500 cartons together with a printed colour reference card at each press. Customer-reported defects fell from 2.4% of cartons shipped to 0.5%, and on annual shipments of 3,000,000 cartons that was the difference between 72,000 and 15,000 defective cartons, or 57,000 fewer reaching customers.
The added inspection cost roughly $60,000 a year in labour, while credit notes and reprints fell by $310,000. In this fictional example the return was more than five times the cost, and the sales team recovered two accounts that had been lost the previous year.
Watch out
Common mistakes
- Using quality control and quality assurance as synonyms, when control inspects the output and assurance builds the process that produces it.
- Trying to inspect quality into a product, since an unstable process simply produces more defects and more inspection finds them without reducing how many are made.
- Measuring only the units rejected, when the real cost includes rework time, scrapped material, delayed shipments, expedited freight and customer returns.
Questions
People also ask
What is an acceptable defect rate?
It depends entirely on the product, since a few per cent may be tolerable in low-value packaging while medical or aerospace parts are measured in defects per million.
Does quality control apply to services?
Yes, through call monitoring, file review, mystery shopping and peer review, all of which check delivered work against a defined standard.
Is 100% inspection always better?
Not usually, because it is expensive, it slows output, and on repetitive manual checks inspector fatigue means a share of defects still slips through.