March 3, 2022
March 3, 2022
I had the most challenging time learning and understanding the synthesis element of the user research process. Even though I came from a research background, I struggled. It wasn't the same kind of work with which I was familiar.
There was so much qualitative data to sift through my head would spin just looking at it. In addition, it was tech, not academic research, and my colleagues expected a very particular type of report so I couldn't hide behind my verbose ramblings. Every word had to be concise and meaningful.
How was I meant to shove so much qualitative data into a neat, straightforward report?
It all started with sound synthesis. However, I had no idea how to synthesize this type of data. Plus the worst part was—no one could tell me! So many terms were floating around: affinity diagrams, coding, tagging, patterns, clusters—but there was no one to learn what these meant.
I went to my best friend, Google, and more frequently than not it would let me down. No one shared their synthesis process because they couldn't. Many people speak generally about the synthesis process because the data in those steps is confidential. Therefore, unless you have others to learn from at your organization, you might struggle as I did.
To make it more accessible and less overwhelming, I’m sharing my process and unveiling this mystery.
Also, I recommend checking out my Introduction to User Research course to learn the foundational skills you need to become a junior user researcher, whether you are looking to transition into the field or are just starting in your career.
I learned synthesis through a lot of trial and error. Throughout my career, I’ve experienced roles where I’ve been a team of one with no user research manager to guide me. So trying and iterating became my go-to.
First, I had to learn what synthesis was and the point of it. Synthesis was about conjoining information across different participants to see the overlap. When we found this overlap, it could indicate a pattern.
From there, patterns meant that we might have to create, improve, or change something in our product. So, I had to assemble a load of qualitative data to find similarities across people to make our product better—simple right? (Not.)
Before I dove into my first synthesis, I researched the different components. Although most articles were theoretical, I was able to piece together a few critical elements of the user research synthesis process:
After reading what must have been 100 articles, I could recite the definitions of these terms, but had no idea how to apply them to all the data I had. I had questions milling around my head and no one to answer them:
For a while, I tried to take notes during my research sessions. It was a mess. Even though I typed quickly, I couldn't keep up with my participants, especially when they launched into a juicy soliloquy. Taking such detailed notes during the session also inhibited my ability to listen actively.
Eventually, I abandoned my frantic scribblings and just listened to the session after and transcribed what the participant said. I used a transcription tool or paid someone to transcribe the interviews when I was under time pressure. I would only note down crucial points or topics I wanted to revisit during the session. I focused on the participant with this approach and left the heavy lifting for after when I would transcribe the interview.
Over time, I learned that codes and tags are words that represent something you found in your data. We use codes/tags to categorize raw research data.
Remember that transcript I wrote? I would use tags/codes to help turn users' statements, observations, or attitudes into categories. These tags/codes would later allow me to pick up on the patterns I needed to create recommendations for my team.
Through this article, I will use the terms codes and tags interchangeably.
It was easy to learn what codes were, but I had no idea what to do with them. How was I meant to create these? How did I know which were right or wrong? There are two ways to make codes:
Now, this further understanding didn't particularly help me. How was I meant to develop codes/tags before the data? And, for the inductive method, were the codes just going to jump out at me? Hint: they didn't.
I typically employ a deductive method, and I recommend starting with this. I use global tags in a lot of my research synthesis. The global tags I use are:
So, I went through my transcript and looked at what the participants said. Then, I tagged that part of the data when their expressions or statements matched the above tags.
For example, let's say we were researching people's mental models on stretching after the gym, and we heard the following statements:
If we used the inductive method and created tags after reviewing the data, we would use this same information to create relevant tags. For example, if this information above was repetitive, we could create tags like:
I use the inductive coding method by reviewing a few transcripts and seeing what similar ideas or thoughts people are having. As I did above, I then try to create a larger category for similar thoughts or statements.
Now we come to affinity diagrams. I use affinity diagrams for everything from making job decisions, writing, organizing my life, and, of course, user research.
Affinity diagramming is the process of bringing together all of your data into groups. Remember the tagging? This is when we use that work. If you used global tags, these could serve as broad categories.
For example, you would bring all the goals participants had under a goals section. Let's look at what this could look like:
So, we've identified all of these as goals and brought them under the goals category. But there are a few variations in these goals. Through affinity diagramming, we can cluster similar goals into smaller subgroups.
We do this because we want to get to more specific recommendations. If we break the data down into similar, smaller points, we can have more apparent action items.
So, let's look at how we might break these goals down:
Now we have clusters, or patterns, of information that we can go away and think through with our teams. Instead of presenting them with all of these goals, we can say, "people are scared about aging and being stiff, what can we do to help them?" or, "people want to become more flexible in general, how might we help them?"
Check out my Miroverse template to get you started!
I'm using a fake and minimal dataset that I just came up with while writing this article, so these clusters aren't particularly robust. Usually, I define patterns or trends by the sample size. I typically think of a trend if 1/3 (round up) of the total sample size said something similar.
So, for seven participants, if three or more said they had a fear of aging and getting stiff, I would categorize that as a pattern. If I spoke to 15, I would consider a pattern after five people said the same thing.
Now, this isn't an exact science, and there can be exceptions, so you don't need to use this as a hard-and-fast rule. For example, one-off insights can be compelling. As you advance in your process, you will better judge patterns and trends, but the 1/3 rule is a nice place to start.
This topic is a whole other article, which I've already written! Dive into the step-by-step guide of how to write compelling insights.
Synthesis, like everything in user research (and life!), takes practice. You don't have to know the exact, perfect process to get started. Eventually, you will create a rhythm and style of this process that is all your own!
Nikki Anderson-Stanier is the founder of User Research Academy and a qualitative researcher with 9 years in the field. She loves solving human problems and petting all the dogs.
To get even more UXR nuggets, check out her user research membership, follow her on LinkedIn, or subscribe to her Substack.