Quality
and the Law of Diminishing Returns: When Your Organization Keeps
Polishing a Process Past the Point of Value — and the Perfection Nobody
Needed Became the Investment Nobody Recovered
The engineering manager stood in front of the executive committee
with a slide that should have made everyone proud. Defect rate: 0.03%.
Process capability index: 2.1. Customer complaints: down 94% from three
years ago. Gauge R&R: under 5%. Every single metric was world-class.
Every single metric had been world-class for over a year.
The CFO leaned forward. “So why are we asking for another eight
hundred thousand dollars in process improvement funding?”
The engineering manager blinked. “To get to 0.01%.”
The room went quiet. Not the impressed kind of quiet. The kind of
quiet that happens when someone realizes they’ve been watching a team
sprint on a treadmill — enormous effort, impressive speed, zero forward
motion.
That moment — the moment when an organization realizes it has been
optimizing something that was already good enough while a dozen other
processes quietly decayed in the background — is the moment the Law of
Diminishing Returns stops being an economics textbook concept and starts
being your quality strategy’s most expensive blind spot.
What the Law
of Diminishing Returns Actually Means
The concept is straightforward. In any system, there comes a point
where each additional unit of input produces less output than the
previous unit. The first 80% of improvement takes 20% of the effort. The
last 20% of improvement takes 80% of the effort. And the final 1% — the
perfection that your best engineers chase with religious intensity — can
consume more resources than everything that came before it combined.
In manufacturing quality, this law operates with ruthless precision.
Reducing defects from 10% to 1% is hard but straightforward. You
identify the biggest sources of variation, you implement controls, you
train operators, you fix the obvious problems. Each intervention
produces visible, measurable results. The feedback loop is
satisfying.
Reducing defects from 1% to 0.1% is harder. You need better
measurement systems, more sophisticated statistical methods, tighter
process control. But the returns are still tangible. Customers notice.
Warranty costs drop. Your reputation improves.
Reducing defects from 0.1% to 0.01%? That is where the curve bends
sharply. The investments required — additional inspection layers, more
expensive equipment, tighter tolerances, more frequent calibration
cycles — grow exponentially while the actual benefit shrinks to
something that may not even be statistically distinguishable from random
variation.
And reducing defects from 0.03% to 0.01%? That is the territory where
organizations don’t just encounter diminishing returns. They encounter
negative returns — because while they were obsessing over that last
fraction of a percent, three other lines drifted out of spec, two
suppliers changed their material without telling anyone, and a new
competitor entered the market with a product that was merely good enough
but cost half as much.
The Perfection Trap
I watched this unfold at a Tier 1 automotive supplier in Slovakia.
The plant produced precision-machined housing units for transmission
systems. Their quality system was, by any reasonable standard,
exceptional. They had achieved zero PPM on their primary customer
delivery metric for eleven consecutive months. Their process was stable,
capable, and well-controlled.
But their continuous improvement team — a group of six talented
engineers — had been working on the same process for two and a half
years. Their mission was to achieve a Cpk of 3.0, which for those
unfamiliar with statistical process control is the equivalent of a
gymnast trying to add a half-point to a perfect score. It requires
eliminating variation that is already so small it barely registers on
the measurement system.
The cost of this pursuit was staggering. Custom tooling modifications
every six weeks. Premium-grade raw material specifications that added
12% to material costs. Measurement equipment that cost more than the
machines it was measuring. And six brilliant engineers who could have
been solving problems on three other production lines that were quietly
running at Cpk 0.9 — below the minimum acceptable threshold.
When I asked the plant manager why they were pursuing Cpk 3.0, his
answer was revealing: “Because we can.”
That is the perfection trap. The pursuit of excellence becomes
disconnected from the pursuit of value. The metric becomes the mission.
The score becomes the game. And an organization that should be deploying
its improvement capacity where it matters most instead channels it into
the place where it matters least.
Where Diminishing Returns
Hide
The Law of Diminishing Returns doesn’t announce itself. It doesn’t
show up on a dashboard with a warning label. It infiltrates quality
organizations through several predictable channels:
Over-instrumentation. A measurement system that
provides enough resolution to control a process is valuable. A
measurement system that provides ten times the necessary resolution is
expensive. One that provides a hundred times the resolution is wasteful.
I’ve seen organizations spend six figures on coordinate measuring
machines to inspect features where a go/no-go gauge would have provided
all the information needed to control the process.
Over-documentation. Quality management systems
require documentation. But there is a point where additional procedures,
work instructions, and forms don’t improve execution — they simply
create compliance burden. Operators who once internalized the correct
method now follow a forty-page work instruction by rote, understanding
less about what they’re doing than they did when the instruction was two
pages. The documentation was supposed to capture knowledge. Instead, it
replaced it.
Over-training. Training is essential. But training
the same people on the same topics to ever-higher levels of theoretical
detail while neglecting the practical skills they actually need is
investment without return. I once observed a week-long Six Sigma Black
Belt refresher course attended by engineers who hadn’t run a designed
experiment in two years because their daily work didn’t require one.
Over-auditing. Internal audits are a cornerstone of
any quality system. But when a process is audited four times a year and
the last three audits found zero nonconformities, the fourth audit is
not providing new assurance. It is consuming resources that could audit
a process that hasn’t been examined in eighteen months and is overdue
for attention.
Over-specification. Engineering specifications
should reflect what the product actually needs. When tolerances are
tightened beyond functional requirements “because we can hold them” or
“because it looks better on the drawing,” the organization pays for
capability it doesn’t need. Every tighter tolerance demands better
tooling, more frequent tool changes, slower cycle times, and more
inspection — all to satisfy a requirement that was arbitrary in the
first place.
The Opportunity Cost Nobody
Counts
Here is what makes the Law of Diminishing Returns so dangerous in
quality management: the cost isn’t just what you spend. It’s what you
don’t do.
Every engineer hour spent driving Cpk from 2.1 to 2.3 on a stable
process is an engineer hour not spent on the unstable process running at
Cpk 0.8 that just shipped a nonconforming lot. Every dollar spent on
additional inspection of a well-controlled characteristic is a dollar
not spent on preventing the defect that your customer will find next
month. Every management review minute devoted to polishing a gold-star
metric is a minute not devoted to the early warning signs that a
critical supplier is in financial distress and about to cut corners.
Opportunity cost is invisible on financial statements. It doesn’t
have a line item. It doesn’t generate a variance report. But in quality
organizations, it is often the single largest cost — the improvement
that never happened because the resources that should have driven it
were busy perfecting something that was already good enough.
I worked with a pharmaceutical manufacturer that had spent three
years and considerable investment optimizing the fill weight accuracy of
a liquid medication. The process was already operating at a level that
exceeded regulatory requirements by a comfortable margin. But the
validation team kept finding smaller and smaller sources of variation to
eliminate.
Meanwhile, their packaging line — the operation that actually touched
the finished product in its final form — had a label accuracy rate of
98.2%. That sounds acceptable until you realize that at their volume, it
meant thousands of incorrectly labeled packages per month. A recall
triggered by a mislabeled product would cost exponentially more than
everything they had ever spent on fill weight optimization. But the
packaging line wasn’t glamorous. It wasn’t high-tech. It wasn’t the kind
of project that looked impressive in a quarterly review.
The validation team’s talent was real. Their technical skill was
genuine. But their deployment was catastrophically misaligned with
organizational risk.
Recognizing the Inflection
Point
The most important skill in quality management isn’t knowing how to
improve something. It’s knowing when to stop improving one thing and
start improving something else.
Here are the signals that you’ve passed the inflection point where
returns begin to diminish:
Your improvement rate is decelerating. If your last
three improvement projects on the same process each delivered less
benefit than the one before, and the cost of each project was equal or
greater, you are on the diminishing part of the curve. The data is
telling you something. Listen to it.
Your remaining variation is dominated by common
causes. When your process is in statistical control and the
remaining variation is inherent to the system — material lot variation,
ambient temperature fluctuation, normal machine wear — further reduction
requires fundamentally changing the system, not tweaking it. That might
be worth doing. But it’s a different kind of investment than incremental
improvement, and it should be evaluated as such.
Your customer hasn’t noticed the last three
improvements. If your defect rate dropped from 0.05% to 0.03%
and your customer didn’t change their behavior, their orders, or their
satisfaction rating, the improvement may exist statistically but not
practically. You optimized a number. You didn’t optimize value.
Your team is creatively bored. The engineers working
on the process are capable people. If they’re going through the motions
— running the same types of analyses, proposing the same types of
countermeasures, generating the same types of reports — it’s because the
well is dry. Not because they lack skill. Because the problem no longer
warrants their skill.
Other processes are deteriorating. This is the most
critical signal and the one most commonly missed. If your overall plant
quality metrics are holding steady but your best process keeps getting
better while your average ones keep getting worse, you are not
improving. You are redistributing quality — robbing Peter to pay Paul,
if you’ll forgive the expression. The aggregate looks fine. The reality
is fragile.
The
Strategic Alternative: Marginal Analysis for Quality
The solution isn’t to stop improving. It’s to improve where the
returns are highest.
This requires a fundamentally different mental model. Instead of
asking “How can we make this process better?” the question becomes
“Where would the next dollar of improvement investment produce the
greatest benefit?”
This is marginal analysis applied to quality. And it changes
everything.
Instead of a continuous improvement team that adopts a pet process
and polishes it for years, you get a team that operates like an
investment portfolio manager. They assess the entire landscape of
quality opportunities. They rank them by potential impact relative to
required investment. They deploy resources to the highest-return
opportunities first. And when the returns on a particular process
diminish below the returns available elsewhere, they move.
This is not abandoning improvement. It is intelligent improvement. It
is the difference between a farmer who keeps fertilizing the same field
long past the point of diminishing yield and one who recognizes when
it’s time to rotate crops.
The practical implementation looks like this:
Maintain a quality opportunity register. Not just a
list of nonconformances. A comprehensive, regularly updated assessment
of every process, every supplier, every customer touchpoint, ranked by
risk and potential improvement benefit. This becomes your investment
menu.
Quantify the return curve for major improvement
initiatives. Before committing resources, estimate the expected
benefit and cost. Track actual against estimate. When the actual return
drops below the return available from the next best alternative,
transition resources.
Set explicit “good enough” thresholds. This is the
hardest part culturally. It requires leadership to say, publicly, “A Cpk
of 1.67 on this process is sufficient. We are not pursuing 2.0. We are
deploying those resources to the line running at 1.1.” This feels like
settling. It is actually optimizing.
Review resource allocation quarterly. Not just
project status. Resource allocation. Where are our best people spending
their time? Is that where the biggest quality risks are? If not, why
not? This conversation is more valuable than any individual improvement
project.
The Psychology of
Over-Optimization
Understanding the economics of diminishing returns is necessary but
not sufficient. You also have to understand why organizations fall into
the trap — because it is not primarily an economic failure. It is a
psychological one.
The first driver is the sunk cost commitment. Once an organization
has invested years in optimizing a process, stopping feels like wasting
that investment. The team has developed deep expertise in this
particular process. They’ve built relationships with the equipment
suppliers. They’ve published papers. Admitting that further investment
isn’t warranted feels like admitting the previous investment was wrong.
It wasn’t. The previous investment was valuable. It’s the next
investment that isn’t.
The second driver is skill comfort. Engineers are problem solvers.
When they’ve been working on a process long enough to understand it
deeply, solving problems within it becomes comfortable and efficient.
Moving to a new, unfamiliar process means starting over — building new
understanding, developing new relationships, confronting new
uncertainty. It’s harder work. It’s less pleasant work. And it’s exactly
the work that produces the highest returns.
The third driver is visibility. Improving a process from Cpk 1.0 to
1.5 is dramatic and impressive. Improving it from 2.0 to 2.1 is
incremental and barely noticeable. But fixing a process from 0.8 to 1.3
is transformative. Organizations gravitate toward what looks impressive
on a slide rather than what creates the most value. The marginal
analysis approach requires rewarding impact over optics.
The fourth driver is inertia. Continuous improvement programs develop
momentum. They develop cadence. Monthly kaizen events become scheduled
regardless of whether there’s a high-value target available. Quarterly
improvement reports demand content regardless of whether the
improvements are meaningful. The system sustains itself long after the
returns have diminished.
A Framework for Smarter
Investment
Here is a practical framework for applying marginal analysis to your
quality improvement portfolio:
Step 1: Map your quality landscape. For every
significant process, determine where it sits on the return curve. Is it
in the steep improvement zone where investment produces high returns? Is
it in the flat zone where additional investment produces minimal
additional benefit? Or is it in the declining zone where further
investment actually creates negative value through complexity,
over-control, and opportunity cost?
Step 2: Establish your opportunity cost. What is the
return available on the best unaddressed quality opportunity in your
organization? That number — that potential benefit per unit of
investment — becomes your hurdle rate. Any improvement initiative that
can’t clear that hurdle doesn’t get resources. Full stop.
Step 3: Time-box optimization efforts. Before
starting any improvement project, define the expected benefit, the
expected cost, and the point at which you’ll reassess whether continued
investment is warranted. This isn’t pessimism. It’s discipline.
Step 4: Rotate talent. Your best quality engineers
should be working on your biggest quality challenges, not your smallest.
Establish regular rotation cycles that move expertise from the optimized
to the critical.
Step 5: Celebrate stopping. This is the cultural
lever. When a team leader recommends redeploying resources from a
well-optimized process to a struggling one, celebrate that decision as
loudly as you’d celebrate a zero-defect month. It is, in fact, the more
mature and valuable decision.
The Deeper Lesson
The Law of Diminishing Returns in quality management is ultimately
about a fundamental truth that most organizations resist: perfection is
not the same as excellence. Excellence is knowing what good enough looks
like and having the discipline to redirect your energy toward the places
where it’s not good enough yet.
The organizations that master this — that build the analytical
capability to recognize the inflection point and the cultural courage to
act on it — don’t just improve faster. They improve smarter. They get
more benefit from less investment. They build systems that are resilient
not because every process is perfect, but because improvement resources
are always flowing toward the point of greatest need.
The engineering manager from our opening story? She eventually got
her eight hundred thousand dollars. Not for the process that was already
at 0.03%. She got it for three other production lines that, between
them, were generating more customer complaints, more warranty costs, and
more internal rework in a single month than the optimized process had
produced in three years. Within six months, overall plant quality
improved more than it had in the previous two years of polishing.
The Law of Diminishing Returns didn’t go away. She just stopped
fighting it and started working with it.
That is the most powerful quality strategy there is: not the
relentless pursuit of perfection everywhere, but the intelligent
deployment of improvement capacity where it actually matters. Not doing
more with less, but doing the right things with what you have.
Your best process doesn’t need your best people. Your worst one
does.
Peter Stasko is a Quality Architect with 25+ years of experience
transforming organizations across automotive, aerospace, and
pharmaceutical industries.
