Blog › Guide

How to analyse qualitative data: a step-by-step guide to interviews, thematic analysis, NVivo and Excel

By Eliakim Anzizi, PhD · 6 November 2026

Qualitative data is information in words, not numbers: interview transcripts, focus group notes, open answers in a questionnaire, diaries, documents or field notes. To analyse it, you read it closely, label the parts that matter to your research question (coding), group those labels into patterns (themes), and then explain what the patterns mean, with quotes as evidence.

This guide shows how to analyse qualitative data step by step. It focuses on thematic analysis, the method most students and researchers start with, and covers interview data, open survey answers, and how to do the work in NVivo or Excel. (If you are in the US or Canada, you may know this as how to analyze qualitative data. The steps are the same.)

What is qualitative data analysis?

Qualitative data analysis is the process of turning a large amount of text, audio or images into a small number of clear, well-supported findings. It answers questions such as how and why: how people experience something, why they make the choices they do, or what a policy means to the people it affects.

Unlike statistics, there is no single formula. The analysis depends on your judgement. That is why examiners and reviewers want to see exactly how you moved from raw data to themes. A clear, documented process is what makes qualitative findings trustworthy. If your study also has numbers, our guide to data analysis in research covers the quantitative side.

Main methods for analysing qualitative data

Choose the method in your proposal, and make sure it fits your research question. The most common are:

Thematic analysis. Finds and reports patterns (themes) across the data. Flexible and widely used, so it suits most dissertations and applied studies.
Content analysis. Codes text into categories and often counts how often they appear. Useful for documents, media and open survey answers.
Grounded theory. Builds a new theory from the data through repeated rounds of coding and comparison.
Interpretative phenomenological analysis (IPA). Explores in depth how a small number of people make sense of a lived experience. Common in psychology and health.
Narrative analysis. Looks at the stories people tell and how they tell them.
Discourse analysis. Studies how language is used to create meaning, identity or power.
Framework analysis. Sorts data into a matrix of cases and themes. Popular in applied policy and health research, especially in teams.

Each method has its own literature and rules. If your department expects a named method, cite a key source for it and follow its steps. The rest of this guide uses thematic analysis, because its steps are a good base for most of the others.

How to analyse qualitative data: thematic analysis step by step

A widely cited approach to thematic analysis comes from Virginia Braun and Victoria Clarke (2006), who describe six phases. The steps below follow that outline, with practical notes added. The phases are not a straight line; you will move back and forth between them.

Step 1: Prepare and organise your data

Transcribe audio recordings, or check transcripts made by software or a transcriber against the recording. Decide how much detail you need: word-for-word with pauses, or a clean version. Remove or replace names and other details that could identify participants, as your ethics approval requires. Give each file a clear name, such as P01_interview_2026-03-12, and keep a simple table of participants and their key characteristics.

Step 2: Get to know the data

Read every transcript at least once, ideally twice, before you code anything. Listen to recordings again if you can. Make short notes on first impressions, surprises and ideas. This step is easy to rush, but it is where you start to see what the data is really about.

Step 3: Generate initial codes

A code is a short label for a piece of data that is relevant to your research question. Work through the data line by line or paragraph by paragraph, and attach codes to meaningful sections.

Inductive (bottom-up) coding lets codes come from the data itself.
Deductive (top-down) coding starts from a list of codes based on your theory, framework or interview questions.
Many studies use both: a starting list, plus new codes for anything unexpected.

Keep a codebook: each code’s name, a one-line definition, when to use it and an example quote. Code the whole data set, not just the parts that fit your expectations. Keep the codes close to what participants actually said.

Step 4: Search for themes

A theme is a broader pattern of meaning that brings several codes together and says something important about your question. Group related codes, and look for a central idea that links them. A mind map or table helps. Some codes will become themes, some sub-themes, and some will be set aside.

A theme is not the same as a topic. "Workload" is a topic. "Feeling that workload is invisible to managers" is a theme, because it makes a claim about the data.

Step 5: Review and refine themes

Check each theme against the coded extracts, then against the whole data set. Ask: is there enough data to support it? Is it clearly different from the other themes? Does it tell a coherent story? Merge, split or drop themes as needed. Note cases that do not fit; they are part of the finding, not a problem to hide.

Step 6: Define and name the themes

Write two or three sentences for each theme that say what it is and what it is not. Give it a short, clear name. Decide how the themes relate to each other and to your research questions.

Step 7: Write up the findings

Present each theme with a short explanation and well-chosen quotes from different participants. Do not let quotes speak for themselves: after each one, explain what it shows. Link the themes back to your research questions and, in the discussion, to the literature. A table of themes and sub-themes with a sample quote for each is a clear way to give an overview.

How to analyse qualitative data from an interview

Interview data follows the same steps, with a few extra points:

Analyse as you go. If possible, read and note each interview soon after it happens. Early ideas can sharpen later interviews.
Do not simply code by question. Your interview guide gives a useful starting structure, but themes often cut across several questions.
Look within and across cases. Understand each participant’s account as a whole, then compare across participants.
Keep context. Note who said what (by code number), and any context that changes the meaning of a quote.
Record your decisions. Keep a short memo or reflexive journal of why you created, merged or dropped codes. It becomes your audit trail.

An illustrative example

This example is illustrative and not taken from a real study. Imagine a student interviewing 12 secondary school teachers about online teaching. One teacher says: "By the time I had uploaded the notes, half the class had no data to download them." The student codes this data costs and access gaps. Across the interviews, similar codes appear: unreliable power, shared devices, parents’ phones. Grouped together, they form a theme such as "Learning depends on the household, not the lesson". The write-up explains the theme, gives two or three quotes from different teachers, and notes the teachers whose experience was different.

How to analyse qualitative data from a questionnaire

Open-ended survey questions often produce many short answers. A practical approach:

Export all answers to a spreadsheet, one row per respondent, one column per open question.
Read a sample of answers and draft a list of codes.
Code every answer, allowing more than one code where needed.
Group codes into categories or themes.
Where it helps, report how many respondents mentioned each category, alongside example quotes. Be clear that these counts describe your respondents, not the wider population, unless your sample allows that.

How to analyse qualitative data in NVivo

NVivo is software for organising and coding qualitative data. It does not find themes for you, but it makes it much easier to manage many transcripts, retrieve every quote under a code, and show your process. A typical workflow:

Create a project and import your transcripts, documents or survey file.
Set up cases for participants and add attributes such as age group, role or location, so you can compare groups later.
Code the data by selecting text and coding it to codes (NVivo calls them nodes in some versions). Build a hierarchy of themes and sub-themes as you go.
Write memos to record ideas and decisions.
Explore with queries, such as a coding query or a matrix coding query that shows how themes appear across groups.
Export your codebook and coded extracts to support the write-up.

Menus and names change between versions, so check the help pages for the version your university licenses. Other packages, such as ATLAS.ti and MAXQDA, work in a similar way. In your methodology, name the software and say what you used it for. Our guide to dissertation methodology shows how to describe this clearly.

How to analyse qualitative data in Excel

Excel (or Google Sheets) works well for small or medium data sets and for open survey answers, and almost everyone already has it. A simple set-up:

One row per data segment: a sentence, paragraph or answer.
Columns for participant ID, question, the quote, code 1, code 2, theme and a notes column.
Use filters to pull together every segment under one code or theme.
Use a pivot table or COUNTIF to see how codes spread across participants or groups.
Keep the codebook on a separate sheet, and use a drop-down list (data validation) so codes are spelt the same each time.

The limits: Excel becomes hard to manage with long transcripts or many codes, and it is easy to lose the context around a quote. For a large project, dedicated software usually saves time.

How to show your analysis is trustworthy

Qualitative research is judged on trustworthiness rather than statistical reliability. You can strengthen it by:

keeping an audit trail of codes, memos and decisions;
having a second person code part of the data and discussing differences;
checking interpretations with participants, where appropriate;
including cases that do not fit the main pattern;
describing your setting and participants in enough detail for readers to judge how far findings may apply elsewhere;
reflecting honestly on how your own position may shape the analysis.

Common mistakes to avoid

Themes that are just interview questions. Reorganise by meaning, not by the order of your guide.
Too many themes. A few well-supported themes are stronger than a long list.
Quotes without analysis. Always explain what each quote shows.
Counting as if it were statistics. Frequency can describe your sample, but a theme’s importance is about meaning.
A vague method section. "Data were analysed thematically" is not enough. Name the approach, the steps, the software and how you checked quality.
Leaving analysis to the last week. Coding and reviewing themes takes longer than most people plan for.

Qualitative analysis checklist

Method chosen and justified in the proposal
Transcripts checked and anonymised
All data read before coding
Codebook with definitions and examples
Whole data set coded
Themes reviewed against the data and named clearly
Quotes from a range of participants, each explained
Cases that do not fit reported
Audit trail of decisions kept
Software and steps described in the methodology

Want an expert to review your qualitative analysis?

Erudite Digital Team is a research consultancy led by Eliakim Anzizi, PhD. We analyse qualitative data, including interview and focus group transcripts, and quantitative data in SPSS, NVivo, Excel and Stata. We also review coding frameworks and findings chapters for clarity, and explain results in plain English. See our data analysis and proofreading and editing services, or start with our free services. Still planning your study? Read our guide on how to write a research proposal, and use our editing and proofreading checklist before you submit.

Sources

Braun, V. and Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77–101.

Frequently asked questions

What are the steps in analysing qualitative data?

Prepare and organise the data, read it closely, code it, group codes into themes, review and name the themes, then write up the findings with quotes and explain what they show.

How do you analyse qualitative data from interviews?

Transcribe and anonymise the interviews, read them all, code meaningful sections, build themes across participants and check them against the data. Report each theme with quotes from different participants.

Is NVivo necessary for qualitative analysis?

No. NVivo helps you organise and retrieve coded data, especially in large projects, but the analysis is your own thinking. Small studies can be analysed well by hand, in Word or in Excel.

Can I analyse qualitative data in Excel?

Yes. Put one data segment per row, with columns for participant, quote, codes and theme. Use filters and pivot tables to group and compare. It works best for small data sets and open survey answers.

What is the difference between a code and a theme?

A code labels a single piece of data. A theme is a broader pattern of meaning that brings several codes together and says something about your research question.

Want your qualitative analysis checked? Send your research questions and a short description of your data on WhatsApp or by email. We reply with a clear plan. See our data analysis service.

Write it Right, Choose Erudite.