Sensitivity Analysis with Quality Function Deployment
Sensitivity Analysis Integrated with Quality Function Deployment · Also known as: SA-QFD, QFD sensitivity analysis, robust QFD, House of Quality sensitivity analysis
Sensitivity analysis integrated with Quality Function Deployment (QFD) tests how stable the prioritization of engineering characteristics remains when customer requirement weights or relationship matrix scores are varied. By systematically perturbing the inputs of the House of Quality, teams identify which design parameters are truly critical and which rankings would flip under different assumptions — turning QFD from a one-shot prioritization tool into a robust decision framework.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use this method when QFD is being applied to a high-stakes product development or process design decision and the input data — customer weights or relationship scores — carry meaningful uncertainty or disagreement. It is especially valuable in cross-functional teams where different stakeholders assign different importance weights, or when relationship matrix entries are engineering estimates rather than measured data. Do not apply it as a standalone method; it requires a completed or near-complete House of Quality as its starting point. Avoid it when the team's QFD input data are already derived from large, representative customer studies with high inter-rater agreement — in that case the computational overhead of sensitivity analysis adds little insight.
Strengths & limitations
- Reveals which QFD input assumptions most heavily drive final engineering priorities, enabling focused data collection.
- Transforms QFD from a deterministic snapshot into a robust, defensible decision framework.
- Quantifies the confidence the design team can place in each engineering characteristic's rank.
- Helps resolve stakeholder disagreements by showing whether competing weight assignments actually change the actionable outcome.
- Compatible with fuzzy QFD extensions, allowing continuous uncertainty representation rather than point estimates.
- Requires a fully constructed House of Quality; it cannot compensate for poorly defined customer requirements or incomplete relationship matrices.
- Computational burden increases rapidly with the number of customer requirements and engineering characteristics being varied simultaneously.
- Sensitivity ranges must be specified by the analyst; if the assumed bounds are too narrow, genuinely fragile rankings may appear falsely robust.
- Does not optimize the design — it evaluates the robustness of rankings, leaving design trade-off resolution to the engineering team.
Frequently asked
Do I need specialized software to run sensitivity analysis on a QFD matrix?
No dedicated software is required. A well-structured spreadsheet (Excel or Google Sheets) is sufficient for one-way sensitivity runs on moderately sized matrices. For large matrices or Monte Carlo joint sensitivity, Python or R scripts are practical. Several QFD software tools (QFD Capture, QFD Designer) include basic sensitivity features.
How large a perturbation should I use when varying customer weights?
A common practice is to vary each weight by ±10–20% of its baseline value, or to use the observed range of disagreement across stakeholders as the perturbation interval. The key principle is that bounds should reflect realistic uncertainty, not arbitrary numbers — document the justification for your chosen range.
What does a rank reversal mean, and how serious is it?
A rank reversal means that under a different but plausible set of inputs, a lower-ranked engineering characteristic would become higher-ranked than the current top choice. This is a serious signal: it indicates the design team's resource allocation decision depends heavily on assumptions that have not been firmly established, and those assumptions need to be resolved before committing to a design direction.
Can this method be used with fuzzy QFD?
Yes, and it is often more natural in a fuzzy QFD context. Fuzzy sets already represent input uncertainty; sensitivity analysis then examines how the defuzzified priority ordering changes as the membership function parameters shift. The combination is well-documented in the engineering literature, particularly for multi-objective product design problems.
Is this method applicable outside engineering?
QFD itself has been applied in healthcare, education, and service design, and sensitivity analysis travels with it to any of these domains. Wherever a House of Quality is built on estimated or debated input weights, testing the robustness of the output ranking is good practice regardless of the application sector.
Sources
- Fung, R. Y. K., Tang, J., Tu, Y., & Wang, D. (2006). Product design resources optimization using a non-linear fuzzy quality function deployment model. International Journal of Production Research, 44(12), 2483–2504. link ↗
- Akao, Y. (Ed.). (1990). Quality Function Deployment: Integrating Customer Requirements into Product Design. Productivity Press. ISBN: 978-0915299416
How to cite this page
ScholarGate. (2026, June 3). Sensitivity Analysis Integrated with Quality Function Deployment. ScholarGate. https://scholargate.app/en/experimental-design/sensitivity-analysis-with-quality-function-deployment
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Quality Function DeploymentExperimental design↔ compare
- SENSITIVITY-ANALYSISDecision-making↔ compare