What is Thematic Analysis in Qualitative Research?

TL;DR

• Thematic analysis is a qualitative research method for identifying, analyzing, and interpreting patterns of meaning (themes) in qualitative data: interview transcripts, focus groups, survey responses, and customer feedback.
• Braun and Clarke's six-phase thematic analysis process (familiarization, coding, searching for themes, reviewing themes, defining themes, reporting) remains the gold standard. The phases are recursive, not linear.
• The most common mistake in published thematic analysis: treating themes as topic summaries ("participants talked about X") instead of meaningful patterns ("participants framed X as a personal failing rather than a structural issue").
• Trustworthiness comes from documentation: reflexive memos, audit trails, and coding decision logs satisfy reviewers more than inter-coder reliability scores alone.
• AI-assisted thematic analysis accelerates the coding process for large datasets while preserving human control over theme development and interpretation.

Thematic analysis is the most cited qualitative research method in the world. But being widely used and being used well are two very different things.

Braun and Clarke's 2006 paper has been cited over 200,000 times across social science, health research, psychology, and education. Don't believe us? Braun and Clarke themselves have documented the evidence. The majority of published thematic analyses contain methodological shortcomings: themes described as "emerging" from data (they don't: the researcher constructs them), reflexivity documentation absent, quality criteria missing or borrowed from incompatible theoretical frameworks.

The method is used widely but practiced rigorously far less often than it should be. Thematic analysis is a flexible approach that helps turn messy qualitative data into a structured narrative, but that flexibility becomes a problem when qualitative researchers treat it as an excuse to skip the rigor.

This guide covers how to conduct thematic analysis in qualitative research with the depth the method deserves. It addresses Braun and Clarke's thematic analysis process in detail, trustworthiness criteria, how thematic analysis compares to other qualitative research methods (grounded theory, IPA, content analysis, discourse analysis), a worked example showing the entire analysis process from raw data to defined themes, and the reporting mistakes that undermine otherwise solid research.

What Is Thematic Analysis in Qualitative Research?

Thematic analysis is a qualitative data analysis method for identifying, analyzing, and interpreting recurring patterns of meaning (themes) across qualitative data. It works with interview transcripts, focus groups, open-ended survey responses, customer feedback, field notes, textual data, and visual data. The output is a set of defined themes, each capturing a distinct meaningful pattern that answers the research questions.

What separates thematic analysis from simple categorization is the interpretive layer. Counting how many participants mentioned "wait time" is categorization. Identifying that participants consistently framed long wait times as evidence of organizational disrespect: that's a theme. The pattern of meaning, not the topic itself, is what makes thematic analysis a method rather than a sorting exercise.

In simple terms: thematic analysis asks "what does this data mean?" rather than "what does this data contain?"

The method was formalized by Braun and Clarke in 2006 as a standalone qualitative analysis method with its own procedural thematic framework and quality criteria. Thematic analysis generally follows a systematic six-phase framework popularized by Braun and Clarke, which has since been refined in their 2022 Thematic Analysis: A Practical Guide and the 2024 RTARG reporting guidelines.

What makes the method especially versatile is its epistemological flexibility. Thematic analysis works within various theoretical frameworks: realist, constructionist, and critical. This flexibility across diverse research contexts is why it remains the default qualitative research method for applied thematic analysis in fields ranging from healthcare to CX to policy research.

Inductive vs. Deductive Thematic Analysis: Choosing Your Approach

Before conducting thematic analysis, you need to decide your analytical direction. This choice shapes every phase of the analysis process that follows.

Inductive thematic analysis is a bottom-up approach where themes emerge directly from the qualitative data, making it particularly useful for exploring new or under-researched areas without preconceived notions. You read the data, generate initial codes based on what you find, and let the broader themes develop from the ground up. The research questions guide what data you collect, but they don't predetermine what themes you'll find.

Deductive thematic analysis is a top-down approach that begins with predefined themes or existing theories, allowing researchers to test specific hypotheses or apply existing concepts to their data. You start with a thematic framework drawn from existing literature and code the data against it. Deductive analysis works well when you're testing whether established patterns hold in a new context.

Most real qualitative research projects use a blend. You might start inductively (letting the data speak) and then check your developing themes against existing theories. Or you might start deductively (applying a known framework) and remain open to inductive themes that don't fit the framework. Both thematic analysis approaches can work alongside quantitative data: use thematic analysis to explain why your survey scores changed, while the scores tell you how much.

Braun and Clarke's Six-Phase Thematic Analysis Process

Braun and Clarke's thematic analysis process provides a systematic and rigorous approach to qualitative data analysis. Here's what each phase involves, where qualitative researchers commonly get it wrong, and what the analysis process looks like in practice.

Phase 1: Familiarization with the Data

The thematic analysis process begins with data collection and then immersion: reading and re-reading your entire data set before coding anything. Most researchers underinvest here. They read transcripts once and move straight to identifying themes.

The practical minimum: read each data item at least twice. During the first read, note initial observations without categorizing. During the second, start noting meaningful patterns and potential themes. If you're analyzing interview transcripts, listen to the recordings at least once. Tone, pauses, and emphasis carry meaning that textual data alone doesn't convey.

Phase 2: Generating Initial Codes

Generating initial codes is the second step in thematic analysis, where researchers assign labels to specific portions of text that capture the essence of the ideas expressed in the qualitative data. After generating initial codes, researchers collate these codes with supporting data, grouping related excerpts to identify connections and potential themes.

Codes aren't just labels applied to data. They're the researcher's first analytical decisions about what matters. Labeling a segment "work pressure" describes it. Coding it as "normalizing overwork" interprets it. For the coding process to produce rich theme development, codes should be analytically rich, not merely descriptive.

A practical tip: code at a fine grain initially (more initial codes than you think you need). You can always merge codes later. Splitting codes retroactively requires recoding data you've already processed. For a deeper walkthrough of coding mechanics, see our guide to thematic coding.

Phase 3: Searching for Themes

This is the phase where most quality issues originate. Researchers often struggle with moving from codes to themes, as basic tools may not support the organization of complex data, leading to difficulties in identifying overarching themes. The task: start grouping codes into potential themes by looking for meaningful patterns across the coded data.

The common mistake: treating themes as topic summaries. "Participants talked about work-life balance" is a topic. "Participants framed work-life balance as a personal failing rather than a structural issue" is a theme. In simple terms: if your themes could be generated by a statistical analysis tool, they're topics, not themes.

Phase 4: Reviewing Themes

Check candidate themes against the coded extracts and the entire data set. Merge, split, or discard themes that don't hold together. Reviewing themes is where you verify that themes accurately represent the data rather than the researcher's assumptions.

The most common Phase 4 failure: keeping too many themes. A useful test: if you can't articulate the distinction between two potential themes in one sentence, they're probably one theme. Braun and Clarke recommend creating a candidate thematic map at this stage: examining data visually to see how different codes relate to broader themes and to the research questions.

Phase 5: Defining and Naming Themes

Refine what each theme captures and write a clear definition. Naming themes well matters: compare "Communication Issues" (a topic label) with "The Silence Spiral: How Unacknowledged Feedback Teaches Customers to Stop Talking" (a pattern of meaning). The second tells you what was found. The first tells you nothing a keyword search couldn't.

Phase 6: Producing the Report

The final step in the thematic analysis process involves writing a report that connects the identified themes back to the research questions, illustrating findings with direct quotes from the qualitative data. A coherent narrative weaves themes together in relation to the research questions, supported by data extracts. Common failures: listing thematic analysis findings without analytic narrative, selecting quotes that illustrate rather than support the argument, and failing to connect themes back to the research questions or existing literature.

Worked Example: The Thematic Analysis Process From Raw Data to Themes

Theory is useful. Seeing the method applied to raw data is better. Here's a compressed walkthrough showing conducting thematic analysis on customer feedback responses. The same research process applies to interview transcripts, focus groups, or any qualitative data.

The Data: 5 Open-Ended Responses

1. "Sarah on chat was incredibly helpful, but I had to explain my issue three separate times before getting transferred to someone who could actually help."
2. "I love the product but every time I need support it's a runaround. I've started just Googling the answer instead of contacting you."
3. "Fast resolution this time, but only because I specifically asked for the same agent who helped me last month. Shouldn't have to do that."
4. "Your knowledge base is outdated. I found the answer on Reddit faster than on your help center."
5. "The agent was great once I finally reached one. But your automated system made me repeat information I'd already typed in the chat form."

Phase 1: Familiarization Notes

After reading the entire data set twice, initial observations (no codes yet):

• Praise for individual agents appears alongside frustration with the system
• Repetition of information is a recurring pattern
• Customers are finding workarounds rather than relying on official support
• The frustration seems directed at process, not people

Phase 2: Generating Initial Codes

Notice: "Forced repetition as disrespect of effort" is an analytic code. "Customer had to repeat" would be descriptive. The analytic version tells you what the repetition means to the customer.

Phase 3: Grouping Codes into Candidate Themes

Grouping codes into broader themes produces two candidates:

Candidate Theme A: "The competence gap: individual agents are capable, but the system around them creates friction that undermines their effort."

Candidate Theme B: "Learned bypass: customers develop workarounds to avoid a support system they've stopped trusting."

Phase 4: Reviewing Themes Against the Entire Data Set

Theme A holds across all five responses: every one contains "the person was fine, the system wasn't." Theme B holds for responses 2, 3, and 4 (customers creating their own solutions). Both themes are distinct. They stay as defined themes.

Phase 5: Theme Development and Definition

Theme 1: "The Competence-System Gap": Customers consistently distinguish between individual agent capability and structural barriers created by the systems agents work within. Frustration is directed at the support process (transfers, information loss, outdated resources), not the people.

Theme 2: "Learned Bypass": When the support system repeatedly fails, customers develop independent workarounds (Googling answers, requesting specific agents by name, using community forums). These workarounds signal eroding trust, not satisfaction with the alternative.

What This Example Shows

Five responses. Two defined themes. Neither would surface from a CSAT score. Neither would surface from a word-frequency count. They surfaced because the researcher read the raw data closely enough to identify patterns of meaning, not just what was mentioned.

Reflexive Thematic Analysis vs. Codebook Approaches

One of the most common methodological confusions among qualitative researchers is conflating reflexive thematic analysis with codebook thematic analysis. They share a name but differ fundamentally in how themes are generated, how quality is assessed, and what role the researcher plays in the analysis process.

Maintaining analytical rigor is a challenge in thematic analysis, as researchers must ensure that their interpretations are grounded in the data. If you're doing reflexive thematic analysis and reporting Cohen's kappa scores, you're mixing incompatible theoretical frameworks. If you're doing codebook analysis and reporting reflexive memos, you're adding valuable rigor but not replacing the reliability measures your method requires. Braun and Clarke have explicitly called this conflation inappropriate.

The distinction isn't a quality hierarchy. A multi-site clinical study coding patient feedback across 20 hospitals needs codebook consistency. A single qualitative researcher exploring how first-generation college students navigate barriers needs reflexive thematic analysis's interpretive depth. Neither approach is superior.

For CX and product teams analyzing customer feedback at scale, codebook approaches usually fit best: consistent categorization, trackable themes, reproducible results. Purpose-built thematic analysis software and qualitative data analysis software effectively automate this approach, applying a persistent taxonomy across every response. Paired with sentiment analysis, theme-level detection becomes even more powerful. The interpretive layer (what do these themes mean for your roadmap, your NPS program, your retention strategy?) still requires human judgment.

Trustworthiness in Thematic Analysis: What Reviewers Look For

Lincoln and Guba's trustworthiness framework (credibility, transferability, dependability, confirmability) is the most widely referenced quality framework for qualitative analysis. Here's what each criterion means for thematic analysis specifically, and what practical steps satisfy it.

Credibility: Are the themes a plausible interpretation? Credibility comes from prolonged engagement with the qualitative data (multiple readings, iterative coding), triangulation where possible (comparing themes across multiple data sources or qualitative methods), and respondent validation. In simple terms: would someone who examined the same data recognize your themes as reasonable? The research questions should clearly connect to the themes you present.

Transferability: Can readers assess whether findings apply to their context? Provide enough thick description (participants, setting, data collection methods) that readers can judge for themselves. Describe your data collection approach, sample characteristics, and the research context in enough detail for future research to assess relevance.

Dependability: Is the research process documented clearly? An audit trail satisfies this: your coding decisions, theme development, reflexive memos, and changes to your analytic approach should all be documented. If a reviewer asks "how did you get from these codes to these themes?", your documentation should answer without explanation from memory.

Confirmability: Are findings grounded in the data? For reflexive thematic analysis, this means documenting how your perspective shaped the work. The goal is transparency, not objectivity.

A 2025 ScienceDirect review proposed a 16-item quality checklist for thematic analysis quality. The three items qualitative researchers miss most: documenting why codes were merged, recording when theme boundaries changed, and verifying that every theme traces back to at least three independent data segments. The most common quality failures are documentation failures, not analytical ones.

Practical Trustworthiness: What Takes 30 Minutes vs. 30 Hours

Trustworthiness isn't an all-or-nothing investment. Some practices take minimal time and produce significant quality gains. Others are resource-intensive and appropriate only for published research.

30-minute investments (do these for every qualitative research project):

• Write a one-paragraph positionality statement describing your background and assumptions before you begin the coding process
• Keep a running memo where you note coding decisions and evolving interpretations throughout the analysis process
• After constructing themes, write one sentence per theme explaining why it matters to your research questions

Half-day investments (do these for published or high-stakes research):

• Create a formal audit trail: version your codebook, document every merge and split decision, save intermediate theme maps showing how different codes relate to broader themes
• Conduct peer debriefing: have a colleague review your themes against a sample of the qualitative data and challenge your interpretations
• Check theme coverage: verify every theme is supported by extracts from at least 3 independent data items across the entire data set

Multi-day investments (for PhD dissertations and funding applications):

• Member checking: share themes with participants and incorporate their responses into further analysis
• Triangulation: compare themes across data sources, qualitative methods, or analysts
• Full reflexive journal: detailed memos for every coding session documenting analytical development

The 30-minute investments transform the quality of any thematic analysis project. Skipping them is the single most common reason otherwise solid qualitative analysis gets criticized during review.

Common Challenges in Conducting Thematic Analysis

Even experienced qualitative researchers encounter predictable obstacles in the thematic analysis journey. Naming these challenges helps you plan for them rather than discover them mid-project.

Data overwhelm. One common challenge is the sheer volume of qualitative data and initial codes, which can make it difficult to identify patterns and move toward broader themes. The fix: code in passes rather than trying to capture everything at once. First pass for descriptive codes, second for analytic codes, third for grouping codes into potential themes.

From codes to themes. Researchers often struggle with the leap from codes to themes. Maintaining analytical rigor during this transition means constantly asking: "Is this a topic label or a pattern of meaning?" If you can't articulate what the theme tells you about the research questions that a simple category count couldn't, it's still a topic.

Team coordination. Coordinating a team effort in thematic analysis can lead to inconsistencies in the coding process and theme development, especially when team members have different interpretations. The solution: for codebook approaches, establish clear coding guidelines and check inter-coder agreement regularly. For reflexive thematic analysis, have one lead analyst construct themes and use the team for peer debriefing.

Knowing when to stop. There's no formula for thematic saturation. For most qualitative research projects, saturation arrives when new data produces codes that map to existing themes without revealing new ones. Document the point where you decided further analysis wouldn't change your thematic framework.

Thematic Analysis vs. Grounded Theory, IPA, Content Analysis, and Discourse Analysis

Qualitative researchers often choose thematic analysis by default. Here's how it compares to other qualitative research methods, and when each approach fits.

Choose thematic analysis when your research questions ask "what patterns exist?" rather than "what theory explains this?" (grounded theory), "what is it like to experience this?" (IPA), "how frequently does this appear?" (content analysis), or "how does language construct meaning?" (discourse analysis). Thematic analysis is a flexible method that can be adapted to diverse data sources and research contexts without adhering to a rigid theoretical framework.

Thematic analysis is widely applied in healthcare research to explore patients' experiences, healthcare provider perspectives, and healthcare policy analysis. It also plays a critical role in policy analysis by extracting key themes from policy documents and public opinion. In market research and CX, applied thematic analysis helps extract comprehensive understanding from consumer feedback, product reviews, and focus groups.

For CX researchers specifically, thematic analysis offers the most practical methodology because it maps naturally to feedback workflows: code qualitative data, construct themes, connect themes to business outcomes. Grounded theory's theoretical sampling doesn't align with fixed survey datasets. IPA's small-sample focus doesn't scale. Thematic analysis handles both depth (reflexive approach on selected data) and breadth (codebook approach on large datasets from diverse data sources).

From our research: when we tested reflexive thematic analysis on a customer feedback dataset, the analyst's CX background surfaced themes like "effort fatigue" that rule-based coding missed entirely. The theme wasn't in any codebook because nobody had named it yet. This illustrates how reflexive thematic analysis's interpretive lens can identify patterns that codebook approaches applied to the same raw data simply can't detect. The thematic analysis journey from data to insight isn't just about identifying themes: it's about the quality of interpretation that shapes those themes.

8 Common Thematic Analysis Reporting Mistakes

Braun and Clarke have documented the most common reporting failures across hundreds of published thematic analyses. These undermine the entire analysis process and are the reason reviewers reject otherwise competent qualitative research.

1. "Themes emerged from the data." This is the most criticized phrase in thematic analysis reporting. Themes don't emerge passively. The researcher constructs them through iterative engagement with the qualitative data. Saying themes "emerged" obscures the researcher's active analytical role. Instead: "Through iterative coding and reviewing themes, three patterns were developed that capture [description]."

2. Themes as topic summaries. A section headed "Work-Life Balance" that simply catalogs what participants said is a topic summary, not a product of developing themes with interpretive depth. Themes should capture meaningful patterns that answer your research questions. "The impossibility narrative: how participants framed work-life balance as a personal failure" is a theme. "Participants discussed work-life balance" is a topic heading.

3. Quotes replacing analysis. Quotes should support an analytical argument, not replace it. A common pattern: state a theme, provide three quotes, move to the next theme. The analytic narrative explaining why these quotes matter, how they connect, and what they reveal about the research questions is what's missing. The quote should never do the interpretive work the researcher should be doing.

4. Missing reflexivity. For reflexive thematic analysis, the researcher's positionality statement, reflexive memos, and documentation of how interpretation evolved are required elements. Many published analyses claim reflexive thematic analysis but provide no evidence of reflexive practice in their research process.

5. Incompatible quality criteria. Reporting inter-coder reliability for reflexive thematic analysis, or claiming reflexive depth for codebook TA, signals methodological confusion. Identify which thematic analysis approach you're using and apply the matching quality framework.

6. Insufficient data engagement. If your thematic analysis findings could be replicated by a keyword search (listing what participants mentioned most frequently), the analysis process lacks the interpretive depth the method requires.

7. Under-specifying the approach. "We used thematic analysis (Braun and Clarke, 2006)" is insufficient. Specify: reflexive or codebook, inductive or deductive thematic analysis, semantic or latent, and which theoretical frameworks guide your analysis. Reviewers increasingly expect this level of methodological specificity when examining qualitative research.

8. Too few data extracts. Each theme should be supported by extracts from at least three independent data segments from the entire data set. A theme supported by quotes from only one or two participants may represent individual experience rather than a meaningful pattern across the qualitative data.

Quality Checklist for Thematic Analysis in Research

Use this checklist before submitting or publishing your thematic analysis. It synthesizes criteria from Braun and Clarke's quality framework (2021), the RTARG reporting guidelines (2024), and the ScienceDirect 16-item review. This checklist isn't exhaustive, but it covers the items most likely to be flagged by reviewers and examiners. A qualitative research project that satisfies all 12 items demonstrates the methodological care that distinguishes rigorous thematic analysis from procedural coding with a thematic label.

1. Is the thematic analysis approach named clearly (reflexive, codebook)?
2. Is the epistemological position stated?
3. Are the six phases described showing recursive engagement, not just sequential completion?
4. Does each theme capture a meaningful pattern (not a topic)?
5. Is the researcher's positionality documented?
6. Is the coding process documented (merges, splits, renames)?
7. Does every theme trace to at least three independent data segments?
8. Are quotes used to support arguments, not replace them?
9. Do themes connect back to the research questions?
10. Are quality criteria appropriate to the approach?
11. Is the analytic narrative interpretive rather than descriptive?
12. If team-based: are coder agreement processes documented?

Scaling Thematic Analysis: From Research Projects to Continuous Feedback Programs

Thematic analysis was developed for qualitative research projects with bounded datasets. But the method's logic (code qualitative data, identify patterns, interpret meaning) applies equally to ongoing customer feedback, support tickets, and product reviews.

The challenge is scale. A qualitative researcher analyzing 30 interview transcripts can apply the analysis process manually. A CX team processing 2,000 support comments a month can't. But the potential for meaningful insights doesn't disappear at volume. At scale, you need codebook consistency: a defined taxonomy applied reliably across the entire data set. Qualitative data analysis software can auto-discover themes from the raw data (like a researcher would in Phase 3), then apply them persistently across further analysis.

What AI handles well: the initial coding process, consistent taxonomy application, trend detection, and identifying themes across diverse data sources. What AI doesn't replace: the interpretive judgment that turns "billing complaints increased 40%" into a comprehensive understanding of whether the billing change was poorly communicated, poorly designed, or both.

For teams moving from manual qualitative data analysis to AI-assisted analysis, the transition works best when the human sets the thematic framework and the technology scales execution. Zonka Feedback's AI-powered thematic analysis applies the codebook approach at scale: auto-discovering themes, applying a persistent taxonomy, and detecting experience signals at both the response level and the theme level. Schedule a demo to see how it handles your data.