Qualitative content analysis: method overview, steps and examples

Qualitative content analysis is a research method for systematically interpreting the meaning of qualitative data (text, audio, video, or images) by coding content into categories and analysing patterns across those categories. Unlike quantitative content analysis, which counts how often predefined terms or themes appear, qualitative content analysis focuses on interpreting what the content means in context. It is one of the most widely used methods in social science, media studies, health research, and applied business research.

How does qualitative content analysis differ from quantitative content analysis?

Both forms of content analysis begin by systematically coding data, but they diverge in what they count and what they interpret.

Quantitative content analysis applies predefined categories to a corpus and produces frequency counts. It can process very large amounts of content and produces results that are replicable by other researchers using the same coding scheme. The limitation is that it cannot interpret meaning: coding "health" as present or absent in a newspaper article does not tell you whether health is framed positively, as a threat, as a policy problem, or as a consumer aspiration.

Qualitative content analysis attends to meaning, context, and latent content (what the text implies as well as what it states). The same article might be coded for the framing of health, the agency attributed to different actors, or the assumptions embedded in the language used. The result is a richer, more interpretive analysis that requires researcher judgement at every step.

The two approaches are not mutually exclusive. Many studies combine qualitative content analysis of a smaller subset with quantitative coding of the full corpus, using the qualitative findings to interpret the quantitative patterns. This sits at the intersection with mixed methods research.

For a direct comparison of qualitative content analysis with thematic analysis, the other major method for analysing qualitative data, see content analysis vs thematic analysis.

What are the 3 main approaches to qualitative content analysis?

Hsieh and Shannon (2005) identify three distinct approaches, each suited to different research purposes.

1. Conventional qualitative content analysis

In conventional content analysis, the researcher does not apply predefined categories. Codes emerge from the data through an inductive process: the researcher reads through the material, identifies patterns, assigns preliminary codes, and groups codes into categories.

This approach is most appropriate when existing theory is limited or when the researcher wants to avoid imposing prior frameworks on the data. It resembles inductive coding in thematic analysis.

Academic example: A researcher studying how patients describe their experiences of a rare disease in online forums has no existing framework to apply. They read through 400 forum posts, coding passages inductively. Categories emerge: "searching for diagnosis," "negotiating with healthcare providers," "managing social stigma," and "finding community."

Business example: A product team analyses 600 customer support tickets for a newly launched product without any preconceived idea of what the main issues are. The inductive analysis surfaces three unexpected categories: configuration confusion during onboarding, specific integration failures, and a pricing model that customers find misleading.

2. Directed qualitative content analysis

Directed content analysis begins with a theory or prior research that provides the initial coding framework. The researcher applies predefined codes to the data, then looks for patterns within and beyond those codes.

This approach is appropriate when an existing theory needs to be validated, extended, or refined using new data. It resembles deductive coding in thematic analysis.

Academic example: A health researcher uses the Health Belief Model (perceived susceptibility, perceived severity, perceived benefits, perceived barriers, cues to action) as a directed framework to code interview transcripts on vaccine hesitancy. Passages are initially coded into these predefined categories, and the researcher then looks for evidence that challenges or extends the model.

Business example: A customer experience team uses the established VOC framework (functional needs, emotional needs, social needs) to code customer interview transcripts, then looks for which categories are most associated with churned versus retained customers.

3. Summative qualitative content analysis

Summative content analysis begins with identifying and counting specific words or content, then interprets the underlying context and meaning. It starts quantitatively but moves toward qualitative interpretation.

Example: A researcher counting mentions of "risk" in corporate annual reports, then interpreting how "risk" is contextualised: is it framed as financial risk, reputational risk, strategic opportunity, or regulatory burden? The count establishes prevalence; the qualitative analysis reveals meaning.

What are the steps in qualitative content analysis?

The process varies by approach, but the core steps are:

Step 1: Define the research question and select your data

The research question determines what kind of content you need and what analytical lens you will apply. Be specific: "How do employees describe work-life balance in performance reviews?" is a better foundation than "What do performance reviews say?"

Select a corpus (the full dataset to be analysed) that is appropriate to your question. Consider whether you are analysing all available content or a sample.

Step 2: Define the unit of analysis

The unit of analysis is the piece of content you will code. Common units include: individual words, sentences, paragraphs, full documents, or thematic units (a passage that contains a single coherent idea, regardless of its length).

Thematic units are the most common choice in qualitative content analysis because they preserve meaning better than word or sentence units.

Step 3: Develop your coding scheme

In conventional analysis, codes develop from the data through an iterative reading process. In directed analysis, codes come from the theoretical framework. In both cases, the coding scheme should be documented clearly enough that another researcher could apply it.

A good coding scheme includes: a code name, a definition, inclusion and exclusion criteria, and an example. This documentation supports consistency if multiple coders are working on the same corpus.

Step 4: Code the data

Apply codes systematically to the corpus. In manual coding, this typically means working through documents line by line, highlighting passages and assigning codes. Allow for the possibility that a passage may receive multiple codes, and document cases where a passage does not fit any existing code (these often become new codes or refine existing ones).

Tools for coding include QDA software (NVivo, MAXQDA, ATLAS.ti), AI-assisted platforms (Skimle), and basic tools like Excel or Word. For a comparison of options, see qualitative data analysis tools complete comparison.

Step 5: Analyse categories and interpret findings

Once coding is complete, analyse the patterns within and across categories. How frequently does each category appear? Which categories co-occur? Are there meaningful differences in how different subgroups use different categories?

Move from description (what the categories are and how often they appear) to interpretation (what the patterns mean in the context of your research question). This is the step where the qualitative dimension of content analysis asserts itself: categories and their patterns need to be interpreted, not just counted.

Step 6: Report findings

Present findings with evidence: direct quotes or examples from the data that illustrate each category, accompanied by your interpretation. Describe the analytical process transparently so that readers can evaluate the credibility of your findings.

Academic examples of qualitative content analysis

Media studies: A researcher analyses 200 news articles about climate change published across five European countries over ten years. Using directed content analysis with a risk-framing framework, they code articles for whether climate change is framed as an environmental risk, an economic risk, a national security risk, or a technological challenge. The analysis reveals that economic framing has become progressively more dominant in three countries while environmental framing has remained dominant in two.

Health research: A team analyses patient narrative essays submitted to a chronic pain programme, using conventional qualitative content analysis to identify the categories through which patients describe the impact of pain on identity, relationships, and work. The findings inform the design of a psychosocial support intervention.

Organisational research: A researcher analyses 15 years of CEO letters in annual reports from 20 companies, coding the language used around corporate responsibility. The summative analysis shows that mentions of "stakeholder" have increased by 340% since 2010; the qualitative analysis examines how the meaning and commitment implied by this language varies.

Business examples of qualitative content analysis

Customer feedback analysis: An e-commerce retailer codes 10,000 product reviews for six product lines, using a directed framework based on known quality dimensions (durability, value for money, ease of use, appearance, delivery, packaging). The qualitative interpretation of category content identifies that negative durability mentions cluster around specific product variants and appear disproportionately in reviews written after a supplier change, pointing directly to the root cause of the quality issue.

Policy and regulatory analysis: A legal team analyses 300 regulatory submissions and public comments using conventional qualitative content analysis to identify the main concerns raised by different stakeholder groups. Categories include compliance burden, competitive fairness, consumer protection, and implementation timeline. See stakeholder consultation analysis for policy teams for more on this application.

Employee communications analysis: An HR team analyses internal survey open-text responses and exit interview transcripts using a directed framework based on the organisation's known culture pillars, then compares the categories that emerge from leavers versus stayers.

How AI is expanding what qualitative content analysis can do

Larger samples

AI analysis platforms can process hundreds or thousands of documents in hours rather than weeks. A corpus that would have required a team of five researchers working for a month can be analysed in a day. This opens up qualitative content analysis for datasets that were previously only accessible through quantitative approaches: ten years of customer reviews, five years of social media posts, a national archive of policy documents.

The analytical implications are significant. When qualitative content analysis was constrained to small samples, findings were necessarily context-specific and difficult to generalise. With larger samples, the same method produces findings that are both richer (qualitatively) and more representative (statistically). This is the core of the argument for how AI is changing qualitative research.

Audio and video content

Traditional content analysis was largely limited to text because audio and video required manual transcription before they could be coded. AI transcription tools have changed this: audio and video can now be transcribed accurately and quickly, making audio and video content accessible to content analysis at scale.

This opens up entire categories of data that were previously too expensive to work with: recorded customer service calls, video testimonials, podcast content, video social media, broadcast media. A brand safety team can now analyse what is actually said in 5,000 hours of customer calls rather than relying on a manual sample of 50. See best AI transcription tools for researchers for an overview of the options.

Skimle supports audio and video upload directly: files are transcribed and then processed through the same structured AI analysis pipeline as text documents, with full traceability from every theme back to the specific timestamp in the original recording.

Multilingual analysis

Manual content analysis across languages requires either bilingual researchers or translation, both of which add cost and introduce reliability concerns. AI analysis platforms with multilingual support can analyse content in multiple languages within a single corpus, automatically detecting language and applying consistent analytical criteria. Skimle's multilingual analysis capability allows a single project to include documents in 30+ languages with consistent coding.

What AI does not change

AI does not eliminate the interpretive work at the heart of qualitative content analysis. The coding scheme still needs to be developed thoughtfully. The researcher still needs to evaluate whether AI-generated categories reflect the data or reflect patterns from the model's training. The findings still need to be interpreted in the context of the research question.

AI handles the volume; the researcher handles the meaning. For more on what responsible AI-assisted qualitative analysis looks like, see the AI qualitative data analysis checklist.

Frequently asked questions

What is the difference between qualitative content analysis and thematic analysis?

Both methods involve coding qualitative data and identifying patterns. The main differences are in origin and emphasis: qualitative content analysis developed in communication studies and is often used for systematic, structured analysis of defined corpora (media content, policy documents, archives). Thematic analysis developed in psychology and is more commonly used for interview and focus group data. Content analysis tends to be more structured in its coding procedures; reflexive thematic analysis (Braun & Clarke) is more explicitly interpretive. The methods overlap considerably in practice. See content analysis vs thematic analysis for a detailed comparison.

Can qualitative content analysis be done with quantitative methods?

Yes. Qualitative and quantitative content analysis are often combined. A common approach is to use qualitative coding to develop a category scheme, then apply that scheme quantitatively to a larger corpus to establish frequencies and test patterns. This integration is one form of sequential mixed methods research.

How many documents do you need for qualitative content analysis?

Sample size in qualitative content analysis depends on the research question and the diversity of content you need to capture. For exploratory analysis of a defined corpus (a company's customer reviews, a set of policy documents), include the full corpus if manageable. For research on a broader phenomenon (how a topic is covered in national media), purposive sampling with a rationale for the inclusion criteria is appropriate. Samples of 20-100 documents are common in published academic studies; AI-assisted analysis makes it feasible to work with several hundred to several thousand.

Is qualitative content analysis inductive or deductive?

Both are possible, and this is one of the method's strengths. Conventional qualitative content analysis is inductive (categories emerge from the data). Directed qualitative content analysis is deductive (categories come from theory). Many studies combine both: start with a theoretical framework, apply it to the data, and remain open to categories that the framework does not anticipate.

Ready to run qualitative content analysis on hundreds of documents without months of manual coding? Try Skimle for free. Upload text, audio, or video content and see AI-assisted systematic coding with full traceability from every category back to the source.

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