How to easily code your interviews
Learn interview coding with examples of open, axial and selective coding. Go from transcripts and initial labels to research findings you can support.
Interview coding means giving relevant passages a short label. Comparing those labels helps you find patterns and differences. Here is a quick overview of three familiar types of coding:
Open coding: label passages to describe what the participant says.
Axial coding: group codes and examine relationships between categories.
Selective coding: connect the categories around a central explanation.
This three-part approach belongs to certain research methods that use interviews to develop a theory. Not every interview analysis requires all three. The explanations and examples below help you understand the differences.
Need to turn your recording into text first? Transcribe your interview with Audiogest and use the transcript as the basis for your analysis.
How to start coding interviews for your research
Whether you are analysing interviews for a dissertation, customer research or a workplace study, start with your research question and a transcript you have checked. Then select passages, compare codes and write up your findings.
Step 1: Transcribing your interview
A transcript is a text version of your interview. You can type the recording yourself or use automatic transcription. Audiogest turns audio and video into text, with speaker labels and timestamps.
Check the text: listen again to unclear passages. Check names, numbers, negatives and who is speaking.
Keep the source reference: record the participant and a timestamp or line number for each passage.
Use the full transcript: a summary can leave out details and different experiences.
Are you also studying pauses or hesitation? Record those separately and check them against the recording. Read more about interview transcription.
Step 2: Preparing for interview coding
Write down your research question. For example: ‘How do writers use AI when writing articles?’
Read the whole interview first. A statement can appear to mean something different without the rest of the conversation.
Choose your approach. Develop codes from the interviews, start with existing concepts or combine both.
Create a record. For each passage, note the participant, source reference, text and code. Use a spreadsheet or a research application.
Describe your codes. Record what each label means and when to use it. Revisit earlier passages when you change a code.
You do not need to decide every code at once. Read again and compare new interviews with earlier conversations. Include statements that do not match your initial expectations.
Types of coding in interviews
The difference is what you do with the text: open coding assigns labels, axial coding examines relationships and selective coding develops a central explanation. You can move between these activities during your research.
Open coding: giving interview passages a label
In open coding, you read the interview and give relevant passages an initial code. Stay close to what the participant says. A label such as ‘Editing time’ is more specific than ‘Experience’. One passage can have several codes.
Example of open coding
Imagine you are studying how writers use AI for articles. The passages below are invented examples, not research findings or experiences of Audiogest customers.
Interview passage | Code |
|---|---|
‘I use an AI tool to create the first draft of my articles.’ | AI use in the writing process |
‘Sometimes the generated text is surprisingly coherent and useful.’ | Quality of AI text |
‘Editing AI text takes more time than I expected.’ | Editing time |
‘I use AI to move forward when I get stuck writing.’ | Help when writing gets difficult |
‘First, I had to learn how to use the software well.’ | Learning to use AI |
Tip: do not add a judgement that the passage does not support. ‘Editing time’ is an appropriate label here; ‘AI saves no time’ goes beyond what the participant says.
Axial coding: grouping codes and examining relationships
In axial coding, you compare codes and organise them into broader categories. Then examine the relationships: under what conditions does something happen, what does the participant do and what happens next? Grouping labels alone is not enough.
Example of axial coding
You could provisionally organise the example codes like this:
Broader category | Related codes |
|---|---|
AI in the writing process | AI use in the writing process; Quality of AI text |
Time and productivity | Editing time; Help when writing gets difficult |
Learning to use AI | Learning to use AI |
Next, examine relationships such as these:
Does editing take less time when the first draft is more useful?
Do writers with more practice experience different benefits from beginners?
Does AI help someone start writing, even when editing takes a long time afterwards?
These are questions, not conclusions from the table. Compare the interviews and look for experiences that contradict your possible explanation.
Selective coding: developing a central explanation
In selective coding, you connect the categories around a central category: the concept that helps explain the main relationships. This belongs to research aimed at developing a theory. It is not simply the code you used most often.
Example of selective coding
In our invented study, a provisional explanation could be: ‘AI supports writing, but requires practice and review.’ You would then examine how the categories relate to it:
AI in the writing process: how does someone use AI, and how useful is the text?
Time and productivity: what does writing support offer, and how much editing is still needed?
Learning to use AI: what changes as someone gains experience?
The example passages are not enough to support this explanation. Examine it across your full study and revise it when experiences contradict it. If your aim is to describe recurring experiences, you can develop themes without developing a central theory.
After coding: writing up results in your research report
Use your codes to answer your research question. Keep what participants say separate from your explanation and your recommendations.
Describe your approach: who you interviewed, how you chose participants and how you coded the interviews.
Explain your findings: describe patterns and differences between interviews.
Support findings with passages: include short quotes or clearly marked paraphrases, with source references.
State the limits: what other explanations are possible, and what can you not establish?
Describe the next step: what do you want to discuss, investigate further or try?
Many passages with the same code do not necessarily mean that many participants share that experience. Ten passages can come from one interview. Check quotes against the recording and follow the agreements about consent, names and identifying details.
From interview to a first report with Audiogest
Audiogest helps you transcribe the recording and create a first summary or report. With your own instructions for the AI, you choose what information to extract. Check the text and supporting passages yourself: AI can miss or misrepresent passages and does not replace your research method.
To organise and code many interviews, you can also use a research application such as ATLAS.ti, MAXQDA or NVivo.
Upload your interview to Audiogest and start with a transcript you can check and code.