Transcripts
1. Introduction: Hi, welcome. I'm Pascal. I've spent over a decade doing human centered design for some of the world's
biggest brands, and now I help teams use AI to do faster,
better research. This course has already
helped thousands of learners get up to speed
with AI in UX research. I keep it updated
as tools evolve, and I'm teaching you skills and frameworks that transfer
across platforms. So what you learn here stays relevant no matter which
AI tools you end up using. Whether you're a UX researcher, designer, product manager, founder or just someone who wants to understand
customers better, you know you should
talk to users. But planning,
interviewing, analyzing, and writing it all up can
feel like a whole extra job, especially if you're new to UX and you don't know
what good looks like yet. AI has changed what's possible. We can move faster and still
do work we're proud of. In this course, we're going
to use modern AI tools, including Chat GPT, to
make the messy parts of qualitative research
feel a lot more doable. We keep it simple and
confidence building. So from day one, we'll use
a few practical guardrails that keep you safe and make your work easier
to stand behind. Privacy, consent,
human judgment, and a clear evidence trail. In this course, we'll
go step by step for the real UX research
workflow in plain language. We'll start by turning
a vague problem into a sharp research objective so you know what you're
trying to learn from users. Then we'll use AI to write interview questions that
get useful answers, run into the confident and analyze transcripts without
losing your own judgment. And finally, we'll
turn that into insights that stakeholders
can actually trust, and you won't be
starting from scratch. You'll get templates
that you can reuse, including
disclosure language, ethics checklist,
and prompt patterns that don't depend on
one specific tool. If you're doing research
as a dedicated researcher, part of your product
or design role or because you're
building a business and you need answers fast, and you want AI to
help you move quicker without cutting corners,
you're in the right place. I'll see you inside the class.
2. Class Project: Hello again, and
welcome to the course. During this course, you
will actually plan, run and analyze your own AI
powered UX research project. This is where the magic happens, and I'm thrilled to guide you through it. The goal is simple. Run a short qualitative
UX research loop, using AI as a
practical assistant, and produce a set of outputs
you can confidently share. Let's break down the steps. Firstly, you choose a product or experience that you're
passionate and curious about. Then guided by the class, you use an AI assistant
like chat GPT, claw, Gemini, or similar to help you form
compelling questions. Interview friends, colleagues, volunteers, or actual customers, and then we use
AI tools again to interpret the data
and uncover insights. And finally, you'll summarize your discoveries in a brief
report or presentation. So by the end of the class, you'll have not only
practiced all the skills, you have also applied them to a real world scenario
and uncovered valuable customer insights for your product or area that
you're passionate about. To illustrate, here's an example
of a successful project. One of our past students explored a local
Jim's mobile app. They interviewed users,
analyzed their feedback with ChechBT and found
actionable insights for improving the
user experience. The result, a comprehensive
understanding of what users love and
what could be improved. A created in hours, not weeks, thanks to the efficiencies
achieved through AI. Now, how to make your project
shine? Here are some tips. Have a discovery mindset. The field of AI is
constantly evolving, so there are no right or
wrong ways to do this. What matters is, are you learning things you
didn't know before? Experimentation is key here and choose a subject you're
really curious about. It doesn't have to change the world or be $1 billion idea. Start small and get curious. Want to find out what
makes people choose a right chair or a
taxi, for example. Or get curious about
what customers would change about your
website, if they could. Then choose an AI tool to be your personal
research apprentice. Following the guidance
from the course, don't just expect
the magic to flow, but have a bit of
a back and forth, pro and challenge, bounce
ideas around with it. Now, have some fun.
So as a first step, start thinking about something
that you're curious to learn about your customers or the product
you're researching. Brainstorm some questions you really want to
find an answer to. What attracts people
to this product? What are their pain points?
What would they like to see? And lastly, don't forget
the power of community. Share your project.
Engaging with others and receiving feedback is an essential part of the
learning process. Remember, I'm here to help, and together, we elevate
your UX research game. So are you ready to speed up your customer research using AI? Let's make your research more insightful, efficient,
and engaging. I can't wait to see
what you discover.
3. The fundamentals of UX Research: So, welcome to the
fascinating world of user experience
or UX research. At its core, UX research is a systematic investigation aimed at understanding users needs, behaviors, and pain points. But what does that truly mean? Now imagine a product or a service that you
interact with daily. What makes it enjoyable
or frustrating? What guides your choices and
what leaves you confused? These are the types of questions that UX research
seeks to answer. You see, the goal of UX research isn't merely to
gather information. It's about delving deep into the human aspects of
technology and design. It's a bridge that connects the creators of a product with the people who will use it, ensuring that these
products are not just functional but
truly satisfying. In a way, UX research is a
bit like detective work. A mission to uncover
clues, solve mysteries, and reveal insights that lead to well designed products
that resonate with users. It's an exciting journey that prioritizes empathy,
curiosity, and innovation. The best part, it's
a journey that we're all a part of whether
we realize it or not, because at the end of the
day, we're all users. So as we go into this lesson, let's keep our minds open
and our perspectives broad. Let's learn to see
through the eyes of the user appreciate
their needs and desires and explore
how we can use UAx research to create
experiences that people love. After all, the art of enhancing user satisfaction is not
just about technology. It's about humanity,
connection, and understanding. And that's what
makes UX research so powerful and essential
in today's world. The goals of UX research are the beating heart of
this exciting field. First and foremost, it's
about understanding users. Who are they? What do they want? Wo frustrates or delights them. By digging deep into
these questions, we can create products that
truly resonate with people. Next, the insights we gather
don't just sit on the shelf. They directly inform
design decisions. It's like having a roadmap that guides us towards creating experiences that align with real human needs
and expectations. And this is where
the magic happens, where empathy translates
into innovation. And lastly, and perhaps
most importantly, UX research aims to enhance
overall user satisfaction. We're not just building
something that works, we're crafting experiences
that people love that feel intuitive and that add
real value to their lives. This is the ultimate goal and the pinnacle of what UX
research strives for. By understanding,
informing, and enhancing, we become not just creators, but compassionate
problem solvers, turning ordinary products into
extraordinary experiences. As we delve deeper
into UX research, it becomes essential to recognize the different
types and approaches that we can employ each serving a unique purpose in our quest
to understand the user. We often categorize
UX research into two broad types qualitative
and quantitative. Qualitative research helps you explore behavior
and motivations, letting us ask the why
and the how questions. Quantitative research,
on the other hand, is about gathering
numerical data. It's where we count, measure, and compare to uncover
trends and patterns. Within these broad categories, we classify research into different stages
like exploratory, evaluative, formative,
and summitive. For the purpose of this class, we will zero in on
exploratory research. This is where we venture
into the unknown. We ask open ended questions and discover new insights
and opportunities. It's about laying
the groundwork, identifying the
needs and sparking the ideas that will guide
your design process. The other stages evaluative, formative and summitive each play vital roles in
the research process, guiding us, testing our ideas, and assessing our final product. But it's the exploratory phase
that often sets the stage, providing us with the raw
materials and inspirations to create something truly
unique and user centered. These distinctions help
us tailor our approach, choose the right methods and ask the right questions
at the right time. By focusing on
exploratory research, we're opening the door to
a world of possibilities, laying the foundation for innovation and setting the stage for a rich and engaging
exploration of UX research. Quick reality check
before we go any further. In the real world, UX research is rarely a big dramatic
project you do once a year. Most teams do what's called
continuous discovery. That just means you're
talking to users regularly in smaller loops and feeding those learning straight back
into product decisions. And it's also usually
mixed methods. So yes, we'll do
interviews because they're brilliant for understanding why people behave the way they do. But you'll often pair that
with things like analytics, support tickets, surveys,
and usability testing. In this course,
we're focusing on exploratory interviews
because they're the fastest way to build
deep understanding. Then we'll learn how to
keep our quality bar high as we move from
conversations to decisions. The UX research process is a structured journey
that takes us from the initial spark of curiosity to the final presentation
of insights. Let's break down this process
into six crucial steps that guide us in uncovering
valuable user insights. It all begins with planning. What do we want to discover? Who are our users? What
questions will we ask? Planning lays the foundations by setting clear objectives, defining our scope, and crafting a roadmap that will guide
us through the research. The next step is recruitment, where we identify and select the participants who will
be part of our research. So we're looking for individuals who represent our
target audience and those who can provide the insights that align
with our objectives. Now comes the exciting part,
conducting interviews. Whether you're face
to face or virtual, this is where we engage
with our participants, ask our questions,
and listen intently. It's a process of exploration, a dialogue that
uncovers the thoughts, feelings, and experiences
of our users. And once the interviews
are complete, we move on to the analysis. Here we dissect the data, break it down into parts, and look for patterns,
connections, and themes. It's like piecing
together a puzzle, finding the hidden truths
that lie beneath the surface. Synthesis takes our
analysis a step further. We're not just
identifying parts, we're putting them together
to form a coherent story. We're drawing conclusions. We're connecting the dots and translating
our findings into actionable insights that can guide our design and
decision making. And finally, we arrive at the point where we
share our findings. Whether with the team, your
client or broader audience, this is where we present
our insights, our story. It's the culmination
of all our efforts, the moment where
research comes alive, sparking conversations,
inspiring action and influencing the way
we create and innovate. By following these six steps, we go on a journey that's
both methodical and creative. It's structured yet flexible. The UX research process isn't
just a series of tasks. It's a way of thinking, a mindset that puts the user at the center
of everything we do. It's how we turn curiosity
into understanding, questions into answers, and insights and into experiences
that resonate with people. And in this class, as we focus
on exploratory research, this process will be our guide, our path to discovering what lies in the hearts and
minds of our customers. But the adventure
doesn't stop here. As we move forward
in this class, we're going to explore
something truly revolutionary the intersection of UX research and
artificial intelligence. Imagine leveraging the power of AI to delve even
deeper and uncover insights faster and enhance our understanding in ways
never thought possible. AI is incredible at
speed and structure. It can help us turn a
messy brief into a plan, draft a screener, turn an interview guide
into better questions, and help us organize analysis. But AI can also make research worse if we use it
in the lazy way. Two big failure modes are, one, people paste sensitive data into tools they're
not allowed to use, and two, they accept confident sounding answers
without checking the evidence. So our rule for this
entire course is simple. AI can assist, but
we own the judgment. We keep an evidence
trail, we sanity check. And when something matters, we verify it like
a real researcher. And here's the quality
bar we'll use as we go. One, evidence. If we
can't point to a quote a behavior or a concrete
observation, it's not a finding. It's a guess. Two,
triangulation. Do we see the same thing
in more than one place, like across multiple
participants or in behavioral data or
in support tickets? And three, review. Could another human look at our notes and understand
how we got there? If we hit those three, we can move fast and stay credible. So as we concluded this lesson on the fundamentals
of UX research, let's reflect on the
journey we've just begun. UX research isn't just about
data, methods, or processes. It's about experience. It's
about connecting with people, understanding their
needs, and crafting solutions that make a real
difference in their lives. The essence of UX
research lies in its ability to bridge the
gap between users and creators to turn empathy
into innovation and to transform ordinary products into extraordinary experiences. It's not just about
what we make. It's about how we
make people feel. I encourage each one of you to think critically to
ask bold questions, and to imagine how you can apply the principles of UX research in your projects and
professional work. Whether you're a seasoned expert or just starting
on your journey, there's a world of
opportunity waiting for you. So get excited, stay curious and embrace the
UX research journey. Together, we're going to
explore, innovate, and create. With the help of
AI, we're going to take UX research to new heights, unlocking potentials and shaping the future of user experience. And it all starts here with you, ready to go on this
exciting adventure. So what are we waiting for?
4. Unleash your AI Research Assistant: Welcome back. In this lesson, we're going to talk about AI
in a really practical way, not the hype and not the fear, how do we use it as
solid researchers? Here's the headline.
AI is not the boss. We are. AI can
help us move fast. It can help us get unstuck, and it can help us stress
test our thinking. But it can also do this
thing where it sounds incredibly confident while
quietly making things up. And if we're not careful, it can nudge us into conclusions that feel tidy and convincing, but that aren't actually true. So our goal is not to
get answers from AI. Our goal is to use AI
to speed up the work, to keep the quality bar high. And for the rest of this course, we're going to use
one simple loop, draft, critique, verify,
and document. That's it. That loop is what lets
us use hechPT, Claude, Gemini, whatever is next without tying our
skills to one tool. And I want you to hold
on to one metaphor. We keep our hands on
the steering wheel. AI is the satnav. Alright, let's start with the most useful question of all. Or even is an LLM. An LLM, a large language model, is basically a system trained
to predict the next word. It has seen a massive
amount of text, and it has learned patterns
of how humans tend to write. So it can produce language
that feels fluent, coherent, and honestly,
sometimes a bit spooky. But here's the key bit.
Fluent does not mean true. LLMs are great at the
shape of an answer. They can draft, summarize, reformat, brainstorm, and help
us spot possible patterns. And for research work, that's
genuinely useful because so much of our job is turning messy inputs into
something clearer. What they can't do automatically is know what's accurate
for your project. They don't know your project.
They don't know your users. And they definitely weren't in the room with your participants. So the best mental model is
a very fast collaborator. Helpful, creative, sometimes
surprisingly sharp. Not a witness and not a source and not the person we quote
in a stakeholder meeting. And once we really get that, a lot of the confusion and hype around AI starts to calm down. Now, if we're going to use AI, we need to know how it fails because it fails in
pretty predictable ways. There are three
classic ones that I see all the time
in research work. Number one, hallucination. That's the polite word
for it made something up, a feature that doesn't exist, a quote that no one said, a neat little key insight
that sounds plausible but isn't actually grounded in anything you actually collected. Number two, overgeneralization. This is the two
people mentioned it, so it must be a
universal truth problem. Suddenly, we got users hate on boarding or everyone is
confused by pricing, maybe. Or maybe it was just
two people with a specific context
on a specific day. Research is all about context, and AI will flatten that
context if we let it. Number three, confident
tone. This one is sneaky. Even when the model
is just guessing, it can sound calm, certain,
and authoritative. When you're moving fast, it's really easy to mistake
confidence for correctness. So the fix isn't become
a prompt wizard. The fix is to use a
workflow that makes it hard to accidentally believe
something that isn't true. And that's where
our loop comes in. So let's do it. Alright, so here's
the thing that we're going to lean on for
the rest of the course. It's simple, it's repeatable,
and it keeps us honest. It's this loop, draft,
critique, verify, document. And the reason why I
love it is because it stops us treating AI
like an answer machine. Instead, we use it
like a power tool. Fast, helpful, and still
something we're responsible for. Let's walk through it. Draft. We use AI to get a first
version on the page. Not because it's perfect, but because it gives us
something to react to. This could be a kickoff agenda, a screener, an interview guide, a theme list, a report outline, anything that normally
starts with a blank page. Step two, critique. Now we switch gears. We ask the tool to critique
what it just made. What's missing?
Where is it vague? What's biased or leading? What assumptions has it
smuggled in without telling us? This is where we turn a knife
output into a useful draft. Step three, verify. And this is the part
that makes it research. We check the output
against reality, against the brief,
against our notes, against transcripts,
against actual quotes. If the tool makes a claim, we ask, where's the evidence? And if we can't back it
up, we don't ship it. Step four, document. Finally, we write down
what we decided and why. What we're confident about.
What's still a hypothesis? What's unknown? This is how
we keep our work defensible, especially when we're
moving quickly. Just to be really clear, this loop isn't a one off thing. It's basically the spine
of our whole process. We use it in planning, analysis, synthesis,
and reporting. If something matters,
we attach evidence. That's the quality bar. Now, tools, you'll hear
a lot of debate about which model is best
HGBT, Claude, Gemini. And by the time you
finish this course, there's probably three more. So instead of trying
to crown a winner, we're going to do something
that's way more useful. We're going to learn
how to evaluate a tool for a specific task. Here's the simple method. It's a three step test drive. Step one, pick the task. We choose a task that
we actually need. For example, draft an
interview guide from a brief, summarize a transcript section, turn themes into insights
and opportunities. Step two, we check the pricing
page. Is a quick filter. Before we test anything, we do the boring grown up bid, but this is going to
save us some time. We open the pricing page. Why? Pricing pages often
show a feature matrix. And that tells us a lot about what we're
actually paying for. Things like file uploads,
longer context windows, team features, data controls, usage limits, two
caveats, though. First, pricing pages don't
tell us output quality. Second, marketing
pages sometimes hide the real constraints
in the fine print. So we use pricing
as a quick filter, not the final decision. And step three, run
a tiny benchmark. This is the test drive. Now we run the same test
prompt in each tool, and we score it on a
few things not vibes, not marketing,
actual usefulness. And here's what we score.
Instruction following. Did it actually do what we
ask evidence, discipline? Did it stick to the provided
text or start inventing? Clarity, is the output
readable and structured? Handling length and files, can it cope with what we're
giving it? Safety fit. Can we use it with our
data and policies? And step four, choose
the best fit for today. We pick the tool that wins for that task in
our constraints, and we keep the workflow because the workflow
is what lasts. So the rubric we
use is input type. Do we need text only or
docs, images or audio? Length. Are we working with short notes or long transcripts? Privacy and permissions, are we allowed to upload this
content? Workflow fit. Do we need reusable templates, custom instructions or team
features? Cost and speed. Do we need good enough, fast or slower and
higher quality? Make this beginner friendly,
here's the Jeet code. If you're not sure
which tool to pick, start with the one you
already have access to. Then run the tiny benchmark. If it struggles,
then you switch. We don't need perfection. We
need a repeatable process. All right, quick, but
genuinely important. Before we paste anything
into an AI tool, we do a small safety check. Not because we're
being dramatic, but because it's very easy to accidentally share something
we didn't mean to. And once it's out there, you
can't really unshare it. So here's the routine
I want us to use. It's simple. And after a couple of times, it
becomes automatic. Step one, classify what
you're about to paste. Just ask, is this
public internal, confidential or personal data. There's participant info in
there, assume it's sensitive. Step two, redact, remove names, company names and anything
that could identify someone. Even small details can add up. Job title plus location, plus one memorable quote
can sometimes be enough. Step three, minimize. Only paste what we actually
need for the task. We're analyzing one
question from an interview, we don't need the
entire transcript. Less input usually means
less risk and better focus. Step four, set boundaries. Tell the tool how to behave. For example, you can
write in your prompt, use only the text I provide. Don't invent quotes. If you're unsure,
ask a question. If we do these
things consistently, we're on much safer ground
and our work stays credible. So, that's the foundation. We use AI to move faster, but we stay accountable
for quality. We don't outsource judgment. We use the loop, draft,
critique, verify, document. If you remember nothing else from this lesson, remember this. When something matters,
we attach evidence. That's how we stay useful, and that's how we
stay trustworthy. Now, before you jump
into the next lesson, I've got a short
assignment for you. This is not project work yet. It's more like a warm up.
We're keeping the excitement high and choosing a starting
tool that you can trust. Here's exactly what we do. Pick one AI tool to start with. It could be chat GPT, it could be Claude or Gemini, or anything that you
already have access to. We can always switch later. Then run a tiny test drive. In your AI tool, paste
a short piece of text. It can be a paragraph
from an article, a snippet of notes or anything. Then ask the tool to
summarize it into three bullet points and then structure it into a
table with columns, a key point, why it
matters and open question. For the critique,
start a new fret and ask the AI to
critique its own output. You can copy the
exact prompt you can use from the prompt
sheet that I'm providing. You can now use
your own judgment and score the output
using the scorecard. Or you can go one step further and ask AI to
help you with that. In that same fret, ask it to
score the tool out of five. Instruction following,
staying grounded, so not inventing stuff
and clarity in structure. Important, ask it
to give reasons for that score so that you know
whether you can trust it. And optional, run the same test in the second tool and compare. Part three of the assignment, capture your
questions for later. So using the process
map from lesson one, write down your thoughts and
questions for each stage. Planning, recruitment,
conduct interviews, analyze, synthesize and share. Just make two rows.
How do you think AI will help at each
of those stages? And the second one, questions
that you have about that. Then bring those
questions with you. We're going to answer them
as we go through the course. Next up, we're going to apply this to a real research plan. We'll take a brief
and turn it into a clear objective and kick off questions.
Alright, let's go.
5. Planning your research with AI: This video, we're going to roll our sleeves up and
dive right in. Let's unpack the
brief and use Chet PT to make your research planning more efficient and effective. Before we jump in, a
quick orientation. Right now, we're in
the planning stage of the UX research process. Our goal today is not to solve the product. It's
to get clarity. We want a short list of
smart kickoff questions, plus the assumptions
and unknowns we need to confirm before
we recruit anyone. And as we use AI here, remember the rule for
this whole course. AI can speed up the thinking, but we decide what matters, and we keep it grounded in the brief. One
quick safety note. If you're working with a real
client or employer brief, don't just paste the whole
thing into a public AI tool. Redact anything sensitive,
and wherever possible, use a short context packet. A context packet is basically just a short sanitized summary. What we're researching,
who we're talking to, then the objectives,
the constraints, and the exact output
format we want, you can use the
template provided as a download with this lesson. This gives the AI model what it needs without leaking
what it shouldn't see. So let's start by understanding the brief and preparing
for the kick off meeting. If you're working for a client, you usually have a detailed or not so detailed project brief filled with objectives,
goals, and expectations. So here's where AI comes in. With Chat GBT, you can
interpret the brief. It can help you distill
the essential information, identifying your key
goals and objectives. Then prepare for the meeting. AI can assist you in outlining
the main talking points, questions to ask, and even potential
challenges to discuss. You'll walk into
your kickoff meeting with clarity and confidence. Now, let's see this in action. Okay, so here we have
our research brief, and I've gone through the
brief already and just highlighted a few sections in it that are really relevant. We want to redact
anything sensitive. If you're working with a tool that is approved by your client and you have the okay to store their information
in those systems, then by all means, you can go ahead and just
work with the original brief. But otherwise, I recommend
that you do a context packet. Context packet will be
useful in any case, because it helps us later on when we need to brief
the AI again and again. So it's a good habit. Sort of done here.
I've gone into the brief and I've highlighted
the relevant section. So in this case, our
client is called Tax Corp, and they are a
traditional taxi company that have a presence in
several major cities, and they are facing competition from ride sharing services. And the objective
of the research is to discover opportunities to enhance the apps features and the experience and compete with those ride
sharing services. There's a little bit
around scope here and some constraints and
some expectations as well and the
participant criteria. So that's what you
would typically find in a research brief. And if you don't have all this
information in the brief, it already gives you some questions that you
can ask the client. Now, over here, I created
a contacts packet, and that's essentially
a sanitized brief. So what we want to
make sure is that all the relevant
information from the brief is also included
in our contacts packet. What I've done here is
I anonymize the client. So I'm saying for
the background, my client is a
traditional taxi company operating in multiple cities. They have an existing
consumer app and want to improve the
enter end experience, so it feels competitive
with rights products. Notice that all I did is I
just replaced the client name, and I've replaced
the name of the app. Then I've pasted it
in the research goals here, once again, making sure that I just say the app rather than
the name of the app. I've also included
the research method and the target participants. And then I've included
the constraints. What we need help with
here is we want to prepare for the
kickoff by preparing thoughtful questions
and checking some assumptions and
unknowns that we can validate before
we start recruitment. Okay, this is where the fun
begins. I'm in Chat GPT. And you can do this in the
AI tool of your choice, so it doesn't have
to be hat chPT. This would work equally well
in Claude or in Gemini, or whatever the
flavor of the day is. Now, on the prompt
sheet provided, you have your first
kick off prompt. So I paste the prompt
right into Chat chPT. I my UX research assistant. Help me prepare the
kickoff meeting for a new client project. Also includes some
constraints because we don't want the AI to
suggest any solutions. We just want to prepare
for the research because the research itself will
inform the solutions. Now, all we need to do is either use the file attachment here and attach our sanitized
research brief that we've just provided. I just added to the top here. So I can copy the brief, the sanitized brief
that we've prepared, paste it to the top, and
just make a new line. What I like to do is just
making three lines so that hhiPT knows that that
is a different section. Okay. And that is
your entire prompt. You can just start and
see what comes out of it. So what we can see now is that ChachiPT has been
interpreting the brief, and we can just check that that matches our understandings. It also points out that
the six weeks suggests a lean qualitative cycle with
limited runs of iteration. So that might be
something we want to discuss in the kickoff meeting. And now here we have
our kickoff questions. GPT gave us some questions
around the golf, some questions around the users. Do you have any existing
personas or segmentation models? Are their priority cities or markets where
insights are critical? So these are all really
good questions that we can now take into
the kickoff meeting. And it's also identified
some assumptions inherent in the brief and some unknowns that would be good to
clarify before recruitment. So the exact definition of regular share use in
terms of the frequency, that will be really
important to know whether the target participants
must have used the client's taxi upp
specifically or any taxi app. So these are all
really good unknowns that Chat GPT worked
out for us here, and that will make
our kickoff meeting so much more productive. That will have me a lot
more prepared to go into the kickoff meeting and to know exactly what kind of
questions I need to ask. Something else that's
going to be really useful is to just have a
concise summary of the research context
here because that'll help us later on in prompting
Chet GBT more efficiently. Just run this as a final prompt. In two sentences, summarize the research context and
goal from the brief. Ignore budget and
timeline details. Great. So now I can
reuse that summary later when I'm drafting
screeners and interview guides. Now, when we go into our kickoff meeting and things change, I can update the brief with the new information that
the client is giving me, and then I can run this
prompt again and get an updated summary of
the research context. So you can see how you
can play with Chet GBT. Almost like your assistant, your real research apprentice. It's really handy.
Remember, preparation is the foundation of success. With AI by your side, you can make this
stage more efficient, allowing you more time to focus on creative and
strategic thinking. So as you've seen in
this demonstration, utilizing ChechPT to interpret a project brief not
only saves time, but also ensures that
you're focusing on the right aspects from
the very beginning. It helps you to ask
the right questions, uncovering nuances that might
otherwise be overlooked. Now let's talk about
your next steps. For your project, I'd like you to take a similar approach. Think of a hypothetical
or real project related to a product or
service you're interested in. Write a brief, keeping in mind the balance between
detail and ambiguity. Now, using what
you've learned today, work on distilling the research objectives from your brief. You can do this manually or you may choose to use
Chat GBT to help you. Now, write down
some questions that you would bring to the kickoff
meeting for this project. Focus on clarifications,
challenges, and any strategic
considerations. And then reflect on the process. How was this experience? What did you learn and how might you apply this in your
professional work? This exercise will
not only sharpen your understanding
of the planning phase in EURk's research, but also provide
you with hands on experience in leveraging AI
for these critical tasks. Always remember,
AI isn't replacing your intuition or expertise,
it's augmenting it. Learning to incorporate
AI into your workflow, you're opening doors to more in depth exploration,
creativity, and precision. That's a wrap for this lesson. Take your time with this
action step and feel free to reach out if you have any questions or need support. I'm looking forward to seeing your progress and catching
up in the next lesson.
6. Recruiting participants with the help of AI: One of the most vital aspects of any UX research project is finding the
right participants. Here's a quick orientation. Right now, we're in
the recruitment stage, and this stage actually
really decides the quality of everything
that follows. Think about it. If we recruit the wrong people, then the interviews are noisy and then the
analysis gets messy, no matter how good
our AI prompts are. So today, we're going to
build a screener that finds the right participants and protects us from bias
and garbage data. But before we touch AI, here are the four
screener rules. One, don't lead the witness. Two, include disqualifiers, so you don't end up
interviewing everyone. Three, practice
data minimization. Only collect what you genuinely need to decide if
someone is a fit, and four, recruit inclusively. If we're not careful, screeners accidentally filter
out whole groups, and then our insights are just
a narrow slice of reality. The quality of your
insights depends on the relevance and diversity
of the people you interview. Today, we'll explore
how AI can be an invaluable tool in crafting research screeners to find
the perfect candidates. Research screeners are sets of questions designed to filter potential participants to ensure that they match the profile
of your target audience. Screeners need to be
precise but not leading, unbiased, and easy
to understand. Now, let's dive into
how you can use AI to help you write
research screeners. You could continue this in the same fret that
you already had. I will just start with
a new fret right now. And here, the first
thing I do is I paste our contact
packet again. So here's the context
of our project. Then I'm just going
to make a new line, and directly underneath that, I paste the prompt
for the screener. And now, this prompt
includes some requirements. So it's asking hGPT to create about eight to 12
questions that are mixed of multiple choice
and short answers. It includes the disqualifiers and all the criteria that
we just talked about. And I want this to be presented as a table because that's
going to be really useful to then translate into whichever platform I'm using
to create those screeners. So let's see what comes back. And here's the result. So I'm
looking for three things. One, is it actually screening
for our target behavior? Two, are the questions neutral, and three, are we collecting
only what we need? And so we have ten
questions here. And we can also see
what the options are. So I can recreate these
questions in my survey software. Whether I use Google
Forms or notion Forms or a survey monkey
or something else, I can choose the
right question type and input all the right options. And I like questions
that anchor on behavior, like how often someone uses a share and whether they've
used taxis recently. So questions like,
in the past 30 days, how many times have you
used a traditional taxi? That helps us to really identify the right target participant here because it is based
on what they actually do. And we also have a
quality question here. What is one reason you choose the transport options
you selected earlier? And Chat GPT included this question here because it
gives us the quality check. I can get a gauge on whether there would be a
good participant to talk to. Now run a QA pass by
pasting this underneath. So right after the
draft screen now. So what we're asking
here, essentially, is to do a QI pass and identify
any leading questions, identify any questions that
could accidentally exclude groups and questions that collect unnecessary
personal data. And so we're really asking
the AI to check itself now. So let's see what comes back. And this is the step
that people skip. So the QA pass is
where we reduce bias and make the
screener actually usable. Notice I'm using AI for
the screener draft, but I'm careful
about using it for the public recruitment copy because wording can really
change who opts in. It's not to replace
your expertise. It's really to augment
your expertise. So you can see how this
is already saving you a lot of time and a lot of brain power because you can
now use these questions, refine them a little bit, put
them into a questionnaire, send that out, and then
recruit your participants. So with the power of AI, you can draft questions quickly, ensuring that
they're unbiased and targeted to your
specific research needs. You can even ask
ChaGBT to review and suggest improvements to make the questions more effective. So you could use a
prompt like this to get some refinement on your already existing
screener questions. So using AI for this process
not only saves time, but it also helps you approach the screener from
different angles, bringing in a level of
objectivity that can sometimes be challenging to
achieve just by yourself. One more important
nuance before we wrap, using AI to draft a
screener is one thing, but using AI to write
the recruitment copy, so the message that you put out to advertise your study to potential candidates is where AI can accidentally create bias. So here's when we should
avoid doing that. The wording could
change who opts in or prime people to talk
about a certain thing, then we keep that copy human. If the topic is sensitive
or high stakes, then we also keep it human. If we need to be precise about what we're
doing with the data, the incentive, recording or consent, then we keep it human. And if inclusion
matters, we're careful. AI often tends to default to one professional voice that
can quietly exclude people. If we do use AI for
recruitment copy, then treat it as a first
draft and then run a QA pass, check for leading language,
check for clarity, check for inclusion, and check that we're collecting
only what we need. It's essential to remember that AI is a tool to
augment your skills. It's not to replace
your judgment. So always review the questions and consider the
ethical implications, such as ensuring
privacy and avoiding potentially sensitive
questions. Use your brain. And now it's your turn. Identify your target audience. Define the profile
of the participants you'd like to interview
for your research project. Then craft your
research screener. Use hechBT or draft manually a set of questions
that will help you identify suitable
participants. Remember, aim for clarity,
relevance, and impartiality. Then reflect on the experience. How did using AI
influence your process? What did you learn and what
challenges did you face? Please share your screeners with a class if you feel comfortable and don't hesitate to ask for
feedback or for assistance. So by now, you've seen
how AI can transform various aspects of the
UX research process from planning to recruitment. As we move forward, we'll
continue to explore more ways to harness
the potential of AI in your
research endeavors. I'll see you in the next liston.
7. Prepare for user interviews with AI: Conducting interviews
is a delicate art. As researchers, we must create a comfortable environment
for participants, guide the conversation
and listen attentively. This lesson will show you
how to leverage AI to prepare for and conduct
successful interviews. But first, a quick orientation. Right now we're in
the interview stage. This is where we collect
the raw material for everything that comes later. Today we're going to
use AI in two ways. First, to help us draft
a moderation guide, and second, to practice our questioning and
listening skills. Just one important note, AI role play is practice. It can help you get sharper, but it's not evidence
about real users. Before we get into questions, we need a solid consent
and recording baseline. In practice, you want to be very clear about
what's being recorded, why you're recording it,
who will have access, and how it will be stored. And you want to remind participants they can
skip any question, pause, or stop at any time. We'll keep this
simple in the course, but in real projects, always follow your
organization's policy and any local regulations. And one more foundation
before we use AI. Interview quality comes
down to question hygiene. We want questions that are
open ended and neutral. Avoid double barreled
questions where we accidentally ask
two things at once. And wherever possible, we anchor on behavior
and real examples, not opinions in the abstract. So let's start with crafting
a moderation guide. A moderation guide is a
roadmap for your interviews, outlining the topics to cover, the questions to ask, and the
flow of the conversation. Ensures consistency and allows you to focus on the interviewee. So let's explore
how AI can assist you in crafting a robust
moderation guide. Start by outlining the key insights that you
want to gather, then write your questions. Use hat GBT to help formulate open ended questions that
encourage detailed responses. Organize the questions to
create a natural progression, starting with general questions and moving then into specifics. Let me demonstrate this process. Here's the pattern I recommend. Give the model a short context packet and
clear research goals. Ask it for topics, open ended questions,
and a sensible flow. Then you do the human
part, edit for tone, remove anything leading and tailor the language
to your style. Before you start asking
research questions, I like to keep consent
simple and explicit. Tell your participant how
long it's going to take. Explain what you're recording,
who will have access, and how it will be stored, and remind participants they can skip anything or
stop at any time. So jump into Chet GBT. First, I'm going to paste
my contacts packet again, which is essentially the brief, just so Chet GBT knows what the project is about or what we're
about to do here. I make a few new lines, and right underneath that, I'm going to paste the prompt. So our prompt is essentially
asking the model to create a moderation guide for 45
to 60 minute interview, you can obviously adjust the
length of that depending on what emerged from
your kickoff meeting. And the same goes for
the core topics here. And the prompt, you may want to include the
topics here that came out of your kickoff meeting that you have discussed with
your stakeholders. And now let's get this
moderation guide produced. So as it generates, I'm scanning for open
ended questions, neutral wording, and a
flow that feels natural. I also like having a short
consent script at the top, so I don't forget the basics in the moment, and that's
what we have here. So it starts with, thank you for taking the time to
speak with me today. I'm interested in learning
about how you usually get around and your experiences
with taxi and share apps. There are no right
or wrong answers, so that's always
important to say. I'm here to learn from you. And with your permission, I like to record
this conversation, so I don't miss anything. The recording will
only be used for research purposes
and will not be shared outside of
the project team. So, yeah, it's just really good to have that included here. So that looks very good. Then we have a couple
of warm up questions. Are related to the topic, so asking about how you
get around the city. What was the last
time we used a car serves to travel
somewhere and so on. And then we're going
into the core topics, the current ride share habits, taxi usage, the end to
end booking experience, something around
trust and safety, pricing and ETA, and so on, all the way to the wrap up and having a few lines
here just to wrap up. Let's see what we have. If
you could change one thing about the way Right booking apps work today, what
would you change? This is always a great
discovery question to ask. And is there anything we did not talk about that
you think is important? And that's always also good for participants to just say
anything that's on their mind. And then it's always
good to thank the participant and
let them know that their input was very helpful.
So that's looking great. And now I run a QA pass by pasting this QA prompt right
after what was generated. So review the moderation
guide that you just created and check
for leading language or judgment or wording, double barrel, question
any questions that are too abstract or missing
follow up probes. And so I want Chet
GBT to revise the moderation guide accordingly.
Let's take a look. So it's included a
section with key changes, rephrased abstract
questions into recent concrete situations to anchor answers in real behavior, remove subtly evaluative
wordings such as especially smooth and replaced
with neutral phrasing like felt very easy, okay? Yeah. This QA passes where the guide really becomes
interview ready. So it's also how we avoid AI generated questions
that sound fine, but subtly lead the participant. Now we've got a solid first
draft moderation guide. Can now take all of
this and copy it into my own document and give it a thorough read through
and adjust it to my style. Fantastic. And then we can go
out and do our interviews. But if you've never
done interviews before, this might be quite daunting. Conducting an interview is
more than following a script. It's about engaging, listening, and adapting. But
you guessed it. You can even practice active
listening and framing open ended questions with
AI. How does that work? Let me show you. This
is a skills drill. It helps you practice
active listening, asking better follow ups, and staying neutral
under pressure. But very important, this is not evidence
about real users, so we don't treat it like data
or use it in our analysis. It's really just a practice
run for you to get into the flow before
talking to real users. A couple of ground rules
make this feel realistic. Keep the simulated
responses brief, limit the role play
to ten questions and always ask for
feedback at the end. And when you can, anchor on
real examples with prompts. Like tell me about a time when
let me create a new chat. In this chat, I'm going
to use this prompt, which is about setting Chat GPT up to do a
role play with you. This is so that we can
practice our interview skills and be really confident that when we're talking
to real people, we're very competent and confident that we're
doing that right. So a little bit of
practice helps. So in this case, we're setting up a role play with Chat GBT. I'm asking CheGBT to
simulate a user I want Chet GPT to provide conversational and brief
responses to my questions. So I don't want it to just
generate long paragraphs. I really just want to practice this really quickly
and authentically. I'm also setting some parameters around how the role
play should go. I don't want this
to go on forever. This conversation will only
contain ten questions. And after ten questions, it would be good to get
some feedback, right? So ChatGPT can provide feedback on our
interviewing skills, which include active listening and asking open ended questions. Active listening means that I'm really listening to
what the user is saying and I'm paraphrasing
and reflecting it back to them to make sure
that I understand it right. Asking open ended
questions means I'm asking questions that cannot be simply answered
with yes or no. They're typically questions
that start with the W. Why, what, when, where, but also how? Now, let's see how this goes. So here's what Chat GPT, in the role of a
user is telling me. The layout of the app is what I find particularly confusing. I can't seem to find where I need to enter my destination. This has caused me
to waste a lot of time trying to figure
out how to book or ride. It's frustrating, to be honest. Now, this is something an
actual user would say. Keep in mind, this is not data. This is just a practice. So let's practice
our active listening and open ended questions. So it's a good idea
to acknowledge the frustration that
the user has voiced. And then I want
to know what does good look like for them?
What are their expectations? So I'm just asking I can see how this would
be frustrating. Which other apps have you
used that do that well? A Che HiPT in the role
of a user tells me, I've used Right Zoom's app, and it has a much more
user friendly interface. Everything is laid out clearly, and I can quickly find what I need. Thank you
for sharing that. Which aspect of Taxi Corp's service are working
well for you? Well, it sounds like you
prefer Tax Corp over other services for their
cleanliness and service. And if the app had a
clearer interface, it would be your preferred
Taxi app, is that right? Okay, now, we've had a simulated conversation for ten questions, and Chat GBT has provided
some feedback. Let's see. You asked thoughtful
follow up questions that were directly related to
my previous responses. Your overall engagement
was excellent, and you were able to
clarify and confirm your understanding of my answers which made me feel heard. That's great. Love getting positive reinforcement.
Thanks, Chat GBT. The progression of
your questions led me to dig deeper into my
experience and feelings, which would be valuable
in real research setting. Now, this is really
useful to now. However, there was a point
where a question felt slightly repetitive and
it's giving me an example, which might have been more
beneficial if it were framed differently to explore
another aspect of the this is really good. I had some positive
view enforcement. Now I can actually
go out and feel confident in conducting
these interviews. And Chat GBT has given
me the little hint on, like what I can improve, which is great because
every interview is a learning opportunity. One last habit that makes you better fast a two
minute debrief. Right after each interview,
capture what you learned. What surprised you,
where you might have led the participant and three
follow ups to try next time. It's simple, but it
compounds quickly. And now it's your turn to put
these concepts into action. Write your moderation guide. Use what you've learned to craft a comprehensive moderation guide for your project and then
conduct an interview. Whether it's with a colleague, friend, or a volunteer, use your moderation guide to conduct at least one interview. Focus on open ended questions
and active listening. And please feel free to
share your experiences and any insights that you gain from this process
with the class. As you can see, AI is more
than a tool for automation. It's a companion in
the creative process, helping you to craft and
conduct meaningful interviews. Remember the real
magic happens when human intuition and empathy meet technological innovation. Happy interviewing. Good luck, and I'll see you in
the next lesson.
8. Analyze interviews with AI: Come back. This is where we take a messy conversation and turn it into something we
can actually use, and we're going
to do it with AI, but we're going to do
it with our eyes open. Here's the core idea. We analyze one interview at a time and that's not because
you're a beginner. That's because it's robust. When you try to analyze all interviews at once,
two things happen. You lose the thread of what this participant
actually said and the AI starts averaging everything into a smooth story that sounds
true but isn't traceable. That's why we go
interview by interview. We make the evidence clear. And then we synthesize across
interviews afterwards. You do a cross
interview pass first? Sure. Sometimes it's
useful for a quick scan. But if we care
about credibility, we don't skip over the
one interview pass. So here's today's workflow. We start with data
handling and reduction. I'm going to talk about
transcription options. There's a small interview
tip I'm going to give you. Then we move from transcript
to evidence table. We can ask AI
analysis questions, and I'm going to give
you some guard rails. We'll talk about confidence
and limitations and how to reset when AI drifts.
So let's get into it. We do anything clever, we do the boring grown up part,
not because it's fun, but because it's about
how we avoid harming people and breaking policies or getting ourselves
into a mess. So here's the reality.
Sometimes we're not allowed to paste interview data
into a public AI tool. So first, we check the
rules in our context. Is it client work, employer
policy, sensitive domains? Just don't risk it. And then we redact. I want this to be
doable, not scary. Reduction in practice usually looks like find and replace, plus a quick sanity skin. For example, replace real
names with participant IDs. So Sarah becomes
participant one. Alex becomes participant two. Then replace specific
organizations and products. So ACM Bank becomes Bank A. Internal tool X
becomes Internal Tool. Remove anything that can
identify someone directly, emails, phone numbers,
addresses, and so on. And then do a second pass
for the sneaky stuff. Just details that could
still give someone away, like the only
midwife in our town or I'm the head of design at that startup
that everyone knows. My favorite low effort
habit is make a copy of the transcript called transcript redacted and keep the
original somewhere secure. Only work with a redacted
one when you're using AI. And finally, we make a
small context packet. This is the bit that you
paste into new frets, so the AI stays grounded. It's short, it's clean, and it includes a one
paragraph project summary, your research objectives, the participant profile
in general terms. What output you're trying
to create, no names, no secrets, just enough context for the analysis to make sense. Okay, let's talk
about transcription. Here's the good news. Most of the time, you don't need a fancy
transcription setup. You can usually start with whatever you used to
record the interview. Zoom often has transcripts, and Google Met can also generate transcripts depending
on your account. And if you're being
scrappy and lean, you can also record
with something like iPhone voice memos or the
voice memos are on your phone. That's a surprisingly
decent option for early stage research. You won't get speaker labels, but you can still
get usable quotes. So what we care about
is not the best tool. We care about whether
the transcript is usable. You tell
who said what? Is it accurate enough that
quotes aren't embarrassing? Can you export it
without a pain? And here's an interview tip. This one is one of
those small skills that makes the rest of your
workflow feel easy. If we know that we're
going to analyze later, we interview in such a
way that makes good text. So instead of letting the participant give
one word answers, we pull for full sentences. And that's not
because we're strict. It's because later
when we're looking for evidence, yeah, it's useless. And in usability moments, we narrate what happened.
Here's an example. So if the participant says, I didn't click that
as a researcher, we can narrate for
the transcript, what stopped you from clicking
the delete plan button. And now we've got a transcript that contains the
thing they avoided, the reason, and the
consequence that they feared. That's a quote you
can actually use. Now the bit that
keeps us honest. When people say AI analysis, what they often mean is
paste the transcript, ask for insights, and
hope for the best. And yes, you'll get an answer. It just might be a beautiful paragraph that you can't defend. So we do something
slightly more disciplined. We build an evidence table. If you've never done this
before, here's how it works. We're making a little
bridge between the raw transcript and the insights that we're
willing to put our name on. And the rule is quote
first, meaning second. So a row in an evidence
table might look like this. I love about this is
that it's simple. It doesn't require
a special tool, and it stops you
from accidentally writing a report
that's 90% vibes. Also, notice what
we're not doing yet. We're not trying to summarize the whole interview
into a grand theory. We're just collecting
the pieces of evidence that will matter later. Okay, so first of all, I'm opening a new text document. So I'm using text
edit here on the Mac. You can use a text document on your PC if that's
what you're using. And now I paste in the prompt
from the prompt sheet. And you can see here there's some space for
our contacts packet. So the contact packet
that we prepared earlier, I'm just going to paste
that right in here. Just checking over
this. Yep. And then let's just focus on the first
research objective for now. Here you can insert some
questions that you have. So for now, let's just
focus on a simple question. What are the key motivations for this customer to choose taxis over right share apps. Just checking this over.
And yes, that looks good. And now, I'm just
going to select all of this and bring up chat
GPT in a new window just to make sure that there
is no existing context. I'm going to paste
all of this in here along with our transcript, which I've redacted before. I just checking it all over. Everything is there. Okay,
let's work the magic. And that could take
a little while hGBT is now reading our transcript and the
prompt, and here we go. It is creating our
evidence table. And now I can go through
this table and can really look at whether I agree with this or what of this is useful and then
take it over and copy it straight into my
working document where I'm putting
together the evidence. Now we can use the
AI like what it is, a fast assistant that
can read and summarize. It's totally fine to
ask questions like, how did this participant do X? What did they struggle
with? What confused them? What did they expect to happen? As long as we add
guardrails that make the output
usable in synthesis. So here are my
favorite guardrails. Answer in a table. Every claim must include
a supporting quote. Separate what happened
versus what it might mean. And if the evidence is
weak, say that explicitly. Yes, you can have a bit
of fun. You can ask. What would you rename
these features based on the participant's
mental model? Or write the frustration
moment as a one line story. Just don't confuse this
creative output with evidence. It's just an aid for
thinking, not a finding. Sometimes you'll
hear researchers say we coded the interviews. That can sound very academic, but coding or tagging
is basically just this. We give little labels
to pieces of what people said so we can
see the patterns later. If a participant says, I didn't trust it
with my bank login, then a simple code or
a tag might be trust. If another person says, I was scared that I
delete everything. That might also be trust or maybe fear of
irreversible action. So the point isn't to be fancy. The point is to stop
your brain from going, I feel like trust was
a thing and instead, be able to say, Cool, trust came up seven times, and here are the receipts. Now, here's the bit that
people don't tell you. You don't need a perfect
coding or tagging taxonomy. You don't need 100 tags, and you definitely don't need
to turn this into a hobby. These days, for a lot of teams, the useful question is, does this label help me find the evidence again and
use it in a decision. So we keep coding lightweight. Sometimes it's literal tags. Sometimes this is just a
column like candidate insight, design issue, open question. There's a tool that
I use that is really great for analyzing
qualitative data and organizing things visually in a way where you can affinity map and where you can see
patterns emerge in your data. It's called condensed.
It's a really useful tool, and I highly recommend using it. To actually just look at
all my tags, in this case, I'm really curious about what
our customers are saying about how they're
making the decision to book a taxi or a ride chair. Given every relevant quote a tag here called decision
making process, and I can look at all the tags about the decision
making process. I can just select all the quotes that are relating to the
decision making process. Pick on this button,
which allows me to copy all the quotes as text. If tagging helps you retrieve
quotes later, do it. If it starts eating your life, skip it and focus on
the evidence table. The goal is not a
perfect taxonomy. The goal is traceable reasoning. Alright, confidence. When we put a confidence
level next to a finding, we're not trying to
sound scientific. We're just doing
something much simpler. We're telling the
truth about how solid the evidence is
because in research, there's a huge difference
between this person had a rough moment and this is a reliable pattern that should influence
product decisions. Confidence is our way of
signaling that difference. So here's a human way
to think about it. When you read a
quote, ask yourself. Did they say it clearly? Did they give a
concrete example? Did it connect to the thing that we're actually researching? If the quote is specific
and unambiguous, that's already a good start. And then ask is this just one moment or did it
show up more than once? That more than once
could be the participant repeated it in different
words in the same interview. Or you've heard the same thing
from other participants, or you can see something
similar in behavioral data. So drop off rage clicks
or support tickets. So a simple confidence scale can be high confidence
means this feels solid. The evidence is
clear and it's not hanging on one fragile quote. Medium confidence
means this looks real, but I want to confirm it. Might depend on the
participant type, the scenario, or the
specific flow we tested. Low confidence
means interesting, but I'm not ready to build
decisions on it yet. It could be a misunderstanding, but one off or just
not well supported. And here's the secret weapon. If you want to sound credible
without being boring, add one more sentence, what would raise confidence? For example, we hear this from two more participants,
this becomes high. If analytics shows drop off at this step,
this becomes high. If we test the revised UI and the confusion disappears,
we can close this. Limitations are just
your honesty clause. They're the reasons. A smart stakeholder shouldn't overgeneralize
like small sample, transcript quality,
unusual participant, very specific context. Stating limitations doesn't
weaken your research. It stops someone else from misusing talk about
the moment where AI goes from helpful assistant
to why are you like this. If you ever found
yourself thinking, No, that's not what I meant. Why are you being
vague? Welcome. That's normal. The
tricky bit is that the tool is good at
sounding confident. It can feel like you're in a conversation with
a stubborn person. And that's where people lose time because we start
trying to convince it, but it's not a person. It doesn't have
context the way we do. It doesn't get embarrassed. It doesn't suddenly become more careful because
we told it off. So when it starts drifting, we don't argue. We reset. And when I say drift, what I mean is this the AI has wandered away from the thing
we actually asked for. It forgets the objective. It starts filling gaps. It gets smooth and generic. So resetting is a small ritual that brings it back to useful. Here's the move.
Start a new Fred, then paste the
contact packet again. State the single objective
in one sentence. Tell it the format that
you want and add the rule. Every claim needs a quote. You'll be shocked how
often that fixes it. So yeah, arguing is
emotional labor. Resetting is just good tool use. Alright, let's make this real. Don't go and analyze
15 interviews. Don't build a massive system. Just do one clean wrap. Pick one interview transcript. Pick two research objectives. Now build yourself an evidence
table with the help of AI. Let's aim for about
ten quotes total. For each quote, we want one sentence on what it means
and a confidence level, so low, medium or high. Then do the part that turns it into actual research, right, two candidate
insights, and under each one pass the two to
three quotes that support it. If you can do that, you basically
learned the core skill. Everything else is just scaling it up and staying
honest while you do. Good luck, and I'll see
you in the next lesson.
9. Synthesize insights with AI: Have come to the crucial stage of synthesizing your findings. You've done the hard work
of conducting interviews, transcribing, analyzing,
and identifying key themes. Now it's time to pull
everything together into actionable insights
and opportunities that can drive your
project forward. But first, a quick orientation. Right now, we're in
the synthesis stage. We're moving from patterns in the data to insights
we can act on. And there's a rule here
that keeps us honest. AI can suggest interpretations
and opportunities, but we decide what's true by checking it against evidence. Here's a simple ladder I use
to keep synthesis clear. A theme is what keeps
happening across interviews. A insight is why that
theme matters and what it tells us about
people's needs or behavior. A opportunity is what we could change in the
product or service. For every insight, we
want to attach evidence. That's quotes and examples, plus a confidence level. Here's one watch out with AI. It tends to smooth things
out into a neat story, but real research is messy. So we want to make sure that
we keep contradictions. We avoid fake certainty
and we make sure every opportunity maps
back to real evidence. Now you've dissected
each interview, breaking them down into themes, patterns, and key insights. The next step is to analyze these findings across
multiple interviews, look for commonalities and differences, contradictions,
and surprises. This is where AI excels. Connecting these dots
will help you uncover a holistic understanding
of your user's experience. Insights are the aha moments
that arise from your data. They are the deep
truths that reveal something profound
about your user. To articulate these insights, you must interpret what
the data is telling you. You'll need to go beyond the
obvious and ask yourself, what does this mean?
Why is it important? Crafting well phrased insights will guide your design decisions and ensure that they are deeply rooted in
user understanding. GBT can assist in translating raw data into profound
truths about your users. You can prompt it to interpret
the themes and findings, probing deeper into what does it mean and why is it important. With clear insights in hand, you are now ready to
identify opportunities. These are the areas where you
can make a real difference, solve a problem, or create a delightful experience
for your users. Brainstorm with your
team or use hGBT to help generate innovative
ideas based on your insights. Think creatively and don't be afraid to challenge
the status quo. It's your chance to turn the insights into something
tangible and impactful. Remember, synthesis is not just a step in the
process, it's an art. It requires critical
thinking, empathy, creativity, and the ability
to see the big picture. As you move through this lesson, embrace the complexity
and allow yourself to deep dive into the nuances
of your user's world. When it's time to transform
insights into opportunities, AI can be a creative partner. Brains domini sessions can be enriched with AI
generated ideas. You can challenge the
model to come up with innovative solutions based on the insights that
you've discovered. It's an exciting
collaboration that can lead to unexpected and
valuable design paths. So let's take a look
at how this can work. Now, from your analysis, you would have an evidence table that might look
something like that. I've created it in
Google Sheets here, but you might have that in
notion or any other format. What's important is that we have some structured
data where we can see the participant ID,
so that's anonymized. The theme that we
have identified, the context in which
that appeared, a quote to back it up with, and what it means
in our own words. And then we ideally also
have a confidence level of how confident we are
in this evidence. Now, if I have that
in Google Sheets, then I can just
export this here. You go to download and
export this essay CSV, because that will
make it easy to then jump into something
like Chat GPT or whichever AI you're using and upload the data in the
same structured format. So let's jump into
Chat GPT here. So the first thing
I will do is I will paste our
contacts packet again. I have that open over here. So once again, that's the contact packet we've
been using all along. It's a recap of what our
brief is. It's sanitized. We stripped away any company
confidential information. And what we're including
here at the bottom is the synthesis outputs
that we're expecting now. So we want insights and
opportunities that are explicitly backed by evidence and include a confidence level. So let's copy this
into a new chat. And at the bottom here, I'm just going to make a few
new lines once again, and now I'm going to add
our synthesis prompt. You can find this synthesis
prompt in your downloads. Just go to the prompt sheet
and copy it from there. The prompt starts with you
and my UX research assistant. You have access to
the CSV I uploaded, which is an evidence
table with these columns. Your task is to focus
only on one theme. And the theme that we want to investigate first is the
booking floor friction here. So I'm going to grab the first theme from
the spreadsheet, and I'm going to
paste that in here. We want to look at one theme
at a time so that the AI doesn't smooth over
too much and it doesn't get entangled
in everything. So taking one theme at
a time allows us to really sense check the output and make sure that there
is evidence attached, make sure that it makes sense, and that it is actually correct. Now, for the output, I'm asking for an inside card. That is essentially
a mini report just for this one theme. And the structure we're looking for here is, know the theme, what we heard the
insight in one sentence, then we want evidence
attached to this. So three to six quotes from the evidence table with participant IDs so that we can cross check it
against the evidence. Then we want the AI to
identify some opportunities here based on these insights
across all interviews. The opportunities usually
start with a verb, so there are things that we can do or things that
our client can do. They're not feature sets they're more things
they can change and things they can act also want the AI to highlight
any contradictions, if there are any and give us
a confidence level and why. That's going to be important
because we want to stand behind the research that
we're conducting here. And using AI as our assistant, we need to ask AI to give us reasons why it
does what it does, or why it makes the assessment. And there's a few rules here. So we only want to use the calls from the CSV
and not invent anything. So just like what
we've done all along. Now, let's add our data here. Now I've attached
the evidence table, and now that we have this prompt narrowed down to one theme,
let's see what comes back. Okay. And here is our
first insight card. So for our booking
flow friction, what we heard is
that users expect the booking process to be
fast, obvious, and reassuring. When steps feel unclear
or feedback is missing, they hesitate or
switch to write share. Our one sentence inside is even small moments of
uncertainty in the taxi apps booking floor quickly
erode confidence and make chair feel like the
safer faster choice. And the evidence for that is from these three participants, one saying I just wanted
it to be two tabs, but I wasn't sure what
I was confirming. Second one said with chair, I can see the driver and
the ETA straight away. Here, I'm kind of
waiting and hoping. And the third one says
I wasn't sure if it had my pickup spot right and I didn't want to end up
on the wrong street. Opportunities that
Chat GPT has now identified from this is to clarify each step of
the booking process, so users always know
what action they're taking to provide immediate
visual feedback after key inputs such as
pickup location and confirmation and to reduce the number of decisions or fels required before a
car is assigned. It didn't find any direct
contradictions in this theme, and the confidence label is medium because the
codes are consistent, but they come from only three participants within this theme. It's a very good point,
and that is something that we should
highlight to our client here to say that what would raise confidence is to conduct additional interviews
that can specifically focus on first time use versus repeat taxi abusers or do some behavioral analytics on the booking drop off points
within the current flow. And now I can take this insight and move it into
my own document and then continue this analysis with the second theme from
our evidence table. Once I work my way
through all the themes, then I can create a
report from that. Then I can decide
which of them are the key insights, which
of them are relevant, and that's really where I
come in as a researcher with my expertise and my
own viewpoint on this. So we don't just let AI determine what's important
and what's not important. This is still in the
hands of the researcher. So take these findings
and think about how you can apply
them to your project. How will you see these insights to guide your design choices? How will you turn
these opportunities into innovative solutions? This is the heart
of Yog's research where understanding
meets innovation.
10. Ethical considerations when using AI: This is about trust. Before we get into tactics,
let's set the tone. Using AI in research can
be genuinely helpful. It can also go wrong in very
boring, very human ways. A participant shares
something personal, and it ends up somewhere where
it shouldn't stakeholder reads an AI summary and
assumes it's the truth. A quote gets cleaned
up a little too much, and suddenly it's not
really a quote anymore. So, this isn't an
ethics lecture. This is a set of habits that protect people and
protect your work. And we're going to
keep it practical. We talk about what to share
and what not to share. What to disclose and how we say honest when
we're moving fast. So hands on the steering wheel, AI can help, but we decide what makes it
into the research. Here's a simple mental model. What can go wrong? To keep this practical, we're going to use a
few plain categories. It's not just because we love
frameworks. I mean, we do. It's because it's easier to spot risks when
you have buckets. So here are the big ones
privacy and security. Are we exposing personal
or confidential data? Transparency. Are we
being honest about AIU? Fairness. Are we missing or distorting certain
groups experiences? Accountability.
Who is responsible if something is wrong?
Human oversight. Are we letting the tool
decide or are we deciding? If you can remember
those, you'll catch 90% of the problems
before they happen. Let's start with the one
that causes real damage. Ah. Here's a rule that'll
save your career. If you wouldn't
post it on red it, then don't paste it
into a random AI chat. So here's some examples
of what not to paste. Names, emails, phone
numbers, and addresses. You'll hear the term PII
used a lot in the industry. That means personal
identifiable information. So don't paste raw session links that include someone's name, or any medical, financial, or highly sensitive
personal info. No clients strategy,
unreleased roadmaps, or internal
intellectual property. Just don't paste anything
that your company or your client has not approved
for external processing. Yes, you can still use AI. You just need to work with
a redacted transcript or a safer tool. And this is why the redaction habits from Lesson six matter. Now let's talk about disclosure. Disclosure is where
people get weird. They either hide AI use because
they're nervous or they overshare in a way that makes participants feel like they're
being watched by robots. Let's make it simple.
Participants need to know. Will AI be used at all? How will it be used during
the session or after? Will they interact
with AI directly? How is the data being protected? Can they opt out
of AI involvement? Will their data be
used to train models, say yes or no, clearly. Also, a quick reality
check telling people AI is involved
can change behavior. Some people will trust you less, some will perform, and
some will hold back. So let's frame it in a
calm and reasonable way. AI is used to save time on admin tasks like
transcription. A researcher reviews
everything. You can opt out. And stakeholders need
something different. They need to trust the work. So we include a short block in the report. What
tools were used? What the AI did and
what humans checked. No drama, transparency. Now let's talk about bias. AI has a personality. It likes things neat. It likes things average. And if you're not careful, it will turn a messy set of human experiences
and turn it into one smooth story that feels true while quietly
sanding off the edges. And that's what we mean
by average washing. So let's make it concrete. Imagine you interviewed
five people for an onboarding flow.
People said it was fine. Two people said
it was confusing. One person said,
I didn't do it at all because I thought it would share my data
with my employer. A lazy AI summary will often
come back with something like users found on boarding confusing and
wanted clearer guidance. And that's not wrong, but it's also not where
the real risk is. That one person's fear
about data sharing might be the difference between
a minor UX tweak and a trust disaster. Example two, the default
user that it invents. If you don't tell the AI
who your participants are, it fills in the blanks. It will sometimes assume the default user
is confident with tech from the majority culture and using the product in
a fairly standard way. So if you're researching a
group that is not that say, accessibility needs, low digital confidence or a niche workflow, then
you have to name it. Otherwise, the AI will
gently flatten it. So here's the habit we build. Every time AI gives you a tidy summary, we do
a quick follow up. Cool. Now show me
the contradictions. So we ask, who had a
different experience? What didn't fit?
What surprised us? What would be easy to miss if
we only looked at averages? And we keep coming
back to quotes because quotes are
where the nuance lives. Let's talk about
human oversight, how we stay responsible
without becoming paranoid. This is where we keep our
hands on the steering wheel. In practice, human
oversight just means we're clear on what the
AI is allowed to do for us. So it can help us
draft the first pass, organize messy notes, summarize
a chunk of a transcript, and suggest possible
interpretation. What it can do is
take responsibility. That part stays with us. So here's a simple way to work before anything
leads to a laptop. So before it goes
into a report or a deck or slack message
to stakeholders, we do a quick integrity check. Asked, Are the codes real
and copied accurately? Can we point to evidence
in each insight? Did we accidentally turn
a hypothesis into a fact? Did we write down
any limitations so someone doesn't
overgeneralize? And when the stakes are
high, we make it social. We ask someone else
to read the outputs, not because we're panicking, but because it's
genuinely hard to spot your own blind spot when you've been
deep in the work. So, this is the vibe. We use AI to move faster, and then we use human
review to stay honest. Reporting integrity. And this is the part where
we stop being cute. If a report includes
made up quotes, it doesn't matter how well designed the slide deck
is. That's not research. It's just storytelling,
and it damages trust fast. So here's the standard
we hold the quote rule. If it's in quotation marks, it came from a participant. Word for word. If you paraphrased it, don't put it in quotation
marks. Write it as a summary. The composite participant trap. Sometimes people try to
be helpful by combining a few participants into one
clean persona like story. That's fine for
internal sense making, but it's not fine if you present it like a
real person said it. So if you ever do a
composite, label it clearly. Otherwise, don't
the evidence trail. The easiest way to
keep this simple is every insight gets
a little anchor, a link to notes, a clip, or a set of quotes in
your evidence table, not because stakeholders
will click every link, but because you should be
able to the AI usage log. And this is the bit
nobody wants to do. And it's the bit that quietly makes your work
feel professional. An AI usage log is just a
small record of what happened. It answers questions like
what tool did we use? What did we ask it to do? Did we feed it raw data
or a redacted version? Did we check it afterward? What did we change? Think
of it as a memory aid. And if a client ever asks, How did you analyze this? You don't have to rely on vibes. Also, it protects you, because if the AI
output was wrong, you can see exactly where
it entered the workflow. All right. Here's where we
actually build the habit. Take your current project
and do two small things. First, write your
participant disclosure. Keep it short and human. Imagine saying it out loud at
the start of an interview. Second, start your AI usage log. Just one entry tool, task, what you checked. If you do those two things, you're not just aware of ethics. You're practicing it. Okay.
See you in the next lesson.
11. Leverage AI for creating a research report: This brings us to the final step of the UX research journey. Here we'll explore the
essential steps to create an impactful
research report. And, of course, we'll leverage hat GPT to streamline
the process. So just in case you're
lost, a quick orientation. Right now, we're at
the reporting stage. This is where we
turn our synthesis into something stakeholders
can actually use, and this is where AI
can save you time, but also where it can
cause real damage if you let it invent certainty
or invent quotes. So we'll use AI to draft, but we'll keep report
integrity tight. Start by gathering
all the insights, themes, and opportunities you've uncovered in the
previous lessons. Organize them in a logical flow that tells the story
of your research. The executive summary is a crucial part of
any research report. It encapsulates the key findings and recommendations
in a concise manner. We can use Che GPT to draft an initial version of
the summary by providing the key insights and
asking the model to summarize them
in a few sentences. The tone of your report should be aligned with your audience. So whether it's a technical audience or business
stakeholders, the language should
resonate with them. GPT can assist you
in this refinement. You can prompt the model to rephrase sections
in different tones, such as formal,
casual or technical. And remember to include visual representation and data
to support your findings. Charts, graphs, or quotes from participants can add
credibility to your report. A simple rule of thumb, let AI help with
structure and wording. But you own what's true. You own the evidence,
the quotes, and the confidence levels. If the model can't
point to evidence, it doesn't go in the report. Okay, so let's look
at how to generate an executive summary and then refine the
tone of the report. So I'm creating a
new threat once again just grabbing the
prompt from the prompt sheet. Which is to help us generate a UX research report or more precisely an
executive summary. So here, first of all, we want to paste a little
bit of context again. So once again, our
context packet just to ground the model. And then in the promptier, you will see some space to paste three to
five inside cards. And these are the
insights that you've generated in the synthesis step. So I will now paste
them in here. These are about
three inside cards. And then the task is to
draft executive summary for senior stakeholders and just
notice the guard rails. So I'm telling the model
to use only what I pasted and not to invent
quotes on new findings. And once it drafts the
executive summary, I treat that as a
starting point. So then I can do a quick
review, just to check. Are the claims actually
supported by the inside cards? Did it overstate confidence? Did it ignore contradictions?
So here we go. This is our executive summary, and we have some key points
that it put together for us. And now what I can do is I can rewrite this for
a specific audience. So for example, I can
rewrite this for executives, so it would be short and outcome focused or for the product team, so it would have more nuance or for customer
support or operations, so it would be more
practical next steps. So using this prompt, rewrite the executive
summary for this audience. So let's say we want to rewrite
it for the product team. Here we go. And let's see
what that would look like for the marketing team. So this is a huge timesaver, but the integrity
rule stays the same. AI helps with
structure and wording, and the evidence stays human on. And you can just continue
having a back and forth with ChaiBT to create the
sections that you need, and you can specifically
ask about specific insights as well to flesh out the different sections of
your research report. So wherever you create
your research report, there might be a
slideshow or a document. You can just copy and paste these sections over and
start writing from there. Here's the structure I use
for most research reports. Start with an
executive summary that gets stakeholders
up to speed fast. Then a short paragraph on
objectives and method, enough so they
understand the scope. Next, come the themes. This is what we heard organized
by topic, then insights. This is where you explain
why each theme matters. After that, opportunities. What should we actually change? Keep these actionable, not
just feature wish lists. Then risks and limitations. Every study has them,
so call them out. And finally, an
optional appendix where you can put your
evidence table or quote bank. This gives stakeholders
a clear path from findings to action. Full traceability
back to evidence. So your research report is not just a collection of facts, it's a narrative that leads
to actionable insights. By using AI tools like
Chet GBT responsibly, you can accelerate the process without sacrificing the
quality of your report. For your project action, I want you to create a draft
of your research report, generate the executive summary, refine the tone of voice, and bring all the
findings together in a cohesive and engaging manner. Whatever format you choose. Remember, a well crafted report can be a catalyst for change, driving improvements
in user experience and informing
strategic decisions. So let's make your
research count.
12. Final words: Alright, that's us. If you're watching this, you earned it. You didn't just consume content. You actually did the work. You turned a messy question
into a research objective. You spoke to users, made sense of what you heard, and turned it into
something useful. That's a real career
defining skill, and you're not doing it alone. You're part of a growing
group of people who are learning how to do UX
research at a modern pace. Using AI as support while
keeping the quality bar high. Remember the vibe, we keep our hands on the steering wheel. AI is the satnav. It
can speed things up. But we decide where we're going. As you take this into the real world, two
quick reminders. First, tools will change. The principals
won't whenever you switch tools or
big update drops, rerun your tiny benchmark. Sanity check the outputs and keep the parts
that actually help. Second, keep it clean
and professional. Protect privacy, get consent, and keep an evidence trail so you can actually
stand behind your work. And please don't disappear. If you've got questions, as you apply this
on real projects, please drop them
in the comments. If something still feels fuzzy, ask. There are no
silly questions here, and I read every single comment. Also, I'd love to know what
you're using this for. What are you researching next? And what are you building
with the insights? Thanks for learning
with me. If the course helps, leave a review. It genuinely makes a difference.
I'll see you next time.
Pascal Raabe, Coaching and UX Design
