Quality control is the set of activities that ensure output meets a defined standard — detection, measurement, and correction of deviation. It can happen at the end (inspection) or inside the process (in-process control). End-of-line inspection catches defects but doesn't prevent them; in-process control reduces defect generation and shortens the feedback loop. The aim is to align what is produced with what is required, at minimum cost of failure and rework.
Quality has two dimensions: conformance (did we build what we specified?) and fitness for use (does it serve the user?). Conformance is easier to measure; fitness for use can require judgment and user feedback. Many organisations optimise for conformance and miss fitness for use — the product passes checks but doesn't solve the problem. The strategic question is: what standard are we controlling to, and does that standard match the outcome we care about?
Quality control interacts with cost and speed. Tighter control usually costs more and can slow throughput; looser control reduces cost but raises the risk of defects reaching the customer. The balance depends on the cost of failure. In safety-critical domains, control is strict. In fast-moving experiments, "good enough" and rapid iteration may dominate. The discipline is making the trade-off explicit: what is the cost of a defect, and how much are we willing to spend to prevent or catch it?
Section 2
How to See It
Quality control appears wherever there is a defined standard and a mechanism to check or correct against it. Look for: specs, checks, measurements, and corrective action when deviation is found. When defects are caught before they reach the customer or the next stage, quality control is at work.
Business
You're seeing Quality Control when a manufacturer samples from each batch and tests against spec. If the sample fails, the batch is held or scrapped. If it passes, the batch is released. The control loop: produce → sample → measure → accept or reject → correct if needed. The same logic applies to service: define the standard, measure delivery against it, correct when it misses.
Technology
You're seeing Quality Control when code goes through CI/CD with tests, lint, and review before merge. The standard is "passes tests and review"; the control is automated and human. Bugs that slip through are feedback to improve the tests or the process. Quality control here is in-process — preventing bad code from reaching production — not only inspection after the fact.
Investing
You're seeing Quality Control when a fund has checklists and veto rights for investments. The standard might be hurdle return, risk limits, or compliance. Deals that don't meet the standard are rejected or escalated. The control is the process that prevents bad bets from getting into the portfolio.
Markets
You're seeing Quality Control when an exchange or clearinghouse marks positions to market and enforces margin. The standard is solvency and risk limits; the control is continuous measurement and action when thresholds are breached. Quality control here protects the system from bad positions accumulating.
Section 3
How to Use It
Decision filter
"Define the standard that matters — conformance and fitness for use. Put checks where defects are cheapest to catch. Prefer in-process control over end-of-line inspection when you can. Match the rigor of control to the cost of failure."
As a founder
Control quality at the bottleneck of customer trust and cost of failure. For product: what does "good" mean, and where do we check? Ship fast, but don't ship broken — define the minimum bar and enforce it (tests, review, staging). For service: define the standard (e.g. response time, resolution), measure it, and correct when it slips. The mistake: no explicit standard, so "quality" is vague. The second mistake: controlling to the wrong standard (e.g. conformance to spec when the spec doesn't match user need).
As an investor
Assess how the company defines and controls quality. Is there a clear standard? Is it measured? Is correction systematic or ad hoc? Companies that treat quality as "we try hard" without measurement and feedback carry operational and reputational risk. Look for in-process control, not only after-the-fact inspection.
As a decision-maker
Use quality control when the cost of defects is material. Define the standard, choose where to check (early vs late), and decide how much to invest in prevention vs detection. In low-stakes or exploratory work, lighter control may be right; in high-stakes or regulated domains, invest in the loop.
Common misapplication: Controlling to a spec that doesn't match user value. You get conformance and unhappy users. Align the standard with fitness for use; update the spec when you learn what "good" really is.
Second misapplication: Inspecting at the end when catching defects earlier would be cheaper. Push control upstream where possible; shorten the feedback loop so defects are caught and corrected before they compound.
Jiro runs quality control through relentless in-process standards: ingredient selection, preparation steps, and presentation. There is no "inspection at the end" — each step is done to a standard, and the next step doesn't begin until the previous one is right. The control is built into the process and the discipline of the team. The result is consistent output and few defects because deviation is corrected at the source.
Ed CatmullCo-founder, Pixar; President, Pixar & Disney Animation
Catmull embedded quality control in Pixar's creative process through candid feedback and iterative review. The standard was "does it serve the story and the audience?" — measured by internal review and later by audience response. The control was in-process: reels, Braintrust meetings, and rewrites rather than a single final inspection. Quality was maintained by making the feedback loop fast and honest, not by inspecting at the end.
Section 6
Visual Explanation
Quality control: define standard, measure output, compare, correct when deviation exceeds tolerance. In-process control catches defects early; end-of-line inspection catches them late.
Section 7
Connected Models
Quality control connects to feedback, process, and failure prevention. The models below either enable it (Feedback Loops, Standard Operating Procedure), extend it (Fail-safes), or warn about bad inputs (Garbage In Garbage Out).
Reinforces
[Feedback](/mental-models/feedback) Loops
Quality control is a feedback loop: measure output, compare to standard, correct. The tighter and faster the loop, the better the control. Feedback Loops describe the structure; quality control is the application to conformance and fitness for use.
Leads-to
[Fail-safes](/mental-models/fail-safes)
Fail-safes are controls that trigger when something goes wrong. Quality control can include fail-safes: automatic rejection, hold, or escalation when measurement exceeds tolerance. The two work together: control maintains quality; fail-safes catch what control misses.
Reinforces
Standard Operating Procedure
SOPs define how work is done — and thus what "in control" looks like. Quality control measures against the standard that SOPs embody. Without a clear standard (explicit or in procedure), control has nothing to compare against.
Tension
[Garbage In Garbage Out](/mental-models/garbage-in-garbage-out)
GIGO: bad inputs produce bad outputs. Quality control on output can catch bad results, but if the inputs (specs, data, assumptions) are wrong, you can conform to a bad standard. Control the inputs as well as the process and output.
Section 8
One Key Quote
"You can't inspect in quality. Quality must be built in."
— W. Edwards Deming
Inspection catches defects; it doesn't prevent them. The leverage is in the process: design and control so that defects are rare and caught early. Measurement and correction are part of building quality in, not a substitute for it.
Section 9
Analyst's Take
Faster Than Normal — Editorial View
Define the standard first. Many teams argue about "quality" without agreeing what good looks like. Is it conformance to spec? User satisfaction? Uptime? Define it, make it measurable where possible, and then design control around it. Without a standard, quality control is theatre.
Control where it's cheapest to catch failure. Catching a defect at the source is cheaper than catching it at the end; catching it before shipment is cheaper than after. Push measurement and correction upstream. That doesn't mean infinite checks — it means placing the loop where the cost of failure and the cost of control are balanced.
Fitness for use beats conformance when they conflict. A product that passes every test but doesn't solve the user's problem has failed. Align the standard with the outcome you care about. When specs drift from reality, update the spec or you're controlling to the wrong target.
Section 10
Summary
Quality control is ensuring output meets a defined standard through measurement and correction. Define the standard (conformance and fitness for use), measure, and act when deviation exceeds tolerance. Prefer in-process control to end-of-line inspection. Match the rigor of control to the cost of failure. Quality must be built in; inspection alone is not enough.
Quality in the sense of fitness for use: build-measure-learn as a control loop for product-market fit. The standard is user value.
Reinforces
[Theory of Constraints](/mental-models/theory-of-constraints)
The constraint limits throughput. Quality failures at the constraint waste capacity. Controlling quality at and around the constraint protects throughput. The two align: improve the constraint, and control quality so the constraint's output is good.
Repeatable systems produce consistent output. Quality control is the mechanism: define the standard, measure, correct. The more repeatable the process and the clearer the standard, the more effective control becomes.
