Thematic analysis

What it is

The approach was formalised by Virginia Braun and Victoria Clarke in a 2006 paper that has become one of the most-cited methodological references in HCI. Their six-phase process: familiarise yourself with the data, generate initial codes, search for themes, review themes, define and name themes, produce the report.

The method sits between two extremes: purely inductive analysis (themes emerge entirely from the data) and purely deductive analysis (themes are pre-specified from theory). Most UX research uses a hybrid — some themes pre-specified by the research questions, others emerging from the sessions.

Why it matters

The alternative to systematic analysis is “selective quoting” — researchers unconsciously surface the quotes that confirm their intuition. Thematic analysis is the discipline that forces engagement with the whole dataset, exposes contradictory evidence, and makes the synthesis defensible to stakeholders who weren’t in the sessions.

When to use it

After any qualitative research: interviews, usability testing, diary studies, open-ended survey questions. Not needed for small sessions where the pattern is obvious from the first two transcripts — but even then, a quick codebook check guards against premature closure.

What it looks like in practice

A product team ran 12 user interviews about why users had abandoned the signup flow. Analysis went through Braun and Clarke’s six phases: familiarisation (reading transcripts twice), coding (highlighting every mention of friction with a tag), theme search (grouping tags into candidate themes), review (checking each theme against the full dataset), definition (naming the three themes that survived), reporting (write-up with quote evidence for each).

Tools: Dovetail, Marvin, Atlas.ti and (increasingly) LLM-assisted coding tools. Spreadsheets and index cards still work. Choose based on team workflow, not vendor pitch.

What people get wrong

• Confusing categorisation with analysis. Grouping quotes under topic headings isn’t thematic analysis — analysis requires interpretation of what the theme means, not just that it exists.
• Premature theme-locking. Deciding the themes in the first three sessions biases the coding of the remaining sessions. Review themes against the whole dataset at the end.
• Over-reliance on quote frequency. The theme that appeared most often isn’t necessarily the most important — severity, decision-relevance and user context matter.
• Treating LLM-generated themes as finished analysis. LLM-assisted coding accelerates the first pass; human judgement is still required for interpretation and verification.

Learn next

Thematic analysis is the synthesis step for most qualitative UX research. It sits alongside affinity mapping (more visual, less textual), framework analysis (more structured), and grounded theory (more theoretical).

Reviewed 2 October 2026 · Editorial standards