You Should be Doing More Generative Research. Here’s a Guide to Get Started.

In addition to this guide, we wrote an "everything-you-need-to-know" generative research how-to. If you're looking for a more substantive deep dive, read it here. 

It’s a big claim, but I’m willing to make it: generative research is the most important user research method.

And I feel this way because at first, I messed it up. 

When I first started my user research career, I fell victim to a common misconception. I was convinced the backbone of UR was usability testing. I could conduct card sorting, write relatively unbiased surveys, and feel validated that usability tests could answer any question I faced.

After these sessions, I knew the gaps and problems users were having with the platform. I understood what parts of the UX flow were clunky, or which pieces of UI seemed more confusing than others. I knew where users tended to drop off, or what “hacks” they put together to accomplish tasks on the platform. I knew the problems of that product in and out. What more would a user researcher need to do or know?

After a few weeks at that first company focused on conducting multiple usability tests on the product, I was asked to put together personas and a customer journey map. 

This was my first real gig as a user researcher, but I had, fortunately, built personas and journey maps for portfolio-building projects and small freelance assignments. In the past, I’d developed what I’d now call “proto-personas” and “proto-journey maps”— personas or journey maps based completely off of assumptions. I’d take my assumptions, talk to real users, and validate/disprove my hypothesis. 

But this time, this wasn’t what I was asked to do. I was asked to start from scratch. 

I was lost. So I found a few templates online and sat for hours, on Sketch, trying to recreate their designs. After the designs were created, I looked at the recommended headings for personas: 

• Goals
• Pain points
• Needs 
• Tasks 
• Motivations

I knew the tasks that users performed on the platform, so I started there. But after filling in the top five tasks of our users, I was stumped again. I moved on to pain points, but they were solely focused on problems or frustrations users felt about the platform. How in the world was I supposed to know these people’s motivations? Or their goals? 

I encountered the same problems with the journey map. I understood the basic flow users took through the platform, but had no understanding of what they were thinking about before or after they used our platform, what they did outside of the platform, or what made them choose our platform over competitors. 

Somehow, I managed to cobble together personas and a journey map based on the usability testing sessions I had conducted (a process which, now, makes me cringe). 

That moment was similar to that of a teenager realizing they don’t hold all the knowledge in the world; it was slightly terrifying and extremely humbling. At that time, however, my biggest problem was that I still didn’t know how to get the information I needed from users in a productive and unbiased manner. 

Luckily, after that job, I was able to begin working with someone who believed in generative or exploratory research. To be honest, when he first explained it to me, I wasn’t 100% sure what he meant and what the exact value was. The first time I tried a generative research interview, I bombed it. The second time was a bit better. Eventually, I became an evangelist for doing generative research more thoroughly and frequently. Here’s what it is, when it works well, and how you can get started: 

What is generative research?

Here’s the best definition I use to explain generative research: 

Through generative research, you use open-ended conversation to get a user to tell stories about experiences. From there, you foster and develop a deep understanding of that person’s overall motivations, goals, needs and pain points—both inside and outside the context of your product.

When you break down this definition, it’s full of powerful information. By understanding a person’s underlying thought processes while they are considering or using your product, you can go far beyond improving your current offerings. Instead of focusing on exactly how people are using a product, as I did in that initial job, you focus on what people are thinking as they use it, so you’re better equipped to answer the why behind their actions.

Here’s an example:

You notice many users hacking your platform to download multiple files at once. What do you do? 

Initially, you might decide to build a bulk download functionality, which should solve the problem. However, users continue to perform their “hacky” solution. Why is this still happening? 

After conducting generative research, you learn that many users had to send files to their managers—who put together and send reports to their overseas colleagues very early in the morning for meetings. The people who were downloading multiple images at once weren’t the same people who were putting together the reports, so the bulk download, while it did save some time, was not generally helpful. They didn’t need to download 50 or 100 images at once. What they actually needed was different levels of access to the platform, so their manager could flag important images for reports, and for users to download later. 

Generative research is the best way to find an insight like this. It gives a clear understanding of the problems your users face, and why they face them. You aren’t asking users for surface suggestions on how to make a product better—you’re trying to understand their feelings and the triggers behind their behaviors. You cultivate a sense of empathy about the people using their product, and they move from being just “users” to actual humans. Generative research allows you to come from the problem space, instead of the solution space, and to create a product that solves real problems. 

When is generative research necessary?

I find generative research useful and necessary all the time—but that’s not useful for setting your research priorities. So instead of leaving you with “it’s always necessary”—I’ll provide four situations when I’ve found generative research especially helpful:

• Before starting a company (or before developing a product). Generative research helps you to understand who your users would be, what problems you would be trying to solve and users’ mental models when they consider those problems. This will give you context for what actions they might perform on a given product, and, most importantly, why they would perform them. With this information, it’s easier to create features users will actually value and use.
• Near the end of the year. Use generative research to determine the upcoming strategy for the next few quarters. It’ll help you understand the most pressing issues your users are facing. From there, you can pivot the product in the right direction, and build features or products that solve their most pressing problems. It ensures you are aligned with your users as you define your strategic vision.

• When you are looking to innovate. Companies should regularly think about new ideas and ways to improve their offerings. If you have the bandwidth, you should always be doing generative research in the background. If you don’t, try to commit to at least 10 generative interviews every six months—where you sit down with your users and discuss what they are missing or have problems solving.
• While creating key deliverables. If you’re building personas, customer journey maps, blueprints, or any information-based visual, do some foundational generative research. It’s very difficult to produce a deliverable based on surface information about how a user views a product. Instead, using generative research, you can paint a real picture of the people who are using your product, and help your colleagues understand what users care about and why.

I would highly recommend conducting generative research at any of these stages.

When is generative research less helpful?

There are also a few points in time when generative research won’t help you meet your goals. Here are a few generative research weaknesses: 

• When you are trying to do fast and iterative changes on a product that don’t require a deep understanding of motivational behavior triggers. In this case, I would recommend A/B testing instead of generative research, as sometimes you simply don’t need foundational information to make a small, positive change.
• When you are simply trying to test prototypes, or test flows on your current product, usability testing or “walking the store” can be more helpful in understanding these concepts. Oftentimes, these tests come after generative research.
• When you have very little time. Generative research sessions are fairly long and take time to synthesize. The fastest I have conducted a generative research and synthesis project like what’s described below is two and a half weeks, from absolute start to finish. But I recommend giving yourself four weeks when you are just starting this methodology. The worst thing is to rush through this type of research and end up with less than ideal results. 

How do I conduct a basic generative research session?

The best way to get started with generative research is to actually do the research. Here is a step-by-step approach I use when beginning or teaching generative research at a company:

1. Define your users and who you want to recruit for generative research. Sometimes this is current users of your product. If you don’t yet have a product, it might be users of a competitor’s product. In some cases you might to balance your recruits with non-users or potential users. 
2. Define the goals and objectives of the research. What questions do you want to be able to answer by the end of this research project? What do you want to accomplish with this research?
3. Create a plan.
   1. How many participants will you talk to? I recommend anywhere from 7-15 participants.
   2. How long will the sessions be? I suggest anywhere from 60-90 minutes. 
   3. What is the timeline you need to adhere to? Is there an absolute end-date you need to have these results in by? Then back up by four weeks to determine when you should start.

1. Write a high-level interview guide or script that contains open-ended questions and conversation starters. I teach the two techniques below for beginners:
   
   
   1. The TEDW approach allows you to phrase questions in a very open-ended manner through the following ways:
   2. The 5-whys technique is very self-explanatory and encourages you to ask why five times to a response, as that will help get to the core of the user’s thought processes
2. Recruit users for the study. Note that this can take up to a week, especially if you are unable to offer a high enough compensation (or value), or have a hard time accessing users. I’ve finished my recruiting within a day, but I have also encountered situations in which it has taken me up to a week and a half to recruit users. If your user base or target audience isn’t especially narrow, a tool like dscout Recruit offers a quick turnaround from recruiting to generative research.
3. Conduct the research! Make sure to test your script internally. There is always time between creating an interview guide and your first research session, so use your colleagues for rehearsal. When it comes to generative research, practice really makes perfect!

It honestly took me months to get used to this style of questioning. Oftentimes, I leave a script behind in favor of just having a conversation with someone. The most common advice I will give students and colleagues is to imagine you are at a social event, and that you are genuinely trying to understand and get to know someone. In this case, you don’t come in with a predefined script but, instead, you ask them questions naturally and learn about them as people. 

With open discovery questions and the right approach to analyzing and synthesizing what you learn, you’re well on your way to transformative insights from real people—which I can tell you from experience beat making assumptions.