Tagging interviews for analysis
Tagging is a key feature of most UX research repositories. We’ll look into why it is worth the effort and how to prepare a tagging system for analysis — without drowning in a sea of labels.
“Why should I tag my data? It’s too time consuming.”
It will increase the life expectancy of your work, beyond the timeframe of your project, from an hour (i.e. your presentation time) to potentially a few years. Systematically and consistently tagging your data will help you build a bank of quotes and insights that can support future projects, facilitate stakeholder buy-in, and be a source of inspiration for UX and marketing copy. UX repos are in my opinion a massive step forward in terms of communication and visibility for UX research. In exchange, you’ll have to put a little bit of extra work when processing your data.
Minimize bias in your analysis. As user researchers and humans, we have assumptions that we might unconsciously be tempted to confirm. Certain participants are also more eloquent or energetic than others and our emotional response can cloud our judgment. There are many ways to reduce bias in analysis, I personally found that checking the occurrence of certain tags is one of them. The point here is not exactly to turn qualitative data in quantitative data (I am still a bit dubious of the charts feature in Dovetail) but instead to use tags as reminders of diverging opinions on a topic.
Should I tag everything?
Not all content is worth tagging. Interview content focusing on workflows and tasks (e.g. answering the question “Can you explain to me how you usually review and choose a browser extension?”) are probably better processed and communicated as visuals. Focusing on text will be a waste of time for you and a pain for the stakeholder to digest and memorise. For such content, tagging can still be used minimally to store a few quotes that will illustrate the high and low points of the tasks.
On the other hand, quotes around values, beliefs, opinions and attitudes are great candidates for systematic tagging.
Now that we know which content we want to work with, let’s talk about how we will tag it.
How do I create a tagging system?
The prospect of labelling hundreds of lines of user interviews can seem overwhelming, but it really doesn’t have to be. With a little bit of preparation beforehand the process is relatively painless.
I personally use the following model, with two tag types :
Tag type 1 = question
Tag type 2 = answer
For ease of use, create the two types of tags in separate boards.
Board type 1 : questions
This board will be the backbone of your tagging system. It will ensure you search for and highlight content that answers your research questions. To create your “question” tags, simply follow your interview guide and extract the topics you want to explore.
Board type 2 : answers
All other boards will be the answers to your interview questions, sorted out by themes.
The simplest way to start creating them is to have one board which lists all the answers you remember for one given question. Don’t worry about being comprehensive, you’ll be able to add the missing ones on the fly while highlighting your notes.
Our tags are ready. How do we proceed from now on?
Tagging interview notes
Notes preparation
The way your notes are organised will play a big part in how efficient your tagging can be. My personal system is : one note = one interview.
For the content to be usable, you will need good transcription. To avoid spending hours transcribing an interview, you can pair up with a good note taker or use automatic transcription (Dovetail just added this feature!).
Highlighting and tagging
In each note, I’ll look into the participant answers, and tag each relevant quote with the corresponding topic + answers.
If you forgot some tags while you were preparing your boards, Dovetail enables you to create them on the fly.
Lastly, if a participant makes the same statement under various forms, pick the most articulate and enticing one, to avoid unnecessary clutter in the highlights table.
How much time will this take me?
Following this process and with a bit of practise, my time estimation is 15–20 minutes per 45-minute interview. The project I used to illustrate this post contained 6 interviews, which represents 1.30 to 2 hours of tagging time. Normally at this stage, we’ll have soaked in so much, that generating insights shouldn’t be too challenging.
In our next blog post, we’ll look into how to analyse the data and create the insights.
Any feedback? Questions? I’d love to know how you tag your interviews in your user research repo!
