What research methods does a ux design agency use?

Which research methods lead?

Agencies rely on a core set of eight research methods, split between watching what people do and asking what people think. Watching methods carry more weight, because actions recorded on screen tell truths that polite answers in interviews often hide.

Every project mixes several methods rather than leaning on one, since each method answers a different question. Method lists published across any UX design agency directory show the same core set appearing again and again, which marks these eight as the working standard of the field rather than the habit of a few studios.

Watching methods come first in most projects.

1. Usability sessions, where a person attempts real tasks while researchers observe.
2. Screen recordings from live products, showing where thousands of users pause or quit.
3. Analytics review, counting which paths people take and which pages lose them.
4. Field visits, watching people use products inside homes and workplaces.

How do asking methods work?

Asking methods gather goals, feelings, and reasons that watching alone cannot reach. A recording shows a person quitting a form, while only a conversation reveals the quit happened because a question felt too personal.

Four asking methods complete the core set.

1. Interviews, held one-to-one, digging into goals and past struggles.
2. Surveys, reaching hundreds of people with short fixed questions.
3. Card sorting, where users group content into piles that make sense in ordinary minds.
4. Diary studies, where people note product moments across days or weeks in their own words.

Answers from these methods get treated carefully, because memory bends and people describe themselves as more patient and more logical than recordings prove. Skilled researchers ask about specific past events rather than habits in general, since one true story beats ten polished self-descriptions.

Matching methods to stages

Method choice follows the project stage, and agencies match the question of the moment against the tool that answers the question the most. Early stages need broad learning, middle stages need testing, and late stages need confirmation.

• Project start favours interviews and field visits, mapping goals before anything gets drawn.
• Structure work leans on card sorting, shaping menus around real mental groupings.
• Draft stages run usability sessions weekly, catching problems while corrections cost hours.
• Post-release time belongs to analytics and recordings, watching real behaviour at scale.

Budget shapes choices too. A survey covers many people at low effort, while field visits cost days per person, so agencies spend heavy methods only where lighter ones cannot answer.

Combining methods for proof

Methods combine because every single method carries a blind spot, and pairs cover what singles miss. Numbers from analytics say what happened without saying why, while interviews say why without proving how often.

A common pairing runs analytics first to find a problem page, then usability sessions on that exact page to watch the failure happen, then a small survey to check how widely the frustration spreads. Three tools, one question, and an answer no single tool could give. Findings crossing two or more methods earn trust, while findings from one source alone stay marked as leads needing confirmation.

Eight methods, split between watching and asking, form the research core of agency practice. Stage decides which method leads, pairing covers each method’s blind spot, and behaviour outranks opinion whenever the two disagree. Teams reading agency proposals against this set can quickly judge whether real research sits behind the process words.