How to Create a Qualitative Study: 7 Steps With Examples

Learning how to create a qualitative study starts with one idea: you are collecting explanations, not counts. Instead of asking how many people chose an option, you ask why they chose it, what they weighed, and what almost changed their mind.

Done well, a qualitative study shows the reasons behind your numbers. Done poorly, it produces a pile of opinions nobody trusts.

In this blog, we’ll walk through seven steps, worked examples, quality checks, and the mistakes that sink most studies.

What is a qualitative study?

A qualitative study is a research project that collects and interprets non-numerical data, such as interview transcripts, open-ended survey answers, field notes, and video, to explain how and why people think, feel, or behave the way they do.

The goal is understanding, not statistical proof. Researchers read the material closely, tag it with codes, and group those codes into themes. A theme is a repeated idea that answers the research question.

Three traits set qualitative work apart:

• Exploratory: It works well when you do not yet know what to measure.
• Context-rich: It records the situation around an answer, not just the answer.
• Interpretive: A researcher reads, codes, and explains the data, so method and transparency matter.

How is a qualitative study different from similar research terms?

People mix up several close terms. A qualitative study is the whole project, from question to report. A method, such as an interview, is only one tool inside it.

If your goal is to compare groups with statistics, read our guide on qualitative vs. quantitative research before you commit to a design.

When should you run a qualitative study instead of a survey?

Run a qualitative study when you need reasons, language, or ideas you cannot predict. Run a quantitative study when you already know the options and need to size them.

Use this decision rule:

• You are exploring a new problem.
  Start qualitative. You cannot write good multiple-choice options for something you do not understand yet.
  
• A number surprised you.
  A drop in NPS or a spike in cancellations tells you what happened. Interviews tell you why.
  
• You are designing a survey.
  Pilot interviews reveal the words customers actually use.
  
• You need to prove how big something is.
  Go quantitative, or use mixed methods.

Stakes matter too. If a wrong guess costs little, five conversations may be enough. If the decision is expensive, plan a larger and more varied sample.

Step-by-step guide to creating a qualitative study

Every strong study follows the same seven steps, whatever the topic. Skipping one rarely saves time, because the gap shows up later as weak data or a report nobody acts on.

Step 1: Define the research question and objectives

Write one open question about one group of people, plus the decision the answer will inform. A good question starts with “how” or “why” and names who you are studying.

• Weak: “Why do customers like our product?” It is vague, and it invites flattery.
• Strong: “What makes small-business owners in the US keep or cancel after the 30-day trial?”
• Objective: “Identify the three biggest reasons for cancellation so product can prioritize fixes next quarter.”

Test your question with one rule: if every possible answer would lead to the same action, rewrite it. For more help, see our guide to writing qualitative research questions.

Step 2: Choose a research design and method

Pick the method that fits the question, not the one that is easiest to run. Each method captures a different kind of evidence.

The method is how you collect data. The design is the logic of the whole study, covered in the next section on choosing a design. Our guide to the qualitative interview covers the most common method in more detail.

Step 3: Write the discussion guide

A discussion guide is the outline of questions and probes a moderator follows. It keeps sessions consistent without turning them into a script.

Build it in four blocks:

• Warm-up: Easy questions about the person’s background and routine.
• Core questions: Five to eight open questions tied to your objective.
• Probes: Follow-ups like “Can you give me an example?” and “What happened next?”
• Wrap-up: “Is there anything I should have asked?”

Avoid leading questions. “Don’t you find checkout confusing?” plants the answer. “Walk me through the last time you checked out” does not. Run the guide with two people first and fix whatever confuses them.

Step 4: Recruit participants and decide the sample size

Recruit on purpose, not at random. Purposive sampling means choosing people who have direct experience with the topic you are studying.

Write a short screener, a set of qualifying questions, so you reach the right people. Then decide how many. Research by Guest and colleagues found that saturation, the point where new interviews add no new themes, appeared within 12 interviews in a fairly uniform group. A systematic review by Hennink and Kaiser of 17 empirical studies found saturation within 9 to 17 interviews or 4 to 8 focus groups when the population was similar and the aims were narrow.

Broader aims, several segments, or multi-country work need more. As a practical habit, recruit a few extra people to cover no-shows.

Step 5: Collect the data

Collect data in a way that protects participants and keeps your own opinions out of the room. Good fieldwork is mostly listening.

• Get informed consent in writing before you start. In the US, federally funded human subjects research follows the Common Rule, and the same consent habits are wise for any study.
• Record with permission, then write short notes right after each session.
• Stay neutral. Do not nod at answers you like or correct answers you dislike.
• Review notes after every three to five sessions, and adjust the guide if new topics appear.

Remote sessions work well for geographically spread groups. Our guide to online qualitative research covers the setup.

Step 6: Analyze the data

Analysis turns raw talk into themes. The most widely used approach is thematic analysis, and the phases give you a clear path:

1. Read the transcripts until you know them well.
2. Generate initial codes, which are short labels for ideas in the text.
3. Search for themes by grouping related codes.
4. Review the themes against the full data set.
5. Define and name each theme.
6. Write up the findings with supporting quotes.

AI tools can suggest first-pass codes and summaries quickly. A person should still review every theme, because software misses sarcasm, culture, and context. Our guide to qualitative data analysis compares the main techniques.

Step 7: Report the findings

A report earns its value when someone acts on it. Lead with the answer to the research question, then support it.

Use this structure: the question, the method and sample, four to six themes, and a clear recommendation. Under each theme, state the claim, add two short quotes, and explain what it means for the decision. Be honest about limits, such as who you did not reach.

A word cloud shows which words appeared often. It does not show what they meant. Use it as a visual, not as evidence.

Which qualitative research design should you choose?

Choose the design that matches the kind of answer you need. A design is the overall logic of the study, while a method is only the data collection tool inside it.

• Phenomenology explores what an experience feels like to the people who lived it. Example question: “What is it like to manage a chronic illness while working full time?”
• Grounded theory builds an explanation from the data instead of testing one you already hold. It suits topics where no good theory exists. Coding happens in stages, from open labels to a connected explanation.
• Ethnography studies a group in its natural setting, often through observation over weeks. Example question: “How do nurses actually hand off patients between shifts?”
• Case study examines one person, team, or event in depth, using several sources of evidence.
• Narrative research collects life or career stories and studies how people make sense of them. Example question: “How do first-generation college students describe their path to graduation?”

When two designs seem to fit, pick the one that answers your question with the least data.

What do real qualitative studies look like?

Qualitative studies look different by industry, but each ties a specific question to a specific decision. These are illustrative scenarios built on common US use cases.

Worked example: A cancellation study

A subscription software team wants to know why trial users cancel. They write one question, recruit 12 recent cancelers through a screener, and run 30-minute interviews.

During coding, three themes repeat: setup took longer than expected, a key integration was missing, and the price was unclear after the trial. Seven of the 12 mention setup. The team fixes onboarding first, then retests with a short survey to confirm the change worked.

What are the most common mistakes in a qualitative study?

Most failed studies break in the design phase, not the analysis. Knowing the usual failure points lets you avoid them early.

• Leading questions: They produce agreement, not insight.
• A sample that is too broad: Mixing very different groups hides every pattern.
• No pilot: One confusing question can ruin a full set of interviews.
• Cherry-picking quotes: Choosing only the lines that support your hunch is bias, not analysis.
• Reporting percentages: “58% of participants” implies a statistic the sample cannot support. Say “seven of 12” instead.
• No link to a decision: A study with no owner and no next step gets filed and forgotten.
• Trusting AI coding blindly: Unreviewed machine themes can be confident and wrong.

How do you protect participants in a qualitative study?

You protect participants by getting informed consent, limiting what you collect, and securing what you keep. Interviews often reveal personal details, so ethics is part of the design, not an afterthought.

Tell people what the study is for, how long it takes, and that they can stop at any time. Replace names with codes in transcripts, and store recordings in a protected location with limited access. Delete raw recordings on the schedule you promised.

Extra care applies in some cases. Health information may fall under HIPAA, the US federal law on protected health information. Studies with children, patients, or employees reporting on their own managers need stronger safeguards. If you upload transcripts to an AI tool, check where the data goes and say so in your consent form.

How do you measure the quality of a qualitative study?

You measure quality by trustworthiness, since statistical tests do not apply. Researchers commonly use four criteria from Lincoln and Guba, summarized in this quality criteria guide.

Add one stopping rule for sample size: stop recruiting when the last two or three sessions add no new codes. For reporting, the 32-item COREQ checklist lists what interview and focus group reports should disclose.

How can QuestionPro support a qualitative study?

QuestionPro Research Software covers the stretch between recruiting and analysis, which is where small teams usually lose time. Four capabilities map directly to the steps above.

• Recruiting and ongoing contact: QuestionPro Communities supports forums, reminders, member rewards, and 50+ languages, so you can keep a pool of willing participants.
• Open-ended data at scale: Use QuestionPro Survey Software to add open-ended and video questions, and Text AI to detect topics and sentiment in the answers.
• Video evidence: AI video analysis summarizes themes and sentiment from video interviews and group sessions.
• Cleaner data: Data Quality AI flags bot, AI-generated, and gibberish open-ended responses before you code them.

Review every AI-suggested theme yourself. The software speeds up reading, but the interpretation stays with you.

Before you launch: A final thought on creating a qualitative study

A strong qualitative study is small, focused, and tied to a decision. The method matters less than the clarity of the question and the honesty of the analysis. Most of what makes a study credible happens before the first interview.

Check three things before you start:

• Can you state the decision this study will inform?
• Does your sample match the people who made that decision?
• Do you have a plan to show how you reached each theme?

Frequently Asked Questions (FAQs)