Quality Function Deployment: When the Customer’s Whisper Becomes Your Engineering Blueprint — and the House of Quality Translates Desire Into Specification

Peter Stasko

You have probably been here before. The marketing team hands engineering a document titled “Customer Requirements.” It is three pages long, full of phrases like “easy to use,” “feels premium,” “reliable,” and “doesn’t break.” Engineering stares at it the way a chef stares at a recipe that says “make it taste good.” These are not specifications. These are feelings.

And yet, somewhere between that vague wish list and the final product, someone has to make a decision. How thick should the material be? What tolerance is acceptable? Which surface finish says “premium” to a human fingertip? How many cycles before “reliable” becomes a lie?

Most organizations bridge this gap with guesswork, politics, or the loudest voice in the room. Quality Function Deployment bridges it with structure. And at the center of QFD sits the House of Quality — a matrix so elegant that it has been turning customer whispers into engineering language since the 1960s.

Let me show you how it works. Not in theory. In practice.

The Origin: Mitsuno and the Voice That Needed Translation

QFD was born at Mitsubishi’s Kobe shipyard in the late 1960s. The challenge was familiar: naval designers had to interpret vague customer desires — “safe,” “comfortable,” “fast” — into thousands of precise engineering decisions about hull geometry, material grades, and weld specifications.

Dr. Yoji Akao formalized the method, and Toyota adopted it in the 1970s. The results were striking. Toyota reported a 61% reduction in development costs and a 33% reduction in development time for projects that used QFD systematically. The method spread to Ford, General Motors, Xerox, HP, and eventually across every industry where the distance between customer desire and engineering reality was measured in millions of dollars.

The core idea is deceptively simple: deploy quality functions — which is to say, translate what the customer wants into what the engineer builds, layer by layer, without losing meaning in translation.

The House of Quality: Architecture of Translation

The House of Quality is the first and most famous matrix in the QFD process. It gets its name because the completed diagram looks like a house — a rectangular body with a triangular roof on top.

Here is what each section represents, and more importantly, what each section does:

1. The Customer Requirements (WHATs) — The Left Wall

This is the voice of the customer, captured and organized. Not what engineering thinks the customer wants. Not what management hopes the customer wants. What the customer actually says when you ask the right questions.

These are typically gathered through interviews, surveys, focus groups, warranty data analysis, and complaint logs. The trick is to dig past surface-level requests to the underlying need. When a customer says “I want a lighter laptop,” they may actually mean “I carry it on my shoulder for two hours daily and it hurts.” The first is a feature request. The second is a human need — and it opens more solution space.

Each requirement gets an importance rating, typically on a 1-5 or 1-9 scale. This rating will drive every downstream decision, so it must reflect real customer priorities, not internal assumptions.

2. The Planning Matrix — The Right Wall

This section adds competitive context. For each customer requirement, you rate:

Customer satisfaction with your current product (1-5 scale)
Customer satisfaction with competitor products (1-5 scale)
Your target level (where you want to be)
Improvement ratio (target ÷ current performance)
Sales point (how strongly this feature influences purchasing decisions)

This produces a weighted importance for each requirement — a number that reflects not just what customers want, but where you are weakest, where competitors are strongest, and where improvement will actually move market share.

This is where QFD stops being a translation tool and becomes a strategy tool.

3. The Engineering Characteristics (HOWs) — The Ceiling

Now engineering enters the room. For every customer “WHAT,” the team identifies measurable engineering characteristics that influence it. These are the “HOWs” — the design parameters, material properties, process settings, and performance metrics that can actually be controlled.

Critical rule: Every HOW must be measurable. “Surface roughness ≤ 0.8 μm Ra” is a HOW. “Good surface finish” is not. “Response time < 200ms” is a HOW. “Fast response” is not.

This is where the translation happens. The team maps each WHAT to one or more HOWs, creating the relationship matrix — the body of the house.

4. The Relationship Matrix — The Body of the House

This is the living room of the structure. Every cell in the matrix represents the relationship between a customer requirement (row) and an engineering characteristic (column).

Relationships are typically scored:

Strong relationship (9): This engineering characteristic directly and significantly determines whether this customer requirement is met.
Moderate relationship (3): This characteristic has a noticeable influence.
Weak relationship (1): There is some influence, but it is minor.
Blank (0): No meaningful relationship.

Why 9-3-1 instead of 3-2-1? Because the original Japanese methodology wanted to make the difference between strong and moderate unmissable. A 9:3 ratio is a statement: strong relationships dominate. This is deliberate. It forces the team to make hard choices.

The weighted scores are calculated by multiplying the relationship strength by the weighted importance from the planning matrix, then summing each column. The result? A ranked list of engineering characteristics by their total contribution to customer satisfaction. You now know which parameters matter most. Not because an engineer said so. Because the customer’s voice, translated through a structured process, said so.

5. The Correlation Matrix — The Roof

This is the triangular roof of the house, and it is where most teams discover something uncomfortable: their engineering characteristics conflict with each other.

Every pair of engineering characteristics is evaluated for synergy or conflict:

Positive correlation (+): Improving one improves the other. Gold.
Negative correlation (−): Improving one degrades the other. This is where the real engineering begins.
No correlation (blank): They are independent.

Negative correlations are the most valuable part of the House of Quality. They surface trade-offs before you have committed to a design. They force the conversation: “We cannot simultaneously minimize weight and maximize rigidity. Which does the customer value more? What is the optimal balance?”

This is the conversation that should happen at a whiteboard in the concept phase. Instead, in most organizations, it happens on the production floor after the first batch fails inspection.

6. The Engineering Targets — The Foundation

The bottom row of the house translates everything into specific, measurable targets for each engineering characteristic. These targets are informed by:

The importance ranking from the relationship matrix
Competitive benchmarking data
Technical feasibility constraints
Cost constraints

This is the final output of the House of Quality: a prioritized, justified set of engineering specifications that trace directly back to customer needs, competitive positioning, and strategic intent.

Beyond the First House: The Four-Phase Deployment

The House of Quality is just Phase 1. The full QFD method deploys through four sequential matrices:

Phase 1: Product Planning (House of Quality)
Customer requirements → Engineering characteristics

Phase 2: Part Deployment
Engineering characteristics → Part characteristics (material grades, dimensions, tolerances)

Phase 3: Process Planning
Part characteristics → Process parameters (machine settings, tool selections, process sequences)

Phase 4: Production Planning
Process parameters → Production requirements (inspection frequencies, control limits, operator instructions)

Each phase takes the outputs of the previous phase as its inputs. The cascade is relentless. By the time you reach Phase 4, every production control on your shop floor traces its justification back to a specific customer need.

This is what “deploying quality functions” actually means: deploying the customer’s voice through every layer of the organization until it reaches the person tightening the last bolt.

When QFD Fails — and It Fails Often

I have facilitated dozens of QFD workshops, and I can tell you the most common failure modes with surgical precision:

Failure 1: Internal people playing customer. The marketing manager fills in the “customer requirements” based on what they believe customers want. No actual customers were consulted. The entire house is built on a foundation of assumptions. When the product launches, nobody buys it, and the team blames QFD.

Failure 2: Too many requirements. A team lists 80 customer requirements and 120 engineering characteristics. The matrix becomes 9,600 cells. Analysis paralysis sets in. The exercise dies of its own weight. The practical limit is 15-25 customer requirements and 20-35 engineering characteristics. If you have more, group them. Abstract them. Prioritize them. Do something other than drowning in them.

Failure 3: Skipping the roof. The correlation matrix is the hardest part to fill in because it demands honest engineering judgment about trade-offs. Teams skip it because it is uncomfortable. Without it, you miss the conflicts, and the conflicts are where the breakthroughs live.

Failure 4: Treating it as a form instead of a conversation. QFD is a facilitated team exercise. The value is not in the completed matrix — it is in the discussions that happen while completing it. When one engineer says “stiffness strongly affects durability” and another says “no it doesn’t, it’s the material fatigue limit,” that disagreement is worth more than any spreadsheet. The matrix is just the structure that makes the disagreement visible.

Failure 5: No follow-through. The team completes the House of Quality, frames it, puts it on the wall, and then ignores it for the rest of the project. QFD outputs must be linked to design reviews, FMEA inputs, and control plan requirements. If the house is a museum piece, it is a waste of time.

A Practical Example: The Electric Vehicle Door

Let me make this concrete. Suppose you are developing an electric vehicle door panel. Here is a simplified slice of what the House of Quality might look like:

Customer Requirements (WHATs) with importance ratings:

Requirement	Importance (1-9)
Easy to close (doesn’t require force)	9
No rattling or vibration while driving	8
Feels solid and premium	7
Quiet cabin (no wind noise)	8
Lightweight (extends range)	6

Engineering Characteristics (HOWs):

Door seal compression force
Panel material thickness
Latch engagement angle
Seal cross-section geometry
Panel material density
Sound deadening mass

Now look at the correlations in the roof. Reducing panel material thickness (to save weight) directly conflicts with “feels solid and premium” and “no rattling.” Increasing seal compression force (for quiet cabin) conflicts with “easy to close.” These trade-offs exist whether you map them or not. The question is whether you confront them during design or during production.

The House of Quality forces you to confront them during design, when the cost of change is a redraw, not a recall.

QFD and the Broader Quality System

QFD does not exist in isolation. It connects to your quality system at multiple points:

FMEA: The engineering characteristics and their importance rankings feed directly into the Design FMEA. The most critical characteristics from QFD become the highest-priority failure modes in DFMEA.
Control Plan: The production-level targets from Phase 4 define what your control plan monitors and at what frequency.
APQP: QFD is explicitly referenced in APQP Phase 1 (Concept and Launch) as a tool for translating customer requirements. If you run APQP without QFD, you are guessing at requirements.
Measurement System Analysis: The engineering characteristics that QFD prioritizes tell you which measurements matter most — and therefore which measurement systems need formal MSA studies.

Think of QFD as the upstream intelligence that makes every downstream quality tool more effective. FMEA without QFD is a list of imagined risks. FMEA with QFD is a focused attack on the failures that would most damage customer satisfaction.

The Human Factor

The most underrated aspect of QFD is not the matrix. It is the cross-functional conversation it forces.

Put a design engineer, a production supervisor, a quality manager, a sales representative, and a purchasing agent in the same room. Ask them to fill in the relationship matrix together. Watch what happens.

The design engineer will insist that a certain parameter has a strong relationship with a customer need. The production supervisor will say, “We cannot hold that tolerance on our current equipment.” The purchasing agent will say, “That material grade costs three times what we budgeted.” The sales representative will say, “Customers don’t care about that parameter — they care about this one.”

This conversation is QFD. The matrix is just the scorecard.

In my experience, organizations that use QFD well do not use it as a one-time exercise. They revisit the House of Quality at every major design review. They update it when new customer data arrives. They use it to evaluate engineering change requests: “Does this change improve a high-priority customer requirement, or are we optimizing something nobody cares about?”

That last question — “are we optimizing something nobody cares about?” — is worth more than any quality tool ever invented. QFD gives you the data to answer it.

When to Use QFD — and When Not To

Use QFD when:

You are developing a new product or significantly redesigning an existing one
Customer requirements are complex, conflicting, or poorly understood
You need cross-functional alignment on design priorities
The cost of design errors is high (automotive, aerospace, medical devices)
You are entering a new market and do not yet understand customer expectations

Do not use QFD when:

You are making minor modifications to an existing, well-understood product
Customer requirements are simple and unambiguous
You do not have access to actual customer data (in that case, go get it first)
Your team does not have the discipline to maintain and use the outputs

The Quiet Power of Structured Translation

Here is what I have learned after 25 years of quality work: the biggest quality failures are not caused by bad manufacturing. They are caused by bad translation. The customer said one thing. Engineering heard another. Production built what engineering specified. The customer was disappointed.

Every recall, every warranty spike, every lost contract has a moment in its history where someone could have asked, “What exactly does the customer mean by that?” and connected the answer to a measurable engineering target.

That is what QFD does. It does not guarantee success. But it eliminates an entire category of failure — the category where you build exactly what you specified, and it is exactly wrong.

The House of Quality is not glamorous. It is a spreadsheet with a roof. But in that spreadsheet lives the most powerful question in product development: “Does every engineering decision we are making trace back to something the customer actually values?”

If the answer is yes, you have deployed quality. If the answer is no, you have deployed hope.

Hope is not a strategy. QFD is.

Peter Stasko is a Quality Architect with 25+ years of experience transforming quality systems from compliance exercises into competitive weapons. He has implemented QFD across automotive, electronics, and industrial sectors — and has seen the House of Quality prevent more recalls than any inspection ever could. His philosophy: translate the customer’s voice before the market translates it for you — with a complaint.