How to do UX scientific research? A Simple 7 Steps Guide

UX without research is rolling dice and hoping. Here is the actual scientific method, in seven beginner-friendly steps that still hold up academically.

Quick answer

Scientific research is a series of phases for answering a problem through empirical observation and logical reasoning. There is no single scientific method, but most research shares the same steps: (1) delimit the scope and state the problem, (2) review the literature, (3) state hypotheses, (4) test hypotheses with standardized, operationalized procedures, (5) analyze the results, (6) draw conclusions and discuss theory, (7) communicate/publish the results so they can be scrutinized and replicated. In UX, the deliverable itself is the ‘publication’. The steps can flex, but you don’t skip any.

Key takeaways

All scientific research starts from a problem, rests on empirical observation, and reasons toward new knowledge.
There is no single scientific method; steps flex by discipline and context, but none should be skipped.
A tight scope (sample, geography, demographics) keeps data relevant, as in an A/B test for over-50 life insurance in New York.
A hypothesis is a tentative answer that must always be tested, never assumed true.
In UX, the deliverable is the publication; every step must be documented for the client.
UX research basics can be learned in about a week, but analysis should be supervised by expert researchers.

The 7 steps

Step 1

Delimit scope and state the problem

Define the research area and boundaries, or results become unreliable or unmanageable. Testing an A/B design for over-50 life insurance in New York against a mixed audience would drown you in irrelevant data. Narrow by sample size, geography, interests and demographics, and choose the methodology and tools here (face-to-face vs remote), explaining why.

Step 2

Literature review

Identify and evaluate published work relevant to your problem. You build on prior researchers, spot gaps, save time, and answer whether your study adds something new (in UX, it always does).

Step 3

State the hypotheses

A hypothesis is a tentative, testable answer to the research question, not an inherent truth. Deduce observational consequences: if it were true, what should we see? In new or exploratory areas the hypothesis can be a point of arrival rather than a starting point.

Step 4

Test hypotheses with standardized procedures

Operationalize every term (specify empirical indicators), define the sample from the target population, select accurate and reliable instruments, design the study to fit the objectives, and collect data in as controlled a setting as possible.

Step 5

Analyze the results

Process the raw data with the right tools (statistics, etc.) and interpret it to see whether it supports or refutes the original hypothesis.

Step 6

Draw conclusions and discuss

Interpret findings against theory. In fast-changing UX environments, prior observations can be proven wrong when tested for real (see A/B testing and Quantum UX).

Step 7

Communicate the results

Publish so others can scrutinize, replicate and reach inter-observer agreement, the bar for knowledge to count as scientific. In UX the deliverable is the publication, and every step (user studies, competition analysis, market research) must be documented.

Conducting UX without research is like rolling the dice and hoping for the best. It might work, but the odds of success are almost nil.
Mariana Arismendi, PhD

Is UX research hard?

General scientific research takes real study, but applying research methods in UX is relatively simple and strictly necessary. You can grasp the basics in about a week; becoming a strong researcher takes years. Beginners can handle simpler methods, but data analysis and research direction should be supervised by expert researchers. For a deeper method, see the factorial design and A/B & multivariate testing guides.

What are the steps of scientific research in UX?

Delimit scope and state the problem, review the literature, state hypotheses, test them with standardized procedures, analyze results, draw conclusions, and communicate the results.

Is there only one scientific method?

No. Methods differ by discipline and historical context, but they share a common structure and none of the core steps should be skipped.

What is a hypothesis?

A tentative, testable answer to a research question. It holds no inherent truth and must always be tested against observable consequences.

What does it mean to operationalize a hypothesis?

To specify the empirical indicators and measurements needed to gather data that can support or falsify the hypothesis.

How long does it take to learn UX research?

The basics take about a week, and simpler methods can be run by beginners, but analysis and direction should be supervised by experts, and mastery takes years.

Want research-backed UX instead of guesswork? Talk to Dorve.

Start a project

How to do scientific research correctly in UX? A Simple 7 Step Guide

Key takeaways

The 7 steps

Is UX research hard?

Want a team that knows the theory?