What needs to be finished first
At least five to eight qualitative interviews with real users from the target segment have been conducted or are firmly scheduled.
The most important stakeholders, a sponsor with decision mandate, and a budget frame for prototypes and tests are documented.
What needs to be ready before start
Workshop room with wall space or Miro board with phase lanes; stickies in five colors; pens; prototyping materials (paper, cardboard, click tools like Figma); test setup; recruiting list for user tests.
One facilitator for overall flow; one to two researchers for interviews and synthesis; two to four designers and engineers; one sponsor with decision mandate (at least at phase transitions); test users for phase 5.
Problem space as one paragraph, not as a solution idea; target group and access to it; known constraints (budget, compliance, technology); prior solution attempts and why they failed.
5-day sprint or 4-8 weeks iteratively
Phase lanes on the wall: Empathize, Define, Ideate, Prototype, Test. Fix sponsor reviews between every phase. Start recruiting for phase 5 in phase 1, not later.
The one question this method answers
Which real user problem is worth solving, and which solution idea survives a test with real users?
Marker: Phase
| Step | Duration | Action | Hint |
|---|---|---|---|
1Phase 1: Empathize | 3-10 days | Conduct 5-10 interviews and 2-3 observations in context. Collect notes, quotes, observed workarounds, and emotional reactions. Ask no solution questions. | When interview questions start with "Would you use X," the phase only produces polite data. Ask about past behavior, not hypothetical intention. |
2Phase 2: Define | 0.5-1 day | Cluster patterns. Formulate a Point-of-View statement: User X needs Y because Z (Insight). Then derive 3-5 How Might We questions and select the most important one. | If the POV statement includes a solution ("needs a mobile app"), re-formulate it. Z (Insight) must be surprising or counterintuitive, otherwise Empathize was superficial. |
3Phase 3: Ideate | 0.5-1 day | Crazy 8s, Brainwriting, or Worst Possible Idea for the HMW question. Generate at least 40 ideas, then cluster and select 2-3 concepts for prototyping. | If the group stops after 20 ideas, divergence is missing. Enforce at least one round of absurd or contrarian ideas, otherwise the group converges on the first obvious solution. |
4Phase 4: Prototype | 1-5 days | Build a low-fidelity prototype that makes the central question directly testable. Paper, click dummy, or Wizard-of-Oz. No production code. One prototype per concept. | Looking realistic beats being complete. The prototype is a question, not an answer. Anyone building backend logic has already left the phase and is already working on the solution. |
5Phase 5: Test | 1-3 days | Moderated one-on-one tests with 5-8 participants per prototype. Observe what users do, not what they say. Synthesize patterns after test 5. Decide: continue, pivot, or drop. | Pattern recognition starts after test 5, not after test 2. Keep the observation matrix filled consistently per test. When behavior and verbal report conflict, behavior wins. |
What comes out at the end
Structured documentation with research notes, clustered insights, POV statement, selected HMW question, ideation outputs, prototype links, test protocols per user, patterns, and final decision with owner and next action.
One iteration per folder with date and problem space as title. Archive previous iterations without overwriting. Document phase-to-phase decisions with date and sponsor sign-off.
Compact working template for Design Thinking with empathy, define, ideate, prototype, and test.
Move from understanding the user context to a testable prototype and learning loop.
When and for what do we use this method?
Which data, observations, decisions, or materials are available?
Owner, date, and success signal.
Concrete filled scenario, fictional example
Concrete filled scenario, fictional example
Empathize: 7 interviews with solo tax advisors, 2 days of contextual observation in Sabine's office. 12 workarounds documented (WhatsApp for documents, Excel lists for follow-ups).
POV: Sabine, a solo tax advisor with 80 clients, needs a way to assign incoming documents to the correct client because she sorts each email manually and loses 6 hours per week.
HMW: How might we automatically assign documents to the right client without clients changing their behavior?
Ideation: 47 ideas, 3 concepts selected for prototyping: (1) email forwarding with tag detection, (2) client upload portal, (3) WhatsApp bot.
Prototype: Figma click dummy for concept 1 (9 screens) plus Wizard-of-Oz setup, where a researcher manually assigns tags.
Test: 6 tax advisors, 60 minutes each. 5/6 understood the concept immediately, 4/6 wanted to use it now, 2/6 had concerns about confidentiality in WhatsApp-like tag routing.
Decision (Sponsor: @marcus, 18.05.2026): Continue with concept 1, with explicit opt-in per client. One-week implementation spike first, then a pilot with 5 advisors.
Recognize symptoms and steer against them
Team starts with solution ideas, interviews are "deferred" or replaced with internal assumptions.
Make phase 1 a hard prerequisite for phase 2. No POV statement without at least five interviews. Sponsor must enforce this rule.
Define phase ends with "User needs a mobile app" instead of a need.
Use a strict POV template: who, needs what, because which surprising insight. Remove solution components and move them to HMW or Ideation.
Group chooses the most obvious idea after 15 minutes because it seems obviously good.
Force at least 40 ideas and at least 5 absurd ones. Only after that move to convergence. Dot voting only after full divergence, not before.
Phase 4 takes 4 weeks instead of 4 days, engineering starts building backend because "we need it anyway."
Fidelity rule: build only what is needed to test the central question. Prefer click dummy or Wizard-of-Oz before code. Anyone building backend has left phase 4.
Phase 5 is filled with colleagues, friends, or existing power users; feedback is biased.
Define recruiting criteria in phase 1 and enforce them in phase 5. Use external recruiters as backup. Better to postpone the phase than test with a wrong audience.
Team runs phase 1-5 once and treats the result as final, without returning to earlier phases.
After phase 5, ask explicitly whether to go back to Define, Ideate, or Prototype, or move to implementation. Iteration is the norm; linear passes are the exception.
Done signals checkable in under a minute
Go to the profile for purpose, similar methods, and sources or continue to the next method in the catalog.