Skip to main content

How to do thematic analysis: A step-by-step guide

Thematic analysis is one of the most widely used ways to develop patterns of shared meaning in qualitative data. This guide walks through Braun and Clarke's six phases, the different families of thematic analysis, and what separates a strong theme from a weak topic summary.

Prepared by:
YouReply Qualitative content team
Published:
Last updated:
{minutes} min read
10 min read

What is thematic analysis?

Braun and Clarke (2006) define thematic analysis as a method for identifying, analyzing and reporting patterns of meaning across a dataset. Much of the paper's influence comes from giving a clear procedure and a shared vocabulary to a method that was already widely used but rarely explained in detail.

An important feature of the method is that it is not tied to a particular theoretical or epistemological framework. That flexibility is an advantage, but it also brings responsibility. Researchers need to state the assumptions they work with, whether they look for themes at the explicit (semantic) or underlying (latent) level of meaning, and whether they aim for a broad description of the whole dataset or a detailed account of one particular aspect.

In their later work (2019, 2021), Braun and Clarke call their approach “reflexive thematic analysis.” In this framing, the researcher's subjectivity is not a bias to be removed but a resource for analysis; themes do not simply emerge from the data but are developed by the researcher through engagement with it. For the broader picture, see what is qualitative data analysis?

There is no single thematic analysis: three families

Quite different practices are carried out under the name “thematic analysis.” Braun and Clarke (2021) roughly distinguish three families to make sense of this variety. The table below is a simplified summary of their distinction; it is not meant to draw hard lines but to help you clarify, in your methods section, which logic you are working with.

Families of thematic analysis (summarized from Braun and Clarke's distinction)
FamilyCore logicRole of the codebookInter-coder agreement
Coding reliabilityConsistency and replicability of coding come first; themes are often set early.A structured codebook prepared in advanceUsually assessed by measuring agreement.
Codebook approachesA structured codebook is combined with a qualitative, interpretive stance.A codebook developed and documented during analysisNot required; use varies by project.
Reflexive thematic analysisThe researcher's interpretation is the engine of analysis; themes are developed after coding.No fixed codebook; codes evolve over the process.Measuring agreement does not fit the logic of the approach.

Guest, MacQueen and Namey (2012), for their part, offer an account of thematic analysis for applied research that emphasizes systematic codebook development and transparent procedures, especially in team projects. Whichever family you choose, writing only “thematic analysis was conducted” in your methods section is not enough; state the approach you followed and the quality criteria that fit it.

Inductive, deductive and hybrid coding

In inductive (data-driven) coding, codes are derived directly from the data without fitting them into a preexisting frame. In deductive (theory-driven) coding, an existing theory, prior research or your research questions guide the coding. Braun and Clarke (2006) discuss these as different orientations to analysis; in practice, most studies sit somewhere in between.

Fereday and Muir-Cochrane (2006) propose that middle ground as an explicitly defined hybrid approach: a codebook is prepared in advance from the research questions and theoretical framework, but new, data-driven codes can be added for meanings that do not fit the template. This suits studies that want to connect with theory without missing what the data has to surprise them with. For how to write code definitions, see the codebook guide.

Fictional exampleExample (fictional)

A fictional study explores teachers' experiences of remote teaching. Drawing on their theoretical framework, the researcher prepares a codebook with starting codes such as “autonomy” and “competence” (deductive). While reading the interviews, they notice that many teachers describe how an open camera brings home life into the lesson; since that meaning has no place in the template, they add a new code, “privacy entering the classroom” (inductive). In the end, codes from both sources are developed into themes together.

The six phases, step by step

The six phases Braun and Clarke (2006) proposed are restated in their later work, especially the 2021 practical guide, with some changes in naming. The newer names stress that themes are not ready-made things waiting to be found in the data but structures the researcher develops. The phases look sequential, yet moving back to earlier phases as the analysis progresses is expected.

The six phases: 2006 naming and later naming
Phase2006 namingLater namingWhat you do
1Familiarizing yourself with your dataFamiliarizing yourself with the datasetRead and reread all the data and note initial ideas.
2Generating initial codesCodingSystematically code meanings relevant to the research question.
3Searching for themesGenerating initial themesCluster codes that share meaning under candidate themes.
4Reviewing themesDeveloping and reviewing themesCheck themes against coded excerpts and the full dataset.
5Defining and naming themesRefining, defining and naming themesWrite each theme's central idea and boundaries, and give it a meaningful name.
6Producing the reportWriting upPresent themes in an analytic narrative, connected to quotes and the literature.
  • 1. Familiarization: Before you start coding, read every transcript at least once, and listen to recordings if you have them. Note what catches your attention and your own reactions in separate memos.
  • 2. Coding: Code the entire dataset systematically. Codes can capture both what is said explicitly and the assumptions underneath; a single excerpt can belong to more than one code.
  • 3. Initial themes: Cluster codes and ask whether a shared idea sits behind each cluster. Themes are drafts at this point, and codes you set aside are not lost.
  • 4. Developing and reviewing: First check whether the excerpts under a theme tell a coherent story, then whether the theme fits your reading of the whole dataset. Most decisions to merge, split or drop themes are made here.
  • 5. Refining and naming: Write a definition of a few sentences for each theme. If you cannot write the definition, the theme is not clear yet.
  • 6. Writing up: Writing is part of the analysis, not something tacked on at the end; as you build the narrative, you rethink the order of themes and how they relate.

It is more productive to read the six phases as a framework that supports analytic thinking than to apply them as a mechanical recipe: what matters is not ticking a phase off as “done” but asking whether its purpose has been met.

What is a theme, and what is a topic summary?

One of the most common problems in thematic analysis is that the “themes” are really topic headings. Braun and Clarke (2019, 2021) describe a theme in reflexive thematic analysis as a pattern of shared meaning united by a central organizing concept. They set this apart from topic summaries, which gather the different, even contradictory, things participants said about a topic under one heading.

Topic summary versus theme
FeatureTopic summaryTheme
NamePoints to an area (“Relationship with managers”).Makes a claim (“Being visible is equated with being trustworthy”).
ContentEverything said about the topic, contradictions includedMeanings that share a common idea
Question it answersWhat did participants say about this topic?What do the data tell us about the research question?
Where it comes fromOften mirrors the interview questions.Brings together meanings from different parts of the dataset.
Fictional exampleExample (fictional)

In a fictional employee experience study, suppose everything said by people who love flexibility and by people worn out by it is collected under “Flexible work.” That is a topic summary. On rereading the excerpts, you see that many participants, even those who like flexibility, complain about having to decide on their own when the workday ends. A theme such as “The job of drawing boundaries has been handed to the employee” names that shared meaning and makes an analytic claim.

A theme's importance is not measured only by how many participants it appears in. Braun and Clarke (2006) stress that what matters is whether a theme captures something important in relation to the research question. Even so, using phrases such as “some participants” or “most participants” consistently in your report helps readers.

Theme maps and writing up

A thematic map is a working tool that shows themes, subthemes and the relationships between them visually. Braun and Clarke (2006) find maps especially useful in the reviewing phase. Rather than drawing a map once and leaving it, updating it as you merge or split themes also documents how your analysis changed.

  • Put the research question at the center and the themes around it.
  • Show relationships between themes (condition, tension, consequence) with an arrow and a short label.
  • Build a theme × participant table; you will quickly see whether a theme rests on only one or two people's accounts.
  • Keep dated versions of the map.

In the report, this structure works well for each theme: a short introduction to the theme's central idea, analytic paragraphs that develop the claim, and selected quotes that support it, each tagged with a participant code. Quotes do not speak for themselves; you need to interpret why each was chosen and what it shows. Braun and Clarke (2006) likewise recommend using extracts as part of an analytic narrative that goes beyond description.

Quality checklist

Braun and Clarke (2006) offer a 15-point checklist for good thematic analysis. Nowell and colleagues (2017) adapt trustworthiness criteria to each phase of thematic analysis and illustrate what can be done at each one. The questions below are a practical summary inspired by these sources, not a reproduction of them.

  • Were transcripts checked against the original recordings before coding began?
  • Was the whole dataset coded with equal care, or were some interviews only skimmed?
  • Does each theme carry a shared central idea, or is it a topic heading?
  • Are themes coherent internally and distinct enough from one another?
  • Are analytic claims supported by quotes, and are the quotes interpreted?
  • Are cases that do not fit or that contradict a theme addressed in the report?
  • Are codebook changes and theme decisions documented in dated memos?
  • Does the methods section clearly state the approach followed and the researcher's position?

Rather than answering these questions once at the end, revisit them regularly, especially in phases four and five. If you keep your answers as short memos, you can show which decision you made and why when a reviewer or advisor asks.

Common pitfalls

Most problems in thematic analysis come not from the technique itself but from a gap between the logic of the method and how it is applied. You can also read the mistakes below as a list to run through while reviewing your analysis; for each one there is usually a safeguard you can put in place early on.

  • Turning interview questions into themes: Themes that map one to one onto question headings are usually topic summaries.
  • Too many themes: Dozens of themes can signal that the analysis has not yet been condensed; consider merging close themes or turning some into subthemes.
  • Trying to prove themes with numbers alone: Frequency tables can add context, but they do not show a theme's analytic value on their own.
  • Reducing the method to a label: Citing Braun and Clarke while following procedures that contradict their logic, for example treating an agreement coefficient as proof of quality in reflexive analysis.
  • Writing that themes “emerged”: This language hides the researcher's decisions; describe how you developed the themes instead.
  • Treating an AI suggestion as a finished theme: Automated suggestions can be a starting point, but suggestions that are not grounded in the source text, or that do not go beyond a topic summary, should not become themes without the researcher's judgment. See ethics in AI-assisted qualitative analysis for more.

Thematic analysis in YouReply Qualitative

In YouReply Qualitative, themes are kept as a separate layer above codes, and a code can belong to more than one theme. A theme × participant matrix shows which participants a theme draws on, and a code × demographic attribute matrix shows differences between groups. Merge and split operations on codes can carry a rationale note, and reflexive memos can be attached to a theme or an excerpt.

If you turn AI on for a study, you can get theme and pattern suggestions; suggestions stay pending until you accept, edit or reject them, and they do not enter matrices or reports before that. The product does not compute an inter-coder agreement coefficient. For this step of the workflow, see the themes stage.

Summary

  • Thematic analysis is a flexible method for developing patterns of shared meaning; that flexibility comes with the responsibility to state your assumptions.
  • Coding reliability, codebook and reflexive approaches rest on different logics of quality; say which one you followed in your methods section.
  • The six phases look sequential but are recursive; you return to the data again and again, especially while reviewing themes.
  • A good theme is not a topic heading but a pattern of meaning united by a central idea that makes a claim.
  • Theme maps, theme × participant tables and dated decision memos make the analysis more traceable.
  • AI suggestions can be a starting point; developing and naming themes is the researcher's work.

References

  1. Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77–101. https://doi.org/10.1191/1478088706qp063oa
  2. Braun, V., & Clarke, V. (2019). Reflecting on reflexive thematic analysis. Qualitative Research in Sport, Exercise and Health, 11(4), 589–597. https://doi.org/10.1080/2159676X.2019.1628806
  3. Braun, V., & Clarke, V. (2021). Thematic analysis: A practical guide. SAGE.
  4. Fereday, J., & Muir-Cochrane, E. (2006). Demonstrating rigor using thematic analysis: A hybrid approach of inductive and deductive coding and theme development. International Journal of Qualitative Methods, 5(1), 80–92. https://doi.org/10.1177/160940690600500107
  5. Guest, G., MacQueen, K. M., & Namey, E. E. (2012). Applied thematic analysis. SAGE. https://doi.org/10.4135/9781483384436
  6. Nowell, L. S., Norris, J. M., White, D. E., & Moules, N. J. (2017). Thematic analysis: Striving to meet the trustworthiness criteria. International Journal of Qualitative Methods, 16(1). https://doi.org/10.1177/1609406917733847

Code your first interview today

The free plan carries a pilot study from start to finish. No credit card required.