A prototype answers one question
The first prototype should not try to be beautiful or final. It should reduce one important uncertainty with as little time and expense as possible.
| Question | Suitable prototype |
|---|---|
| Does the size fit in the vehicle? | A full-scale cardboard volume. |
| Is the working height comfortable? | A provisional top, adjustable on safe supports. |
| Is the storage easy to understand? | A mock-up of drawers and tool outlines. |
| Is deployment too complicated? | A working prototype of the mechanism, unfinished. |
| Does the structure stay stable? | A reinforced prototype with a careful protocol and a safe area. |
Prepare before making
The five tests of the case study
1. Transport
Load the planned tools, move the whole set along the real route and check handles, wheels, latches and loading into the vehicle.
2. Deployment
Time it from the closed position to full locking. Note hesitations, pointless movements and pinch risks.
3. Access to the tools
Ask the user to find, use and put back several representative tools without help.
4. Ergonomics
Observe the posture and gather feedback during the tasks actually planned, without extrapolating to tasks not studied.
5. Stability and strength
Define a careful, gradual protocol suited to the risks. People's safety comes before obtaining a value.
Measurement table
| Test | Planned measurement | Result | Observation | Decision |
|---|---|---|---|---|
| Transport | Weight and route | To complete | To complete | Accept / change / redo |
| Deployment | Time and number of steps | To complete | To complete | Accept / change / redo |
| Storage | Access time | To complete | To complete | Accept / change / redo |
| Ergonomics | Scenario and feedback | To complete | To complete | Accept / change / redo |
Have the results analysed
After the test, send ChatGPT the table, the protocol and the observations. Ask it to distinguish the findings from the interpretations.
Analyse these test results. For each gap with the target, state: the measured finding, the possible causes ranked without excessive certainty, the missing data and a next test to tell the causes apart. Do not recommend any change that is not linked to an observed result.
The improvement loop
Close a version
Version tested: reference and date.
Goals reached: validated criteria.
Failures or limits: results that did not conform.
Changes decided: with reasons.
New risks: created by the changes.
Next milestone: a new mock-up, a reinforced prototype or final validation.
ChatGPT can analyse data files, produce tables and charts and let you review the code used. Always check that the columns, units and missing data are correctly understood. Official reference: Data analysis with ChatGPT.