Evidence-Based Design Iteration and Decisions
A practical guide to using evaluation evidence to improve designs, manage iteration, and communicate decisions with traceable reasoning.
From Evaluation Findings to Traceable Changes
Testing can refine the solution—and the problem
means using evidence from a test to compare a design with user needs and requirements, decide what to change, and test the revised design. It is not always a matter of fixing the product. A test may instead show that a requirement is unclear, incomplete, or based on an assumption that needs to be reconsidered.
To turn an evaluation into a traceable action, record what was observed or measured, who or what was affected, and the task, context, and limitations of the evaluation. Then connect the finding to the relevant need, requirement, or design assumption. This context matters: the same difficulty may have different implications for different users or tasks, and a finding should not be treated in isolation.
Work through the self-checkout example
In a library self-checkout test, several users cannot find the “scan another item” control. That observation identifies a problem, but not its cause. The team investigates whether the control is hard to see, the instructions are confusing, or the task flow is unclear. Those are possible explanations to examine, not conclusions already established by the observation.
For each finding, the team decides whether to keep, revise, add, clarify, or remove a requirement, change the design, or gather more evidence. In this example, the team might revise the interface and test it again, clarify or change a requirement, or gather more evidence before choosing. The original requirement that users complete checkout “quickly” may need refinement if it lacks a measurable target or specified user conditions. The team can then define how completion time and errors will be assessed; no particular target or outcome is implied by the example.
For each chosen action, define a test or analysis that can check whether the issue has been addressed. Keep requirements clear, consistent with stakeholder needs, and verifiable. Record why a requirement changed, and consider effects on related requirements, interfaces, safety, cost, schedule, and verification plans. This traceability makes it possible to follow the reasoning from an observation to a change and its check.
Planning and Bounding an Iteration
Make each iteration answer a question
A deliberate iteration follows four parts: plan, test, learn, and decide. Before testing, state the question and purpose, identify the participants or conditions, and specify what evidence would support a decision. This makes it clearer what the test can establish—and what it cannot.
Choose a that fits the question. A sketch can help test whether an idea is understandable; a mock-up can help test layout; and a working can help test performance or integration. Using an early, small-scale test can reveal issues before greater investment in a more detailed solution.
After testing, interpret the evidence and prioritize findings by their effects on user needs, requirement compliance, safety, and project constraints. Select changes accordingly, then consider whether a change could create a problem elsewhere. If a result is inconclusive, record the uncertainty instead of treating a guess as a finding.
Set boundaries and a next decision
Agree on the next decision, the evidence needed to make it, available time and resources, and any non-negotiable requirements. Also decide what criteria would support moving forward. Continue revising when important questions remain unresolved; proceed or stop when agreed criteria are met, remaining risks are acceptable, and further testing is unlikely to change the decision.
These boundaries keep iteration focused without making the design process rigid. Teams may alternate between exploring options and narrowing choices as evidence accumulates; the work need not follow a strictly straight-line sequence.
Making Decisions Others Can Follow
Compare options openly
Compare alternatives using explicit criteria. These may include user effectiveness, accessibility, safety, feasibility, cost, and schedule. State the trade-offs: one option may improve a measure while weakening another, so the choice should not be presented as an improvement in every respect. Involve affected stakeholders when their needs or expertise are relevant to the decision.
Preserve the reasoning
For each significant decision, keep a concise record of the decision and date; the problem or requirement addressed; the evidence considered and its limitations; the alternatives and criteria; the rationale and important trade-offs; affected requirements, designs, and tests; and unresolved risks, assumptions, or follow-up actions. Recording only which option was chosen leaves others unable to assess how that choice was reached or what remains uncertain.
Communicate the decision in a form suited to its audience. A , annotated drawing, or comparison table can make a change visible; explain its rationale and consequences in plain language. Keep requirements and design records aligned so others can understand why the solution changed and compare future proposals with the original reasoning. A clear record therefore supports both current coordination and later decisions about the design.
: A clear, stakeholder-aligned statement that can be checked through verification.
: An organized cycle of planning a test, learning from its results, and deciding what to do next.
: A concise record of a significant design decision and the reasoning and consequences behind it.
: A stated basis for comparing possible design options, such as safety, accessibility, cost, or schedule.