House of Quality Software, Explained on an Electric Kettle

The brief

You’re designing the next model of a mid-price electric kettle. Budget is limited: you can seriously improve two, maybe three things. Which ones? That’s the question a House of Quality answers. If you want the bigger picture of the method first, start with What is Quality Function Deployment.

Room 1: What do customers want?

The left wall holds customer requirements — in the customer’s own words. Not “2,200 W heater” but “boils fast”. Sources: interviews, surveys, product reviews, support tickets. Customers also tell you how much each one matters; here we use a 1–5 scale.

Our four: boils fast (5), outside stays cool to the touch (4),quiet (2) and lasts for years (4). A real project would have ten to twenty; four is enough to see the mechanics.

Room 2: What can engineers measure?

The ceiling holds technical characteristics: things you can measure and change. Each one has a unit, a direction of improvement (↑ more is better, ↓ less is better, or hit a target) and a target value for the new model.

Notice what’s missing: “time to boil”. That’s an outcome, not a design parameter — it’s almost entirely set by heater power. Add both, and the “boils fast” requirement gets counted twice. Duplicate characteristics are the most common way a House of Quality quietly lies.

Room 3: How strongly are they connected?

The center is the relationship matrix. For every requirement–characteristic pair you ask: if I change this characteristic, how much does it affect whether the customer gets what they want? Strong is 9, medium is 3, weak is 1, and most cells stay empty. The scale is deliberately steep — one strong link outweighs several weak ones, which keeps the result from being decided by a pile of “well, sort of” scores.

Two cells worth explaining. Quiet × heater power = 3: a more powerful heater boils more violently, and violent boiling is loud. Lasts for years × heater power = 1: more power from the same element means more thermal stress. Neither link is obvious from the spec sheet — which is exactly why the team fills in the matrix together.

Room 4: The roof — how characteristics affect each other

The triangular roof records what happens to one characteristic when you push another:

• Heater power vs. element life — strong negative. Squeeze more watts out of the same element and it wears out faster.
• Heater power vs. noise — negative. Faster boil, louder boil.
• Surface temperature vs. noise — positive. A double wall that keeps the body cool also muffles the sound of boiling. One design change, two wins.

On paper, the roof is a reminder. In QFD Tool you can switch it on for a stage so correlations adjust the priorities, and strong negative pairs between important characteristics land in the technical conflicts list.

Room 5: How do we compare?

The right wall holds competitive assessment: customers rate your product and your competitors on each requirement, 1 to 5.

The picture is clear: we lose on boils fast (3 vs. 5) and on outside stays cool (2 vs. 4) — and both matter to customers. We win on quiet, which customers care about least. QFD Tool can feed these gaps into the math: the further you trail the best competitor on a requirement, the more its importance grows. Separately, a technical benchmark compares your characteristic values with the best competitor or with your target.

Room 6: The basement — priorities

Each characteristic’s score is the sum, down its column, of requirement importance × relationship strength. Divide by the total (145) and you get its share of the engineering attention:

• Heater power: 5×9 + 2×3 + 4×1 = 55 → 38%
• Outer surface temperature: 4×9 = 36 → 25%
• Noise at boil: 2×9 = 18 → 12%
• Heating element life: 4×9 = 36 → 25%

What the kettle’s House of Quality says

1. Heater power comes first (38%) — but you can’t just turn it up. The roof shows two negatives: element life and noise. This is a technical conflict, and it’s where engineering creativity goes: an element with a larger heat-transfer area, for example, gives more power without more stress. QFD Tool lists conflicts like this and ranks them by severity; the sharpest ones are good candidates for TRIZ.
2. Outer surface temperature (25%) is where we trail competitors, and the double wall that fixes it also helps with noise.
3. Element life (25%) ties for second, because cutting corners on it would undo the extra power.
4. Noise (12%) can wait. We’re already ahead, and customers rank it lowest.

What the software does that paper doesn’t

Now suppose marketing comes back from a round of interviews: people buy kettles for open-plan kitchens, and noise matters more than we thought. Importance of “quiet” goes from 2 to 4.

On paper, that’s a recount. In QFD Tool, it’s one edit, and the priorities update immediately. Noise jumps from 12% to 21% and ties with surface temperature and element life, while heater power drops from 38% to 36%. The double wall, which helps with both temperature and noise, suddenly looks like the best investment in the model.

Beyond instant recalculation, the software handles the parts that are tedious by hand:

• Roof and competitor gaps inside the math, each switchable per stage. The summary report shows every step — base priority, after correlations, after benchmarks — so the final number isn’t a black box.
• Implementation difficulty: mark characteristics that are expensive to move, and their priority is adjusted for effort.
• A technical conflicts list ranked by severity, so the trade-offs worth solving rise to the top.
• PDF and SVG export of the finished House for a report, a design review or a thesis appendix.
• Autosave in the cloud, so the matrix is the same file for everyone on the team and nothing is lost when a browser tab closes.

Three mistakes software won’t catch for you

• Requirements written as solutions. “Stainless steel body” is a design choice. The customer wants “looks good for years” — and there may be cheaper ways to deliver it.
• A matrix full of 9s. If everything is strongly related to everything, the priorities flatten out and tell you nothing. A healthy matrix is mostly empty.
• Empty rows and columns. A requirement with no strong link is a need nobody is designing for. A characteristic with no links is a parameter nobody asked for.

After the first House

The kettle’s top characteristics — heater power, surface temperature, element life — don’t tell the people designing the heating element or the plastic housing what to do. The next step is to make them requirements for the parts, and then for the process. That’s multi-stage QFD, and once you have a chain, requirement traceability tells you which customer need each parameter at the bottom serves.