Prototyping and Iterative Testing
This domain tests whether you can choose the right prototype fidelity, design a test that produces decision-grade evidence, and iterate without confusing a prototype with a production-ready product.
Fidelity follows the question
- Low fidelity (sketches, foam models, clickable wireframes) tests concept, form and flow.
- Looks-like prototypes test appearance and user perception.
- Works-like prototypes test function, tolerance and reliability.
- Pre-production units test the actual materials and process, and support regulatory or validation evidence. An MVP is the smallest release that delivers validated learning or value; it is not simply a low-quality final product. A prototype built in a different material or process can answer a usability question but not a durability question.
Test design and iteration
Define the question, acceptance criterion, sample size, environment and measurement before building. Change one major variable per iteration where possible; use design of experiments when interactions matter. Record failures and the resulting design change. Iteration is controlled learning, not repeated tweaking without a hypothesis.
Worked example
A battery door must survive a 1.2 m drop onto concrete. The test plan uses 30 production-representative units, three orientations, and a pass criterion of zero openings. First round: 3 failures, all at the latch, all on the corner orientation. Root-cause analysis shows latch engagement varies with moulding tolerance. The team changes from a snap latch to a screw-fixed door, retests 30 units with zero failures, but assembly time rises by 18 seconds. The decision is to accept the screw for the first release and develop a tolerance-controlled snap for the next model. The reasoning links failure mode, root cause, design change and a measured trade-off.
Traps that cost marks
- Testing a 3D-printed prototype for injection-moulded part strength.
- Using too few samples and declaring success after one pass.
- Changing several variables at once and attributing the result to one.
- Confusing prototype testing with validation on production units.
- Reporting anecdotes instead of pass/fail against a stated criterion.