Process
Validation: Proving Your Process Actually Works — Not Just That It
Worked Once
A Story That Started With an Epic
Failure
It was a Monday, six in the morning, when the production manager from
a factory that had just launched a new line for plastic automotive
components called me. “We have a problem,” said the voice on the other
end. “All three batches from the weekend production are unusable.
Dimensions are out of tolerance by a tenth of a millimeter. But when we
started the line on Friday, everything was perfect.”
The year was 2019, and this factory had just experienced a classic
scenario I’ve seen a hundred times: the process was approved based on a
single successful batch. One beautiful moment when everything lined up.
And then reality hit — variation that nobody anticipated because nobody
had systematically looked for it.
That day I learned a lesson I repeat to this day: One good
result is not validation. It’s a coincidence waiting to become a
problem.
What Is Process
Validation — and Why It’s Not Optional
Process Validation is the systematic evidence that your manufacturing
process, under proper control, can repeatedly and reliably produce
outputs that meet predefined specifications. It’s not a one-time test.
It’s not a signature on a protocol. It’s a comprehensive evidentiary
process that connects design, installation, operation, and performance
into one cohesive whole.
According to FDA, ICH Q8/Q9/Q10, and ISO 13485, Process Validation
is divided into three phases:
- IQ — Installation Qualification: Is it installed
correctly? - OQ — Operational Qualification: Does it work as
intended? - PQ — Performance Qualification: Does it work
reliably in real-world operation?
Three letters. Three questions. And between them lies the difference
between a factory that delivers with confidence and a factory that
struggles with every third batch.
Phase
1: Installation Qualification — Are You Building on the Right
Foundation?
IQ is your first line of defense. It answers one simple question:
Is the equipment installed exactly as designed?
In practice, this means confirming that:
- All components are delivered per specification
- Installation matches drawings and schematics
- Utilities (air, water, electricity, cooling) are connected
correctly - Calibrations of measurement systems are completed
- Software is installed in the correct version
- Documentation (manuals, certificates, approvals) is complete
I remember a project where during IQ we discovered that the main
pressure sensor had been installed in the reverse direction compared to
the schematic. The installers thought it looked “aesthetically better”
that way. The sensor worked — but it read values with the opposite sign.
Without IQ, this detail would have remained hidden until the process
started producing defective parts and nobody would know why.
IQ is not bureaucracy. It’s your last chance to catch
installation errors before they become manufacturing errors.
What Goes Into an IQ Protocol?
A good IQ protocol is specific and measurable. It doesn’t say “check
the pump.” It includes:
- Equipment identification — serial number, model,
revision - List of checkpoints — every component, every
connection - Acceptance criteria — exact values, ranges,
tolerances - Evidence — photographs, measurements,
certificates - Signatures and responsibilities — who inspected, who
approved
Phase
2: Operational Qualification — Does It Work, or Does It Just Look Like
It Does?
If IQ answers “is it built right?”, OQ answers “does it
work as designed?”
This is the phase where your equipment undergoes testing under defined
conditions. Not in real production — but in a controlled environment
where you can isolate variables and confirm that every function does
exactly what it should.
OQ verifies:
- Range of operating parameters (minimum, maximum, nominal
values) - Functional tests of all systems (heating, cooling, pressure,
speed) - Alarms and safety functions
- Reproducibility of settings
- Interactions between subsystems
- Software functions and limits
A Real Story: When OQ Saved
a Million
In one automotive project, during OQ we tested a molding process at
the upper and lower limits of mold temperature. At the nominal
temperature of 180°C, the parts were perfect. At 175°C — still fine.
But at 185°C, the surface finish degraded to an unacceptable level.
Without OQ, we would have launched the process with a temperature
range of 175–185°C because “it met the specification.” In reality,
every temperature spike above 183°C would have produced defective parts.
OQ allowed us to set operational limits at 175–182°C — and eliminate
the problem before it even existed.
OQ isn’t about whether it works. It’s about where it stops
working — and how close to that limit you can safely operate.
Worst-Case Testing
One of the most valuable approaches within OQ is worst-case
testing — testing at extreme combinations of parameters. You’re
not simulating average production. You’re testing conditions that could
happen, even if they’re improbable:
- Highest speed + lowest pressure
- Longest time + highest temperature
- Maximum load + minimum calibration
These combinations reveal weak points that nominal tests will never
show.
Phase 3:
Performance Qualification — The Real Test of Reality
If IQ confirms installation and OQ confirms functionality, PQ
confirms performance in real-world operation. This is
the moment of truth.
PQ proves that the process under normal manufacturing conditions —
with real operators, real materials, real variation — produces outputs
that meet specifications consistently and repeatedly.
The Standard Approach to PQ
Traditionally, PQ is performed as three consecutive batch
validation — three consecutive batches that must all meet
specifications. But that’s the minimum, not the maximum.
A proper PQ includes:
- Definition of the manufacturing scenario — who
operates, what material, what conditions - Statistical sampling plan — not “measure 30 pieces,”
but calculate how many pieces you need for 95% confidence - Measurement of critical quality attributes
(CQAs) - Variability analysis — Cp, Cpk, Pp, Ppk
- Trend analysis — not just the average, but stability
over time - Deviation documentation — what happened, why, and
how it was resolved
The Modern
Approach: Continuous Process Verification
The modern approach, supported by FDA and ICH, goes beyond three
batches. Continuous Process Verification (CPV) is a
philosophy that says: validation doesn’t end with a protocol. Validation
is a continuous process.
CPV means that:
- You monitor critical parameters in real time
- Statistical Process Control (SPC) is integrated
- Deviations are analyzed and documented on an ongoing basis
- Periodic revalidations are planned
- Production data continuously confirms (or challenges) the validated
state
This approach is especially important in high-risk industries —
pharmaceuticals, medical devices, aerospace — where “three good batches”
are not sufficient evidence of long-term stability.
The Common Language
IQ/OQ/PQ — How It All Connects
It’s important to understand that these three phases are not isolated
steps. They form an interconnected system where each phase builds on the
previous one:
| Aspect | IQ | OQ | PQ |
|---|---|---|---|
| Question | Is it right? | Does it work? | Is it reliable? |
| Focus | Installation | Function | Performance |
| Environment | Static | Controlled | Real-world |
| Result | Confirmed state | Confirmed range | Confirmed capability |
You can’t do OQ without IQ. You can’t do PQ without OQ. And you can’t
consider a process validated without all three.
The Most Common Mistakes I
See
In 25 years in quality, I’ve seen it all. Here are the mistakes that
come up most often:
1. Validation as a Paper
Exercise
The biggest sin: a protocol written so it can’t fail. Acceptance
criteria set so wide that even a catastrophic process meets them.
Validation is not a checkbox — it’s an evidentiary process. If your
protocol can’t fail, it can’t prove anything.
2. Ignoring Variability
You measure 30 pieces from one batch, under the same conditions, with
the same operator. And you declare the process “validated.” But what
happens when you change the material batch? When the night shift is
running? When the shop floor is 5°C warmer? Variability is reality —
and your validation has to capture it.
3. Forgetting the Human Factor
A process doesn’t run itself. It’s operated by people — and people
are the biggest source of variability in any process. PQ must include
different operators, different shifts, different times of day. Otherwise
you’re validating a process that only works with John on the morning
shift.
4. Missing a Revalidation
Strategy
Validation is not a one-time event. Processes change — materials,
equipment, operators, environment. If you haven’t defined when and how
you revalidate, your original validation loses its validity and you
won’t even know it.
5. Blending IQ/OQ/PQ
Yes, phases can overlap. But they can’t replace each other. Each
phase has its purpose, its acceptance criteria, its documentation. When
you mix them into one protocol, you lose clarity and — worse — you lose
the opportunity to stop the process at the point where you need to.
How to Build a Process
Validation Strategy
Good validation doesn’t start with a protocol. It starts with a
strategy.
Step 1: Process Understanding
Before you validate anything, you need to understand the process.
What are the critical parameters? What are the critical quality
attributes? What is the natural variability? Without this knowledge,
you’re validating blindly.
Use tools like: – Process Flow Diagram — map the
process – FMEA — identify risks – Cause
& Effect Matrix — link parameters to outputs –
DOE (Design of Experiments) — systematically test
influences
Step 2: Master Validation
Plan
Create a document that connects all activities: – What is being
validated and why – What the acceptance criteria are – What the sampling
plan is – Who is responsible for what – What the dependencies between
activities are – What the critical milestones are
Step 3: Execution with
Discipline
Execute each phase with discipline. Don’t change the protocol in the
middle of a run. Don’t ignore deviations. Don’t document “after the
fact.” Every deviation is information — and information is your most
valuable currency.
Step 4: Analysis and
Conclusion
Analyze the data statistically. Not just “is it within tolerance.”
Look at Cp, Cpk. Check trends. Verify stability. And then — and only
then — draw a conclusion.
Step 5: Ongoing Monitoring
Validation continues. Define how you’ll monitor the process, when
you’ll revalidate, and what signals will alert you that the process is
drifting from the validated state.
ROI of Process Validation
I’ve heard the argument: “Validation costs money. Time. People. It
delays the launch.”
My response is always the same: How much does a customer
complaint cost you? How much does a recall cost? How much does losing a
customer cost?
In that 2019 project, where three batches failed after a “successful”
launch — the real cost was €47,000 in scrap, 120 hours of analysis, two
weeks of delivery delays, and a damaged relationship with the customer.
A complete Process Validation would have cost approximately €15,000 and
three weeks.
Investment in validation is not a cost. It’s insurance. And
the best insurance you can buy.
Conclusion: Validation Is an
Act of Trust
When I sign a validation protocol, I’m not just putting down my name.
I’m putting my professional confidence that this process — under these
conditions — with these people — will do what it’s supposed to. Every
day. Every batch. Every piece.
That’s not a small thing. It’s a responsibility that deserves to be
taken seriously.
Process Validation isn’t about paperwork. It’s not about audits. It’s
not about regulations. It’s about confidence. Confidence
that when your customer opens the package, they get what they expect.
Confidence that when your operator presses the “START” button, the
process does what it should. Confidence that your factory stands on a
solid foundation — not on luck.
Three letters. Three phases. One result: trust.
And trust, unlike luck, can be systematically built, measured, and
proven. That’s exactly what Process Validation does.
Peter Stasko is a Quality Architect with 25+ years
of experience in automotive, manufacturing, and continuous improvement.
He helps factories build systems that don’t work by accident — but by
design.
