Braun & Clarke thematic analysis: the complete guide
Are you considering using Braun and Clarke's thematic analysis for your dissertation or thesis? It's one of the most widely cited approaches to analysing qualitative data across research disciplines, and also one of the most misunderstood. Get the six phases right and you have a method examiners trust. Get the “themes emerge from data” language wrong, or hand in a set of topic summaries instead of real themes, and it's the fastest way to lose marks in an otherwise solid project. This guide walks through exactly how to do it well, with a worked example, so you don't have to learn the difference the hard way.
Have your transcripts ready?
Get codes and themes that follow these six phases, each linked to the participant quote behind it. Free for up to 3 transcripts.
What Braun & Clarke thematic analysis is
Thematic analysis is a qualitative data analysis method for spotting, organising, and interpreting patterns of meaning across qualitative data, such as interview transcripts, open survey responses, focus groups, or documents. Braun and Clarke did not invent the method; what they did was provide a clear step-by-step process that any researcher could carry out, with named phases and defined decisions at each stage.
That clarity is why the 2006 paper became one of the most cited articles in the social sciences, and why later teaching pieces such as Ahmed et al. (2025) still build directly on their framework.
Two things set it apart from other methods. It is not dependent on a single theory, making it useful whether you lean realist or constructionist. And it works at two levels: semantic, staying close to what your research participants said, or latent, reading for the assumptions underneath. You decide which, and you say so in your write-up.
Braun and Clarke also stress a point worth holding onto before you start: thematic analysis is better understood as a family of methods, not one standardised procedure (Braun and Clarke, 2022a). Reflexive TA, codebook TA, and coding-reliability TA all use the word “coding,” but they rest on different assumptions about what a code and a theme actually are, which is exactly why mixing procedures from two different traditions without saying so causes the incoherence examiners flag most often. If you're undecided whether reflexive TA is the right method for your research question in the first place, our guide on reflexive TA vs. other qualitative methods covers that choice directly.
The six phases (steps) of thematic analysis
Braun and Clarke's thematic analysis covers six phases; however, this is not a one-way road. Oftentimes researchers have to revisit earlier phases several times. For example, a candidate theme that falls apart in phase four will send you back to recode in phase two. Braun and Clarke are explicit that the process is recursive. Naeem and colleagues give a helpful worked account of moving through the steps toward a finished model in their step-by-step process (2023), if you want a second walkthrough alongside this one.
To make each phase concrete, here is a small worked example running throughout: four excerpts from hypothetical interviews with remote workers, coded and developed into a theme across all six phases, so you can see how one thread survives the whole process rather than jumping between disconnected fragments.
Worked Example Dataset (Remote Worker Interviews)
P1: “I answer emails at 9pm because I feel like I should be available, even though nobody's asked me to.”
P2: “My manager never said I had to be online after 6, but somehow I still feel guilty logging off.”
P3: “There's no clock-out anymore. The laptop's always right there.”
P4: “I used to leave work at work. Now work just lives in my house.”
1. Familiarise yourself with the data
Read everything, then read it again. Before you code a single line you need to know your material well enough to recall where things were said. Most researchers transcribe their own interviews for exactly this reason: the typing forces attention. Jot early notes as you go, but hold off on formal coding until you have been through the whole dataset at least once. Reading the four excerpts above together, an early note might be: several participants describe pressure that nobody explicitly applied.
2. Generate initial codes
Work through the data and label anything that looks relevant to your question. A code is a short tag for a feature of the data. In the example transcripts, initial codes are generated as seen below. Code generously: you can always drop codes later, but you cannot cluster what you never tagged. Keep the extract that each code sits on, because you will need those quotes to defend your themes.
Extract Initial Code P1: “I feel like I should be available” Self-imposed availability P2: “Nobody said I had to... but I still feel guilty” Guilt without an external rule P3: “The laptop's always right there” Physical presence of work at home P4: “Work just lives in my house” Collapse of work-home boundary Stuck on how to analyse your data, or what to write for “data analysis”?
Upload or paste your transcripts and get codes and themes you can defend, with the quote behind each one.
Start free→3. Search for themes
Now shift from individual codes to broader patterns. Sort your codes into candidate themes, grouping ideas that pull several codes together and say something meaningful about your question. This is where sticky notes, tables, or a whiteboard earn their keep. Alternatively, if you coded your data with the tool above, it has already clustered the codes into candidate groups, so this phase starts from a sorted draft instead of a blank table. A theme is not just a topic word like “boundaries”. It is an interpretive claim. Looking at the four codes above, a candidate theme might be: “Remote work replaces external rules with self-policing.” Notice this is not “experiences of remote work,” a topic; it is a specific claim about what the codes, read together, mean.
4. Review the themes
Test each candidate theme twice. First, check the coded extracts inside it hang together: do all four excerpts actually support “self-policing,” or is P3 really about something else (the physical intrusion of the laptop, not guilt)? Second, check the theme makes sense against the full dataset, not just these four excerpts. Themes that overlap get merged; themes with only one thin quote behind them get cut or folded elsewhere. In this case, P3 might split off into a separate theme about the physical erosion of home-as-refuge, distinct from the guilt-driven self-policing in P1 and P2. You should end with a set that is distinct, each earning its place.
5. Define and name the themes
Write a short definition of what each theme is and is not. If you cannot describe a theme in a sentence or two, it is probably doing too much and needs splitting. Theme names should be sharp and readable. An examiner skimming your contents page should grasp the core analysis from the theme names alone. For the example: “Guilt-driven self-policing: participants regulate their own availability in the absence of any stated employer expectation, driven by an internalised sense that they should always be reachable.” That definition tells a reader exactly what is and is not covered.
6. Produce the report
Write the analysis so that the argument, the themes, and the evidence read as one piece. Each theme gets its definition, the quotes that support it, and your interpretation of what it means for the research question. The write-up is analysis, not a data dump: quotes illustrate the point you are making, they do not replace it.
From 2006 to reflexive thematic analysis
Since 2006 Braun and Clarke have refined how they talk about the method, and the current name for their version is reflexive thematic analysis. The phases are much the same. What changed is the thinking behind them. They now reject the idea that themes “emerge” from data on their own, as if waiting to be found. Themes are produced by a researcher who brings a perspective, and that perspective is a resource, not a contaminant to be scrubbed out.
The practical upshot: in reflexive thematic analysis you are not chasing a single “correct” coding that a second coder would replicate. You are building a considered interpretation and being transparent about your analytical role, which we explore in detail in our guide on writing reflexivity in thematic analysis. That is a different goal from a codebook approach, where a fixed set of codes is applied for consistency. If your project needs the codebook style instead, the difference between letting codes surface and applying a fixed scheme is discussed in our guide on inductive vs. deductive coding.
The Limitations of Braun and Clarke's Framework
No method is free of trade-offs, and Braun and Clarke themselves have written extensively about where reflexive TA gets misapplied or misunderstood. Knowing these limitations upfront will save you time and help you defend your methodological choices to an examiner.
It is genuinely slow to do by hand. Reading a dataset multiple times, generating dozens of initial codes extract by extract, then working iteratively through candidate themes across several review cycles is a manual, line-by-line process. There is no shortcut through phases one and two if you are doing this entirely yourself; the depth of engagement the method asks for takes real time, which is precisely why it is so often the stage that pushes a dissertation timeline back. Our step-by-step guide to coding interview transcripts makes that stage more manageable.
Its flexibility cuts both ways. The same theoretical openness that makes thematic analysis usable across realist and constructionist projects alike also means it offers far fewer built-in procedural guardrails than a more structured method like Framework Analysis. Braun and Clarke (2022a) note that thematic analysis is closer to a method than a full methodology, since it does not, by itself, tell you what your research question should look like, how to collect your data, or how large your sample needs to be. That openness is a strength for adaptability and a real risk for a first-time researcher who wants more scaffolding.
Reflexive TA deliberately rejects the reliability checks some reviewers expect. This is the limitation most likely to catch a student off guard. Braun and Clarke are explicit that reflexive TA does not use inter-rater reliability, consensus coding, or a fixed codebook checked by a second coder, because those checks assume there is a single correct, objectively accurate coding to converge on, an assumption reflexive TA's underlying values reject outright (Braun and Clarke, 2022a). If your supervisor or a reviewer comes from a background in coding-reliability approaches, they may expect exactly these checks. The method is not flawed for lacking them; the tension is that not every audience shares reflexive TA's assumptions, so this is worth stating explicitly in your methods section rather than leaving a reviewer to wonder why no second coder was involved.
Topic summaries are the most common failure mode, and they are easy to produce without noticing. Braun and Clarke (2022a) reviewed published papers citing their method and found this to be the single most frequent problem: researchers using reflexive TA's language while actually producing themes that are topic summaries, not meaning-based interpretations. It happens because a topic summary is genuinely easier to write. The next section shows exactly how to catch this in your own analysis before it costs you marks.
Mistakes that cost marks
Three problems show up again and again in student thematic analyses, and examiners spot them fast.
- Topic summaries dressed as themes. “Communication” is a bucket, not a theme. A theme makes a point: “communication broke down whenever roles were unclear”. If your theme name is a single noun, it is probably still a topic. Braun and Clarke (2022a) offer a useful test: if you could have named this theme before you analysed any data, or if it maps directly onto one of your interview questions, it is very likely a topic summary. A theme built the right way cannot be written in advance, because it depends on a pattern you found across the data, not a category you started with. Compare “Barriers to remote work” (a topic, could be named in advance) against “Remote work replaces external rules with self-policing” (a theme, only visible after reading the data).
- Quotes doing the analysis. A string of quotes under a heading is data, not interpretation. Say what the quote shows and why it matters for your question.
- Claiming themes “emerged”. In reflexive thematic analysis this phrasing signals you have misread the method. You constructed the themes; write as though you did.
Running the analysis without losing weeks
You can finish your thematic analysis on time without sacrificing academic rigour. The genuinely slow part, as covered above, is almost always phases one and two: reading everything closely and generating that first full pass of codes across every transcript. The thematic analysis tool on this site codes your text and clusters items into candidate themes with verbatim quotes, so you start phase three with a structured draft instead of a blank page. For a walkthrough of what to paste and how to review the output, see how to use the tool, or read our guide on finding themes across studies in a literature review. If you are coding a full set of interview transcripts rather than the four-excerpt example above, our interview coding tool is built for that; for a literature review instead of interviews, see the literature review synthesis tool. Coming from NVivo or MAXQDA and weighing whether you still need a licence for a single project, our NVivo alternative comparison covers the trade-offs.
Frequently asked questions
How many phases are in Braun and Clarke's thematic analysis?
Six: familiarising yourself with the data, generating initial codes, searching for themes, reviewing themes, defining and naming themes, and producing the report. Braun and Clarke set them out in their 2006 paper, and the phases remain the same in the later reflexive version. What changed over time is the theoretical thinking behind them, not the practical steps.
What's the difference between thematic analysis and reflexive thematic analysis?
Reflexive thematic analysis is Braun and Clarke's own update to their 2006 method, not a different method. The six phases stay the same; what changes is that themes are treated as built by the researcher's judgement rather than as patterns that “emerge” from the data on their own.
How many themes should a thematic analysis have?
There is no fixed number. Braun and Clarke's guidance is that themes should be distinct and each earn its place against the full dataset, which in practice usually lands somewhere between three and six for a typical student project. A study with twelve thin themes almost always has several that should be merged.
Is thematic analysis inductive or deductive?
It can run either way. Coding data-up with no predefined code list is inductive; coding against a fixed scheme you set before reading the data is deductive. Reflexive thematic analysis is usually run inductively, but nothing about the six phases requires it. See our guide on inductive vs. deductive coding for how to justify whichever you pick.
What are the main limitations of Braun and Clarke's thematic analysis?
The method is genuinely time-intensive to do manually, its theoretical flexibility means it offers fewer built-in procedural guardrails than more structured methods, and reflexive TA deliberately does not use inter-rater reliability or fixed codebooks, a choice consistent with its own values but sometimes at odds with what reviewers trained in other traditions expect (Braun and Clarke, 2022a).
References
- Braun, V. and Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), pp.77–101. doi:10.1191/1478088706qp063oa
- Naeem, M., Ozuem, W., Howell, K. and Ranfagni, S. (2023). A Step-by-Step Process of Thematic Analysis to Develop a Conceptual Model in Qualitative Research. International Journal of Qualitative Methods, 22. doi:10.1177/16094069231205789
- Ahmed, S.K., Mohammed, R.A., Nashwan, A.J., Ibrahim, R.H., Abdalla, A.Q., M. Ameen, B.M. and Khdhir, R.M. (2025). Using thematic analysis in qualitative research. Journal of Medicine, Surgery, and Public Health, 6, p.100198. doi:10.1016/j.glmedi.2025.100198
- Braun, V. and Clarke, V. (2022a). Toward good practice in thematic analysis: Avoiding common problems and be(com)ing a knowing researcher. International Journal of Transgender Health, 24(1), pp. 1–6. doi:10.1080/26895269.2022.2129597