Qualitative data is non-numerical information that describes qualities, meanings, and experiences rather than quantities. It takes the form of text, speech, images, or observation — interview transcripts, focus group recordings, open-ended survey responses, documents, field notes. Unlike quantitative data, it is not measured or counted; it is interpreted. Analysis involves identifying patterns of meaning rather than calculating statistics.
What is qualitative data?
Qualitative data captures the qualities of something: what it is, what it means, how people experience it, why they behave as they do. A satisfaction rating of 3.2 is quantitative data. The interview in which a customer explains that the product meets their needs but makes them feel like they are constantly fighting the interface is qualitative data. The number measures; the words explain.
The term covers a wide range of data types — any non-numerical information gathered for research purposes. What unites them is that analysis requires interpretation, not arithmetic.
Qualitative data is the primary material of qualitative research, though it also appears in mixed methods studies alongside quantitative data. It is used across academic disciplines and in applied research contexts including user research, market research, HR, policy analysis, and strategy consulting.
What are the types of qualitative data?
Qualitative data comes in several forms, each suited to different research contexts:
| Type | Examples | How it's collected |
|---|---|---|
| Text data | Interview transcripts, survey responses, documents, social media posts | Interviews, surveys, document retrieval |
| Audio data | Recorded interviews, focus groups, meeting recordings | Recording during data collection |
| Video data | Observation recordings, video diaries, user testing sessions | Video recording in field or lab |
| Images | Photographs, drawings, visual representations | Photography, participant-generated images |
| Field notes | Researcher observations, reflective memos, contextual notes | Written during or after observation |
| Artefacts | Objects, documents, organisational materials | Collection from field sites |
In most qualitative research, text is the primary data type — either collected as text (documents, written responses) or converted to text through transcription (interviews, focus groups). Analysis tools, including AI-assisted tools like Skimle, work primarily with text.
How does qualitative data differ from quantitative data?
| Qualitative data | Quantitative data | |
|---|---|---|
| Form | Text, speech, images, observation | Numbers, ratings, counts |
| What it captures | Meaning, experience, context | Frequency, magnitude, relationships |
| Analysis | Interpretation and pattern-finding | Statistical analysis |
| Sample logic | Small, purposive | Large, representative |
| Output | Themes, categories, narratives | Averages, correlations, significance tests |
| Strengths | Depth, nuance, discovery | Precision, generalisability |
| Limitations | Not statistically generalisable | Misses meaning and motivation |
Neither type is inherently superior. The question is which type of data can best answer your research question. "How many of our customers are dissatisfied?" is a quantitative question. "Why are customers dissatisfied, and what do they most want changed?" is a qualitative question.
Many research designs use both. A company measuring NPS scores (quantitative) alongside open-ended comments about the reasons for those scores (qualitative) is collecting both types. The numbers tell you where the problem is; the words tell you what it is.
Where does qualitative data come from?
Recommended reading
The renaissance of qualitative research
Interviews
One-to-one conversations that produce transcripts. A 60-minute interview generates approximately 8,000-10,000 words of text. Interviews are the most widely used source of qualitative data because they allow for depth, probing, and the exploration of individual experience.
For guidance on collecting interview data well, see how to conduct effective business interviews and how to write a perfect interview guide.
Focus groups
Group discussions that produce a single, richer transcript shaped by the interaction between participants. A two-hour focus group with eight participants generates significant data volume, but the dynamics between participants — agreement, disagreement, elaboration — are part of what the data captures.
Open-ended survey responses
Text responses to survey questions that do not constrain the answer format. A survey of 500 employees with two open-ended questions can produce 50,000+ words of qualitative data. For guidance on writing good open-ended questions, see open-ended questions in research.
Documents and existing texts
Policy papers, corporate reports, social media posts, news articles, historical records. Document analysis uses existing material rather than generating new data through direct engagement with participants.
Observation and field notes
Notes taken during fieldwork — recording what the researcher sees, hears, and experiences in a setting. Field notes capture context, behaviour, and atmosphere that interviews often cannot.
How do you analyse qualitative data?
Qualitative data analysis involves systematically working through the data to identify patterns of meaning. The process is interpretive: the analyst constructs an account of what the data means in relation to the research question.
The most widely used approaches are:
Thematic analysis. Coding segments of text and constructing themes from the patterns in those codes. Braun & Clarke's six-phase thematic analysis framework is the most widely cited guide. See what is thematic analysis? for a complete explanation.
Content analysis. Counting and categorising content in text. More systematic and less interpretive than thematic analysis; produces findings that can be expressed quantitatively (how many times a theme appears). See content analysis vs thematic analysis for a comparison.
Grounded theory. A methodology for building theory inductively from data, through successive rounds of coding. See grounded theory methodology.
Interpretive phenomenological analysis (IPA). An intensive analysis of lived experience, working with small samples. See interpretive phenomenological analysis.
Discourse analysis. Analysis of how language constructs meaning in a social context — not just what is said, but how the saying of it does social work.
The analysis always involves qualitative coding: labelling segments of text to mark what they are about, and then finding patterns in those labels across the dataset.
What are the main challenges of working with qualitative data?
Volume. A study of 30 interviews produces approximately 300,000 words of text. Reading, coding, and analysing that volume manually is time-consuming. AI tools like Skimle substantially reduce the coding burden by processing the full dataset and generating initial themes, which researchers then review and refine.
Rigour. Because qualitative analysis is interpretive, it requires systematic methods and transparent reporting to be credible. Two analysts working with the same data may identify somewhat different themes — this is expected in qualitative research, but the analytical process needs to be documented so readers can evaluate it. See AI qualitative data analysis checklist for a practical audit framework.
Transcription. Audio data must be converted to text before analysis. Manual transcription takes roughly four hours per hour of audio. AI transcription tools have reduced this to minutes, at accuracy rates above 95% for clear audio. See best AI transcription tools for research.
Confidentiality. Qualitative data often contains sensitive personal information. Transcripts may need to be anonymised before analysis, storage, or sharing. Skimle's anonymisation feature automates pseudonymisation of participant names and identifying details.
How does AI change how qualitative data is analysed?
Recommended reading
Hallucinations, limited context and black boxes? Not with Skimle
The bottleneck in qualitative research has historically been analysis. AI addresses this directly.
Modern AI-assisted qualitative analysis tools can:
- Code a dataset of 50 interview transcripts in minutes, rather than the weeks that manual coding takes
- Identify themes across the full dataset without being limited by researcher memory or fatigue
- Surface patterns across metadata variables (comparing themes across departments, time periods, or participant groups) that would require significant manual effort to identify
- Maintain a traceable link from every theme back to the source passages that support it, enabling verification
The important distinction is between AI as a coding engine and AI as an analytical replacement. AI handles the mechanical process of coding and aggregation. The interpretive work — constructing themes that genuinely explain what the data means, identifying the significance of unexpected findings, making judgements about ambiguous passages — remains with the researcher.
For academic researchers, transparency about AI assistance is increasingly expected in methods reporting. For market researchers and HR and people teams working with large volumes of qualitative data, AI analysis is often the only practical way to systematically analyse the full dataset rather than working from a sample.
For a complete guide to AI-assisted analysis, see AI in qualitative research.
Frequently asked questions
Is qualitative data reliable?
Reliability in qualitative research is a different concept than in quantitative research. It does not mean that different researchers would produce identical findings from the same data — some variation is expected in interpretive analysis. It means that the analysis is consistent, documented, and defensible: the researcher can show their working, and a reader can evaluate whether the interpretations follow from the data. Rigorous qualitative research is reliable in this sense.
Can qualitative data be converted to quantitative data?
Yes, in a limited sense. After coding qualitative data, you can count how many participants or responses fall into each theme — which introduces a quantitative layer. Sentiment analysis assigns positive/negative scores to text. But the conversion loses the depth and nuance that makes qualitative data valuable. Use quantification of qualitative data as a supplement to interpretation, not a replacement for it.
What is the difference between qualitative data and anecdotal evidence?
Qualitative data is collected and analysed systematically, using defined methods and documented processes. Anecdotal evidence is informal: a story, an example, a recollection that is not part of a systematic study. The difference matters for what claims you can make. Systematically collected qualitative data can support research claims; anecdotes cannot.
How do you store qualitative data securely?
Interview transcripts and focus group recordings contain personal information. Storage should comply with relevant data protection regulations (GDPR in Europe). Best practices include: storing data on encrypted systems, limiting access to authorised team members, using pseudonymised identifiers rather than real names in analytical files, and having clear data retention and deletion policies. Skimle stores data in compliance with GDPR and supports pseudonymisation through its anonymisation feature.
Ready to analyse your qualitative data? Try Skimle for free — AI-assisted qualitative analysis with full traceability from themes to source quotes.
Related reading:
- What is qualitative research?
- Qualitative coding: what it is and 5 approaches explained
- Thematic analysis: steps, types and examples
About the authors
Henri Schildt is a Professor of Strategy at Aalto University School of Business and co-founder of Skimle. He has published over a dozen peer-reviewed articles using qualitative methods, including work in Academy of Management Journal, Organisation Science, and Strategic Management Journal. His research focuses on organisational strategy, innovation, and qualitative methodology. Google Scholar profile
Olli Salo is a former Partner at McKinsey & Company where he spent 18 years helping clients understand the markets and themselves, develop winning strategies and improve their operating models. He has done over 1000 client interviews and published over 10 articles on McKinsey.com and beyond. LinkedIn profile
Sources
- Using thematic analysis in psychology — Braun & Clarke, Qualitative Research in Psychology (2006)
- Integrating quantitative and qualitative research: how is it done? — Bryman, Qualitative Research (2006)
- Participant Observation as a Data Collection Method — Kawulich, Forum: Qualitative Social Research (2005)
