In this write up, I'll list a few common techniques used to find the very first people you can talk to as potential users

Preparing for your research

The first and foremost is to have an idea about who you want to talk to. Follow these steps:

  • Write down what you want to do. say "I want to build a chatbot for customer service representatives".
  • Then ask yourself "why would they want something like this?". Try to find an answer online by searching for about 10 mins.
  • Then write down the answer: say "because they have to repeatedly asnswer very similar questions and would love having a better solution".
  • Then ask yourself "why would they want a better solution and what would that look like?". Try finding an answer to that.
  • Repeat this process 3-5 times.

After this process (I call it "so what" exercise but people call it "abstraction ladder" because with your repeated challenges your're abstracting away from the first thing that came to mind and trying to get to a fundamental reason).

Once you have gone through this exercise you should have formed a much more informed opinion about the problem you want to solve, and importantly you should have surfaced lots of assumptions you are making. One can categorize those assumptions into the following:

  • assumptions about who the user is (audience hypothesis)
  • assumptions about what problem they have (problem hypothesis)
  • assumptions about what solution they want (perhaps partly informed by what solutions they are currently using)

Your task in talking to users is (in)validating these assumptions so that you're left with a list of assumptions that you can stand behind, a much better understanding of the problem, and a concrete understanding of who would need a solution and why would they use it. So, how do you find these people to talk to?

Finding users

The following is some of the techniques we use at aggregate intellect

  1. Good ol' social media post - the baity kind. Often times I post on social media with a question that is sometimes a little baity: people love to correct you!! eg. "I'm doing some research on using chatbots for customer service and all I see are stupid approaches. Does anyone work on anything interesting?". See, while I know much better, I posed as an uninformed person to get a reaction from people. Then I can see who feels strongly enough to correct me, and then I'll message them directly and see if I can talk to them.

  2. Good ol' social media post - the honest kind. Of course, if the above is too much for you, you could just honestly say that you're looking for feedback and would anyone help you. So, I would say "I'm looking for anyone who works on customer service chatbots for a quick 15 mins call. I'm doing research for an app I'm building". Or even better if I know the problem statement a little more, then I would say something like "I'm looking for anyone who has recently interacted with a customer service chatbot and had a good or bad experience to have a 15 min chat".

  3. Surveys. Sometimes we put out a very short survey asking a few non-leading questions about the problem statement we have, for example about the last time people had to interact with a customer service chatbot. Then at the end we would say, I would love to send you a $10 gift card to talk to you more about your experience. Please leave your email here if you are ok with me contacting you. You could try this without offering the gift card first and if that doesn't work do it again with the gift card. people love giftcards!

    • A more advanced version of this is to use things like paid advertisements and collect a lot of contacts in one go, but that's more complext to pull off. See here for example.
  4. Warm introductions. This is the tried and true approach. Go on Linkedin, find people you want to talk to. Figure out which of your 1st order connections are connected to them. Ask for introductions. Repeat for all other physical / virtual networks you're part of

  5. Cold approaches. If all fails, who is stopping you from sending an email / linkedin connection request / etc, politely laying out why you are contacting them, and chances are they might turn out to be nice people. You wouldn't hit 100% of the shots you dont take, eh?

Covers: theory of Qualitative User Research
Questions this item addresses:
  • How to find your "first" users for user research?
0 comment

Asking Users What Problems They Have

Total time needed: ~2 hours
The purpose of this list is to provide you with contextual information about why user research is important, and what the common techniques to do that are. By the end of this list, you should be able to have plan to identify your users, find ways to access them, and actually carry out interviews with them
Potential Use Cases
product development, user research, design thinking
Who is This For ?
BEGINNERany ml practitioner interested in proactively identifying the problems they should work on and building a narrative about why they want to work on those
Click on each of the following annotated items to see details.
ARTICLE 1. Human-Centered Machine Learning
  • How to stay focused on users when designing ML products?
  • why can't ML figure out if you're solving the right problem?
  • why using ML does not necessarily provide a unique solution to the problem?
  • what are the specific techniques to iteratively move towards a good solution?
15 minutes
ARTICLE 2. The “Why” Behind Qualitative User Research
  • what is qualitative user research?
  • when to perform qualitative user research?
  • what methods you can use for qualitative user research?
15 minutes
WRITEUP 3. Technique for finding your "first" users for user research
  • How to find your "first" users for user research?
10 minutes
ARTICLE 4. What to do when you can't find users to talk to?
  • how to carry out user research without actually talking to users?
10 minutes
ARTICLE 5. Quantitative User-Research Methodologies: An Overview
  • What are the common methods used for quantitative user research?
10 minutes
ARTICLE 6. AI and Design: Putting People First
  • why is it important to think about humans when designing AI products?
10 minutes
BOOK_CHAPTER 7. PAIR: People + AI GuideBook
  • How to define user needs and success?
  • How to collect data and evaluate performance?
  • How to incorporate explainability?
  • How to collect feedback and improve?
60 minutes

Concepts Covered

0 comment