Engineers can tell by starting with a user’s goal and current difficulties, testing a solution with people who plausibly have that need, and checking whether it improves a meaningful outcome. A feature request, positive reaction, or successful prototype task is useful evidence—but none alone proves that the feature addresses the underlying problem.
Start with the user’s problem, not the proposed feature
A request such as “add a dashboard” describes a possible solution, not necessarily the need behind it. First establish who is trying to do what, when the need arises, how they handle it now, and what prevents them from reaching the outcome they want. GOV.UK’s user-needs guidance recommends writing needs from the user’s perspective and focusing on the problem rather than a solution.
A useful starting format is “I need to [do something] so that [outcome].” Add context—such as the user group, trigger, or constraints—where it changes what the team should build. Keep the wording recognizable to users rather than relying on internal product language. A stakeholder opinion or user suggestion that has not been checked against evidence is an assumption to investigate, not yet a validated requirement.
Build a picture from behavior and context
Use more than one evidence source. Review what the team already has, such as product analytics, search logs, support and call-center data, and earlier research. These can show what people do, where they get stuck, or what they seek, but they may not explain why. Interviews and observation help reveal users’ context, workarounds, frustrations, and desired outcomes.
#1 Best Overall
Include people who struggle with existing routes as well as confident users, and consider people who support users in completing the task. GOV.UK’s user research guidance recommends planning research around the questions and decisions at hand, selecting appropriate methods, and considering time and cost. Turn uncertain beliefs into explicit research questions before committing to implementation.
Choose a test that answers the next question
Do not build a polished feature just to find out whether its basic idea is understandable. For an early question, a sketch or paper prototype may be enough; detailed interaction questions may require a higher-fidelity prototype or something closer to the live product. The right level of realism depends on what the team needs to learn.
Rank #2
- Product Condition: No Defects
- Good one for reading
- Comes with Proper Binding
Recruit people who are plausible users and give them realistic tasks with clear success criteria. Let them attempt the task without steering them toward the intended path. Observe whether they complete it, hesitate, make errors, misunderstand labels, invent workarounds, or repeatedly encounter friction. Use the pattern of difficulty to revise the design and test again. The Office for Health Improvement and Disparities describes this approach in its qualitative research guidance.
For qualitative usability testing, that guidance suggests recruiting 5 to 6 participants. GOV.UK’s broader research-planning guidance says many qualitative methods commonly use 4 to 8 participants per round. These are planning suggestions for learning and iteration, not guarantees of representation or estimates of how common a problem is. Surveys, A/B tests, and benchmarking generally need substantially larger samples; GOV.UK notes that hundreds may be needed for clear findings, though the appropriate sample depends on the study and is not a universal power calculation.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Match the evidence to the claim
Different methods answer different questions. Interviews and observation help explain needs and context. Usability testing shows whether people can complete tasks and where interaction breaks down. Analytics can reveal behavioral patterns at scale. An experiment can help determine whether a change caused an outcome shift. No single method should be asked to prove more than it can support.
| Evidence | What it can help establish | What it does not establish by itself |
|---|---|---|
| User interview or observation | Users’ goals, context, frustrations, and current workarounds | That a proposed feature will change behavior or improve outcomes |
| Prototype usability test | Whether participants can use an interaction and where they encounter friction | That the feature is the right intervention or will improve real-world outcomes |
| Analytics or search and support data | Patterns in behavior, demand, and points of difficulty | Why a pattern occurs or whether a particular feature caused a change |
| Experiment or real-world evaluation | Whether an intervention is associated with a measurable outcome change; a well-designed experiment can help assess causality | Results beyond the tested users, conditions, or outcome measures without further evidence |
A participant who completes a controlled task has shown that the task was possible in that session, not necessarily that the feature helps people in everyday use. Think-aloud sessions can clarify comprehension and experience, but spoken impressions may differ from behavior, and participants may say what they think the researcher wants to hear. Pair comments with observed actions and, when appropriate, outcome data. The OHID’s think-aloud study guidance explains this method and its limits.
Rank #4
Define and measure the intended outcome
Before launch or a real-world pilot, state what should improve in concrete terms: for example, whether users can complete a task, how long it takes, whether they need support, or whether they achieve the goal they came for. Choose an outcome that reflects the user need rather than a convenient proxy such as feature clicks alone.
When uncertainty or potential harm warrants it, test the feature in realistic conditions or use an experiment designed to distinguish its effect from other changes. The UK Government’s Magenta Book Test and Learn annex recommends agreeing on a measurable outcome, testing critical assumptions early, learning from real-world evidence, and using feedback loops. It also states that this approach strengthens rather than replaces robust evaluation.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsBest Value
Record why the feature exists and what would change the decision
Keep the user need, evidence, assumptions, intended outcome, acceptance criteria, and test results together. That record lets engineers and product teams explain the rationale, check whether requirements were met, and revisit the decision as the product and its users change. Home Office engineering guidance on designing from evidence says evidence-based decisions improve the ability to meet user needs and calls for evidence and decisions to be transparent and documented.
A practical decision record should make clear which findings are observed, which remain assumptions, and what result would prompt another iteration or a different solution. Keep the evidence current and relevant to the users and conditions in which the feature is meant to work.
Use sample-size guidance as a starting point, not a rule
Small qualitative rounds are valuable for uncovering usability problems and improving a design; they do not estimate population-wide effects. The OHID reports an example of 29 participants across four rounds in an EPIC HIV usability study in rural South Africa, with recruitment refined toward older and more rural participants as barriers emerged. That is an illustration of adapting research to context, not a sample-size formula for other products.
Likewise, a four-session think-aloud study described by OHID—one facilitator and two observers evaluating a digital-health guidance site—is an example, not a general recommendation. Participant fit, inclusion, the question being asked, and the consequences of error should shape the study plan. Guidance from UK public-sector sources should be adapted to the product’s domain, regulatory setting, user population, and risks.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
