Step 6 · Verify

Prototype, measure and improve

Make each prototype answer a precise question, then let the results guide what comes next.

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.

QuestionSuitable 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 precise question the prototype must answer.
A dimensioned plan or sketch and the specifications version.
The list of parts, tools and protective equipment needed.
Pinch, tipping, cutting and falling points identified.
Criteria for stopping the test immediately.
Measurements to record and the person responsible.

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.

Do not reproduce a dangerous test suggested automatically. For heavy loads, mechanisms, locking and stability, have the protocol defined or validated by a qualified person.

Measurement table

TestPlanned measurementResultObservationDecision
TransportWeight and routeTo completeTo completeAccept / change / redo
DeploymentTime and number of stepsTo completeTo completeAccept / change / redo
StorageAccess timeTo completeTo completeAccept / change / redo
ErgonomicsScenario and feedbackTo completeTo completeAccept / 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.

Analysis request

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

PlanPrototypeMeasureUnderstandChangeRetest

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.