What Is Thematic Analysis?
Thematic analysis is a qualitative research method for identifying, organizing, and interpreting patterns of meaning (themes) within a dataset. It involves systematically coding qualitative data, interview transcripts, open-ended survey responses, social media posts, or observational notes, and grouping those codes into broader themes that capture something important about the research question. Unlike methods tied to specific theoretical frameworks, thematic analysis is flexible enough to work across epistemological positions, which is why it's become one of the most commonly used qualitative methods in social science, health research, and market research.
Why Thematic Analysis Matters in Research
Thematic analysis turns unstructured qualitative data into organized findings that teams can act on. Without a systematic approach to coding, qualitative analysis often drifts into cherry-picking quotes that confirm what the researcher already believed. According to a review in Qualitative Research in Psychology, Braun and Clarke's 2006 framework has been cited over 120,000 times, making it the most widely adopted approach to qualitative analysis in published research.
How Thematic Analysis Works
Braun and Clarke's Six Phases
Virginia Braun and Victoria Clarke published the definitive framework for thematic analysis in 2006. It's structured in six phases, though the process is recursive, you'll move back and forth between phases as your understanding develops.
Phase 1: Familiarization. Read through the entire dataset at least twice. Take notes on initial impressions. Resist the urge to start coding immediately. You need to understand the data as a whole before you break it into parts. If you're working with interview recordings, transcribe them verbatim first.
Phase 2: Generating initial codes. Work through the data systematically and assign codes to meaningful segments. A code is a brief label that captures what a data segment is about. For example, a respondent saying "I stopped using the app because it crashed every time I tried to check out" might receive codes like "app crash," "checkout failure," and "churn trigger."
Code everything that's potentially relevant. At this stage, more codes are better than fewer, you can always consolidate later.
Phase 3: Searching for themes. Review your codes and start grouping them into candidate themes. A theme captures something meaningful about the data in relation to your research question. It's not just a topic or a domain, it requires an organizing concept. "Technology problems" is a topic. "Technical friction erodes trust and drives churn" is a theme.
Phase 4: Reviewing themes. Check your candidate themes against the coded data and the full dataset. Does each theme have enough supporting data? Are the boundaries between themes clear? This phase often involves splitting themes that are too broad, merging themes that overlap, or discarding themes that don't hold up under scrutiny.
Phase 5: Defining and naming themes. Write a brief description of each theme, what it captures, what it doesn't, and how it relates to other themes. Give each theme a concise, informative name. "Price sensitivity among first-time buyers" is more useful than "Theme 3."
Phase 6: Writing up. Present your themes with supporting data extracts (quotes) and connect them to your research question. The write-up should tell a coherent story, not just list themes in sequence.
The Coding Process
Coding can be done manually (highlighting and annotating transcripts), with software (NVivo, Atlas.ti, Dedoose), or with AI-assisted tools that generate initial codes for human review.
There are two main coding approaches:
- Inductive coding starts from the data itself. You let codes emerge from what participants say rather than applying a pre-existing framework. This is useful for exploratory research.
- Deductive coding starts from a theory or framework. You create codes in advance based on what you expect to find, then look for them in the data. This works well when you're testing existing models or comparing against known categories.
Most practical projects use a hybrid approach, starting with some deductive codes based on the research objectives while remaining open to inductive codes that emerge from the data.
Thematic Analysis vs. Grounded Theory
Both methods involve coding qualitative data, but they differ in scope and ambition.
| Aspect | Thematic Analysis | Grounded Theory |
|---|---|---|
| Goal | Identify and describe themes | Build new theory |
| Theoretical commitment | Flexible, atheoretical | Specific methodology |
| Sampling | Typically fixed | Theoretical sampling (evolves) |
| Output | Themed findings | Theoretical framework |
| Complexity | Moderate | High |
| Best for | Applied research, market research | Academic theory development |
When to Use Thematic Analysis
- Analyzing open-ended survey responses to understand the reasoning behind quantitative ratings
- Processing interview or focus group transcripts from customer research, UX studies, or employee feedback programs
- Reviewing customer support tickets or product reviews to identify recurring pain points and feature requests
- Conducting content analysis on competitor messaging, social media conversations, or community forums
- Synthesizing findings from multiple qualitative sources into a cohesive set of insights
Common Mistakes to Avoid
- Treating themes as the same as topics or codes: a theme requires an interpretive concept, not just a category label
- Skipping Phase 4 (reviewing) and reporting your first-pass themes without checking them against the full dataset
- Using data prevalence as the only criterion for theme importance, a theme mentioned by two participants can be more analytically significant than one mentioned by twenty
- Failing to document the coding process, which makes it impossible for others to evaluate or replicate your analysis
- Starting with too few codes and forcing data into predetermined categories before you've understood its range
How Quali-Fi Supports Thematic Analysis
Quali-Fi's Research plan ($1,061/month) includes AI-powered open-end coding that generates initial codes from survey responses and interview transcripts, then lets analysts review, edit, and reorganize them into themes. The platform tracks code frequency and co-occurrence across respondent segments using built-in cross-tab tools. For large-scale projects requiring custom taxonomies and multi-coder reliability checks, the Intelligence tier ($2,750+/project) provides dedicated analyst support.
Frequently Asked Questions
How many themes should a thematic analysis produce?
There's no fixed number. Most studies produce between four and eight themes. Fewer than three usually means you haven't analyzed deeply enough. More than ten often means your themes overlap or you're reporting codes rather than themes. The right number is however many it takes to tell a complete, coherent story about your data.
Can thematic analysis be used with quantitative data?
Thematic analysis is designed for qualitative data. However, you can use it alongside quantitative analysis in mixed-method studies, for example, running thematic analysis on open-ended responses while analyzing closed-ended questions statistically. Some researchers also quantify themes (counting how many participants mentioned each one), though this should supplement rather than replace the qualitative interpretation.
Do I need special software for thematic analysis?
No. You can do thematic analysis with printed transcripts and highlighters, or with a simple spreadsheet. Software like NVivo or AI-powered tools speed up the process and make it easier to manage large datasets, but they don't change the analytical logic. The quality of thematic analysis depends on the researcher's thinking, not the tool.