Ideas, comparison, testing and design iteration

Compare different ways of achieving the same purpose

An idea is a possible response to the brief. Alternatives should differ in a way that could matter: a label strip on the current shelf, a tray with outlines, or adjustable compartments. Three colour versions of one arrangement explore less of the problem than three different ways to find and return tools.

The Design Council's Double Diamond describes opening up and narrowing down while understanding a problem and developing responses. Real work can loop back. A trial may reveal that the stated difficulty was incomplete, as when a return problem turns out to involve missing tool names as well as storage positions.

Build only enough to learn what matters next

A prototype is a representation used to investigate an aspect of a proposed solution. A paper layout can explore label placement and spacing. It cannot establish the strength or durability of a finished tray. Choosing the model starts with choosing the question.

Before trying a layout, write a prediction: ‘Pairing each name with an outline will reduce swaps between these two tool positions.’ Define the task and what counts as a swap. Record what happened even if it contradicts the prediction.

The diagram makes the feedback path visible. A revision follows an observed issue and a proposed explanation. It then needs another check; changing the design is not itself evidence that the change helped.

Make the comparison fair enough for its purpose

In this constructed comparison, every candidate is evaluated against the same width and material-cost limits. For a useful comparison of return tasks, keep the starting tool set, task instructions and success rule consistent while allowing each candidate's storage layout to differ. Here success means returning a tool to its correct position without a hint. Record participant experience too: repeating a task may improve performance even when the design has not improved.

A weighted score can be useful later, but it adds judgements about importance and scale. Begin with a visible table. Show failures, missing evidence and trade-offs instead of making one total look more certain than the information behind it.

Keep layout and practice from becoming the same explanation

Suppose four intended volunteers are available for a new exploratory comparison of A and C. If everyone uses A first and C second, layout and practice always change together. A later improvement could come from remembering the tools. For this small plan, prepare two A-then-C order cards and two C-then-A cards, shuffle them and assign one to each volunteer before starting. Each layout will then occur first twice and second twice.

Give each volunteer the same listed ten return tasks on each layout, with the same tools, neutral instructions and success rule. Reset the starting arrangement before each task. For this question, use matched A and C paper layouts with identical positions, outlines and size; change only the added names. These are layout mock-ups, not the earlier width specimens. Record the participant code, layout revision, first/second position, each task outcome, wrong placement, hint and prior familiarity. This is a new comparison, not a reconstruction of the earlier table.

Read the individual error records and compare results within each order before combining totals. If whichever layout comes second improves, practice is a plausible explanation worth investigating. Reversing order stops C from always receiving the second-position advantage; it does not erase memory. Seeing names on C might teach a person where tools belong on A, and the effect need not work equally in reverse.

This plan uses each person as a comparison with themselves, reducing some differences between people. If learning persists strongly, another option is to assign fresh people to just one layout, accepting that differences between the groups then need attention. Four volunteers cannot establish a population-wide effect. Use this round to locate errors and decide the next revision or a more informative comparison, while retaining the order and task records.

Use a failure to choose the next question

A meets width and cost but misses the constructed return target by one successful task. The error notes show two swaps between similarly shaped positions. The team proposes adding tool names beside the outlines.

C is the revised candidate. Its illustrative record reaches nine correct returns, but those totals alone do not prove the labels caused the change. People may have learned the arrangement, and the totals do not document every condition of the comparison.

A useful next comparison would document those conditions and vary the order of layouts where practical. The immediate decision is narrower: C is worth further evaluation, and its evidence should travel with its design revision.

Try the reasoning

1. Can a successful paper layout establish the durability of the finished tray?

   Show the answer and reasoning

   No. It can address layout questions. Durability needs evidence from a model or product appropriate to that question.

2. Why is B's ten-out-of-ten result insufficient to select it under this brief?

   Show the answer and reasoning

   B exceeds the mandatory width limit. Its return result cannot cancel the fit failure.

3. Does C's higher total prove that the added names caused the improvement?

   Show the answer and reasoning

   No. Learning, task differences and other changes could matter. The totals motivate a clearer comparison and inspection of individual errors.

4. A team compares two unfamiliar catalogue screens, X and Y. Every person tries X before Y and Y has fewer errors. Propose a clearer next comparison and explain one remaining limit.

   Show the answer and reasoning

   For a small exploratory round, allocate some intended users to X-then-Y and others to Y-then-X, choosing their assigned orders before the sessions. Keep the search goals, instructions, reset state and error rule the same; retain each user’s screen order, familiarity and task-level errors. Examine both orders because practice or fatigue could favour the second screen. Balanced orders do not remove unequal carryover: Y may reveal information useful on X. Fresh groups seeing one screen each can avoid that carryover, but then group differences and a small sample still limit the claim.

What to carry forward

Use each model to answer a specific question. Compare against agreed conditions, record errors, revise and check again.