DMADV Explained: The 5 Phases of Design for Six Sigma, with a Worked Example
DMAIC fixes a process that exists. DMADV designs one that does not exist yet, or one so broken that fixing it will not be enough. The letters stand for Define, Measure, Analyze, Design and Verify. This guide explains each phase, the tools that fit, and when to choose DMADV over DMAIC. It ends with a worked example, including a test that fails and sends the team back to redesign.
What DMADV is
The ASQ glossary describes DMADV as a Six Sigma design methodology used for developing new products or processes, and notes it is also known as Design for Six Sigma. It says the method applies when existing processes cannot be improved sufficiently to meet customer requirements. See the ASQ glossary entry for DMADV.
The key idea is that the design starts from customer requirements and ends with evidence that the design meets them. Where DMAIC asks “why is this process failing?”, DMADV asks “what must this new thing do, and what design will do it?”
If you are new to the background, read what Six Sigma is first. Our article on design for Six Sigma covers the wider approach.
DMADV or DMAIC?
| DMAIC | DMADV | |
|---|---|---|
| Starting point | An existing, working but flawed process | A new product, service or process, or a redesign from scratch |
| Main question | Why do we get defects or delays? | What must it do, and what design will do it? |
| Phases | Define, Measure, Analyze, Improve, Control | Define, Measure, Analyze, Design, Verify |
| Typical output | A better process with controls in place | A verified design ready for release |
One guide from Learn Lean Sigma says the choice should rest on the nature of the problem rather than a fixed percentage of change. If a stable existing process is the thing to improve, use DMAIC. If a new solution is needed, or the current setup cannot meet future requirements without fundamental redesign, use DMADV. Real projects can straddle the line. See the full side-by-side in DMAIC vs DMADV.
The five phases at a glance
The table below summarizes the phases in our own words, using the description in the Learn Lean Sigma DMADV guide.
| Phase | Key question | Typical tools | Output |
|---|---|---|---|
| Define | What are we designing, for whom, within what limits? | Project charter, stakeholder list | A bounded design brief |
| Measure | What do customers and the business need, in measurable terms? | Voice of the customer (VOC), critical-to-quality (CTQ) measures, QFD | A measurable requirements list |
| Analyze | Which concept best meets the requirements? | Concept alternatives, weighted decision matrix, early risk analysis | A chosen concept |
| Design | What exactly will we build? | Design FMEA, prototypes, simulation, tolerances, design of experiments | A detailed design |
| Verify | Does it work as required under real conditions? | Tests, pilots, statistical analysis | A release, revise or loop-back decision |
The guide stresses that no single tool is mandatory. Pick a tool because the design question needs it, not because a training slide lists it.
Phase by phase
1. Define
- State the opportunity and why existing processes will not do.
- List stakeholders, objectives, constraints and who makes decisions.
- Write the charter. Keep the scope wide enough to allow real options, but bounded.
2. Measure
- Collect voice-of-the-customer data: interviews, surveys, complaints, observation.
- Include business, regulatory and technical needs, not only what customers say.
- Turn each need into a critical-to-quality (CTQ) requirement with a measure and a target. “Fast” becomes “start a return in under 3 minutes.”
3. Analyze
- List the functions the design must perform.
- Generate several concepts, not just the favorite.
- Compare concepts against the CTQs with a simple weighted matrix, and check key risks and assumptions.
4. Design
- Turn the chosen concept into a detailed design: characteristics, interfaces, tolerances and controls.
- Build prototypes or simulations and test the riskiest parts first.
- Keep a trace from each need to a requirement, a design feature and a test. Record changes and who approved them.
5. Verify
- Test against the criteria you set in Measure, under realistic conditions.
- Run a pilot with real users or real volume where possible.
- Decide: release, revise, or go back to an earlier phase.
A worked example
Everything below is made up to show the method. Imagine an online store whose returns are handled by email and take too long. Fixing the email process would help only a little, so the team chooses to design a new returns service.
Measure: from needs to CTQs
| Customer or business need | CTQ measure | Target |
|---|---|---|
| “Starting a return should be quick” | Time to start a return | 3 minutes or less for 90% of customers |
| “I want my money back fast” | Refunds issued within 5 days of the parcel arriving | 95% or more |
| Finance: “Keep handling cost down” | Staff time per return | 5 minutes or less |
Analyze: choosing a concept
Three concepts are scored from 1 (poor) to 5 (strong) against four weighted criteria. The weighted score is the sum of weight × score, divided by 100.
| Criterion (weight) | A: improve email | B: self-service portal | C: outsource to a third party |
|---|---|---|---|
| Meets the CTQs (40) | 2 | 5 | 4 |
| Cost (25) | 5 | 4 | 2 |
| Time to build (20) | 5 | 3 | 4 |
| Risk (15) | 4 | 3 | 3 |
| Weighted score | 3.65 | 4.05 | 3.35 |
Concept B scores highest, mainly because it meets the CTQs. The scores are judgments, so test how sensitive the answer is. If B still wins when you move the weights around, the decision is robust.
Verify: a pilot that fails one test
The team designs the portal and pilots it with 200 customers.
| CTQ | Target | Pilot result | Verdict |
|---|---|---|---|
| Return started in 3 minutes or less | 90% | 92% | Pass |
| Refund within 5 days | 95% | 93% | Fail |
| Staff time per return | 5 minutes or less | 4 minutes | Pass |
One CTQ misses. The team does not release. It looks at why, finds that refunds wait for a manual check, and goes back to the Design phase to trigger the refund automatically when the carrier scans the parcel. It then re-tests. This loop-back is what separates a real Verify phase from a ceremonial one.
Design brief template
DMADV design brief: [project name]
Sponsor: [name] · Lead: [name] · Start: [date] · Target release: [date]
1. Opportunity: [What need or gap are we addressing? Why can’t the current process meet it?]
2. Scope and constraints: In: [..] · Out: [..] · Budget: [..] · Regulatory or safety limits: [..]
3. Customers and stakeholders: [Who will use it? Who is affected?]
4. CTQ requirements: [Measure, target, how it will be tested.]
5. Concepts to compare: [At least three, including a low-cost option.]
6. Key risks and assumptions: [What could make the design fail?]
7. Verification plan: [Tests, pilot size, pass criteria set in advance.]
8. Decision gates: After Define / Measure / Analyze / Design / Verify: [who decides, and what sends work backward.]
Common mistakes
These follow the pitfalls listed in the Learn Lean Sigma guide, in our words:
- Locking in a solution early. Writing the favorite answer into Define or Measure means other concepts never get a fair test.
- Moving the targets. Changing a requirement after a failed test, without recording why and who approved it.
- Tool checklist thinking. Using QFD, FMEA or experiments because they are on the list.
- Verification as a formality. Treating Verify as project closure instead of testing against criteria set in advance.
- Unsupported sigma claims. Announcing a sigma level without a specification, a trustworthy measurement system and suitable data.
- Losing traceability. You should be able to follow each need to a requirement, a design feature and a test.
Learn more
ASQ’s 2022 Green Belt Body of Knowledge lists Design for Six Sigma methodologies in its overview section. For courses, see the Six Sigma Green Belt course guide, and for the levels see Six Sigma belt rankings. For quick, focused improvement work, try a kaizen event. Everything is gathered in the Six Sigma hub.
Online providers such as Coursera and Udemy list lean and Six Sigma courses. We may earn a commission if you sign up, at no cost to you. We are not affiliated with any university. Check each syllabus to see whether design methods are covered.
Frequently asked questions
What does DMADV stand for?
Define, Measure, Analyze, Design, Verify.
What is DMADV used for?
Developing new products, services or processes, or redesigning an existing one so completely that ordinary improvement would not be enough.
What is the difference between DMAIC and DMADV?
DMAIC improves an existing process. DMADV designs a new one. The last two phases differ: Improve and Control in DMAIC, Design and Verify in DMADV.
Is DMADV the same as Design for Six Sigma?
The ASQ glossary says DMADV is also known as Design for Six Sigma. Other guides describe it as one roadmap within Design for Six Sigma, so the terms are often used interchangeably.
What are CTQs?
Critical-to-quality requirements: measurable characteristics that matter to the customer or business, each with a target.
What happens if Verify fails?
Do not release. Find the cause and go back to the phase that needs rework, usually Design, then re-test.
Browse certificate, course and degree programs side by side.