How to Find the Next Big Idea: Deductive vs. Inductive Thinking in Product Research

How do you figure out what features to build into your design?
How do you get those magical insights that actually improve your product—versus just shifting things around?

In this episode, we unpack one key distinction that helps design psychologists and UX researchers choose the right method at the right time: inductive vs. deductive research.

Imagine you have two different ideas for how to design an app for restaurant waitstaff. You think of adding some possible features, like a picture-based layout, or a list of incoming customers.

 So—do you give the waitstaff a prototype of each app version and see which version performs better (deductive research)? Or do you systematically observe the actual waitstaff in action before even deciding which features to build (inductive)?

This choice is about more than methodology—it shapes the kinds of insights you get, and how impactful your design ultimately becomes.

🔍 You’ll learn:

• When inductive research unlocks hidden insights you didn’t even know to look for
• Why deductive research is great for making clear decisions—fast
• How your design phase should guide your research method
• What to consider when you're short on time or budget
• And how to avoid a common trap: testing too early

By the end, you’ll know how to orient your research approach based on where you are in the design journey—so you can uncover insights that actually move the needle.

Transcript

WEBVTT

00:00:00.017 --> 00:00:04.897
How do you figure out what features to build into your design?

00:00:05.237 --> 00:00:12.617
How do you get those magical insights that guide you to designing and building

00:00:12.617 --> 00:00:17.277
a hit product? You know, no one knows how to build one of these hit products.

00:00:17.277 --> 00:00:18.957
It's a mystery how they come together.

00:00:19.057 --> 00:00:24.897
But there are insights that researchers are doing somehow along the way. How are they doing it?

00:00:25.317 --> 00:00:30.857
So picking which research activity to do can be very complicated and confusing,

00:00:30.857 --> 00:00:34.977
especially if it's not your particular area of training.

00:00:35.697 --> 00:00:40.897
But are there ways that we can think about research that makes it easier for

00:00:40.897 --> 00:00:43.157
us to wrap our minds around it?

00:00:43.297 --> 00:00:49.757
So this episode is about that, the underlying logic that can help us foresee

00:00:49.757 --> 00:00:52.537
the right research method to pick next.

00:00:52.737 --> 00:00:56.737
To help orient our thinking here, I'd like to start with a concrete example.

00:00:57.077 --> 00:01:01.777
Picture a busy restaurant where waiters are running around between tables,

00:01:02.117 --> 00:01:06.237
jotting down orders and keeping track of who needs more service.

00:01:06.457 --> 00:01:09.497
So there's lots of potential confusion in the mix.

00:01:09.677 --> 00:01:16.237
Now imagine that you and your design team are designing an app to be used by

00:01:16.237 --> 00:01:22.257
this restaurant's wait staff, and you're working with one particular restaurant to launch this app.

00:01:22.757 --> 00:01:27.937
You sit down for some meetings with the manager and the restaurant owner,

00:01:27.937 --> 00:01:34.777
and you begin to realize that the manager has one idea about how the app should look and behave,

00:01:34.957 --> 00:01:38.197
while the restaurant owner has a different idea.

00:01:38.637 --> 00:01:43.077
Imagine you're designing a tablet app for a restaurant staff.

00:01:43.297 --> 00:01:48.337
Picture a busy restaurant with waiters juggling orders and keeping track of tables.

00:01:48.697 --> 00:01:55.977
Imagine the manager and the restaurant owner offer contrasting ideas about how

00:01:55.977 --> 00:01:58.297
to build the app and how it should come together.

00:01:59.037 --> 00:02:04.177
The manager argues that the app should have pictures of tables and that this

00:02:04.177 --> 00:02:09.197
would give waiters a quick visual cue of the tables that need assistance.

00:02:09.417 --> 00:02:13.237
While the owner, on the other hand, suggests a different approach.

00:02:13.437 --> 00:02:19.657
They want the app to have a list of guests sorted in order of the guest time arrival.

00:02:20.317 --> 00:02:23.657
You, as the designer, decide to compare

00:02:23.657 --> 00:02:26.597
these suggestions against each other to figure out which

00:02:26.597 --> 00:02:30.437
one is the best way to build the app comparing the

00:02:30.437 --> 00:02:36.337
options against each other would be using a deductive research method deductive

00:02:36.337 --> 00:02:41.457
research starts with a hypothesis and you design an experiment to see which

00:02:41.457 --> 00:02:47.257
option performs best but what if you're not sure whether one of the two suggestions

00:02:47.257 --> 00:02:50.317
is right at all what if it's something different altogether.

00:02:51.361 --> 00:02:56.041
Instead of choosing one of the two suggestions, you might decide to show up

00:02:56.041 --> 00:03:01.621
at the restaurant and systematically observe how the waiters interact with customers.

00:03:01.941 --> 00:03:07.041
You gather data about the restaurant and how it operates until you can step

00:03:07.041 --> 00:03:09.061
back and look at the full picture.

00:03:09.281 --> 00:03:14.721
That's inductive research. You're building an understanding based on observation.

00:03:15.021 --> 00:03:20.361
In today's episode, we're diving into one of the most critical questions in

00:03:20.361 --> 00:03:27.101
our field. How do you select the right research method to get the best results for your design?

00:03:27.561 --> 00:03:34.441
When is it better to test competing ideas versus when is it best to take a step back and observe?

00:03:34.821 --> 00:03:36.881
Today we'll unpack these questions.

00:03:37.541 --> 00:03:42.661
We'll explore when it's better to conduct research using a broad exploration,

00:03:42.941 --> 00:03:47.381
inductive, versus using targeted testing, deductive.

00:03:48.021 --> 00:03:50.881
Deduct means to remove. here's a finite set

00:03:50.881 --> 00:03:54.181
of options we're going to get to the truth by

00:03:54.181 --> 00:03:57.101
saying what it isn't it's not option

00:03:57.101 --> 00:04:02.121
c because we did this experiment and it really couldn't be option c and have

00:04:02.121 --> 00:04:06.461
this theory still be true and it's not option a either the only one that seems

00:04:06.461 --> 00:04:12.361
supported is option b so it's using a deductive method you're removing things

00:04:12.361 --> 00:04:15.801
then an inductive you're adding things you're saying,

00:04:16.121 --> 00:04:21.781
okay, I'm going to add a whole bunch of examples, stacking pebbles and building a pyramid out of it.

00:04:21.941 --> 00:04:28.241
And then now I've added enough to where I've reasoned my way to the truth that way.

00:04:28.401 --> 00:04:33.421
Knowing the difference between when to do deductive versus inductive research

00:04:33.421 --> 00:04:39.121
is one of the keys to designing more innovative solutions that are optimized

00:04:39.121 --> 00:04:42.621
for a better user experience. So let's dive in.

00:04:43.448 --> 00:04:51.148
So this comes from Sir Francis Bacon, who is considered the father of inductive reasoning.

00:04:51.148 --> 00:04:58.888
He wanted a move away from sort of a sterile deductive thinking process that

00:04:58.888 --> 00:05:04.768
was common during the time, and he wanted a return to or a shift toward observation of nature.

00:05:04.988 --> 00:05:12.388
And his scientific method involved lots of piecing things together and getting a picture of reality.

00:05:12.388 --> 00:05:18.368
So we have different scientists in history who may have steered more or less

00:05:18.368 --> 00:05:20.388
toward one direction versus another.

00:05:20.708 --> 00:05:25.468
All scientists are doing some of both of these, but some scientists are more

00:05:25.468 --> 00:05:28.928
notable for kind of one method versus the other.

00:05:28.928 --> 00:05:35.868
So if you think, for example, of Charles Darwin, Charles Darwin put together

00:05:35.868 --> 00:05:42.188
his theory and published his research based upon a massive amount of observation.

00:05:42.188 --> 00:05:47.708
When he went to the Galapagos Islands, he's taking bird specimens and drawing

00:05:47.708 --> 00:05:49.988
the different types of beaks that they had.

00:05:49.988 --> 00:05:57.748
And he's looking at turtles and he's amassing this giant set of data, of observational data.

00:05:58.388 --> 00:06:05.188
And he's using that to paint a picture about how evolution happens and there's variants in genes.

00:06:05.628 --> 00:06:10.928
And he needed to really pile on the data because it wasn't really a new idea.

00:06:10.928 --> 00:06:14.308
It just needed a lot of support in order for it to take off.

00:06:14.965 --> 00:06:19.445
And there's other scientists in history as well, like B.F.

00:06:19.585 --> 00:06:23.765
Skinner, the famed behaviorist, one of the most famous behaviorists,

00:06:23.885 --> 00:06:29.725
who gathered tons of different data by doing lots of different experiments where

00:06:29.725 --> 00:06:33.745
he tweaked the conditions slightly as he's studying rats and pigeons.

00:06:33.745 --> 00:06:41.585
And he's trying to paint a picture or really generate knowledge about how the

00:06:41.585 --> 00:06:48.605
parameters of behaviorism show us how we can shape and steer behavior.

00:06:49.165 --> 00:06:54.065
So these are all inductive methods that involve gathering lots and lots of different

00:06:54.065 --> 00:06:56.125
data in order to paint a picture.

00:06:56.345 --> 00:07:00.685
But then there's the deductive method that's also very, very critical for science.

00:07:00.685 --> 00:07:07.105
This comes into play where you have two different theories, and you are creating

00:07:07.105 --> 00:07:10.665
an experiment that will support one theory over the other.

00:07:10.825 --> 00:07:16.045
And this is how science is able to advance and pick the right theories and hypotheses

00:07:16.045 --> 00:07:20.045
to move forward and build our understanding of what's actually happening.

00:07:20.385 --> 00:07:25.065
So how does all of this come into play for us as design psychologists?

00:07:25.425 --> 00:07:31.245
As a designer, you'll go through various phases from from uncovering user needs

00:07:31.245 --> 00:07:33.745
to testing and refining your product.

00:07:33.985 --> 00:07:40.745
So let's go through an example of a full design journey from observing users

00:07:40.745 --> 00:07:46.525
and conducting interviews to running A-B tests, usability studies, and so on.

00:07:46.845 --> 00:07:54.365
We'll explore whether each stage of the process utilizes deductive versus inductive research methods.

00:07:55.165 --> 00:07:59.645
By the way, for each one of these, I'll pause briefly after I ask the question

00:07:59.645 --> 00:08:01.665
to give you a chance to quiz yourself.

00:08:02.165 --> 00:08:08.405
So for our first example, when you immerse yourself in a population and you

00:08:08.405 --> 00:08:12.665
observe their behaviors systematically, what type of approach is that?

00:08:13.885 --> 00:08:20.105
That's an inductive approach. You're gathering data step-by-step to build a

00:08:20.105 --> 00:08:22.085
complete picture of the situation.

00:08:22.465 --> 00:08:26.185
Now let's think about interviews. What does that lean toward?

00:08:27.645 --> 00:08:33.605
In most cases, they tend to lean towards an inductive research approach.

00:08:33.605 --> 00:08:39.725
Instead of testing a predetermined hypothesis, you're systematically gathering

00:08:39.725 --> 00:08:44.285
information to paint a full picture of the user experience.

00:08:44.285 --> 00:08:49.345
Keep in mind, it's not just casual observation, it's methodically collecting

00:08:49.345 --> 00:08:52.085
data to understand what's truly happening.

00:08:52.965 --> 00:08:57.885
Okay, what about A-B testing? Is that deductive or inductive?

00:08:58.925 --> 00:09:02.225
This method leans toward deductive reasoning.

00:09:02.445 --> 00:09:08.105
You start with two or more predetermined options and you do your A-B testing

00:09:08.105 --> 00:09:12.785
to compare the options against each other and see which one performs best.

00:09:13.025 --> 00:09:18.885
And now let's consider usability testing. Does it lean more toward deductive

00:09:18.885 --> 00:09:20.985
or an inductive research approach?

00:09:21.665 --> 00:09:28.205
The answer for this one really depends on context, but generally usability testing

00:09:28.205 --> 00:09:30.765
aligns with a deductive method.

00:09:31.045 --> 00:09:36.445
With usability testing, you've got a hypothesis, even if it's an implicit one,

00:09:36.585 --> 00:09:41.885
and you're investigating a specific question. Does this ScreenFlow work.

00:09:42.105 --> 00:09:47.985
So how does this decision of whether to choose an inductive versus deductive

00:09:47.985 --> 00:09:53.465
research method actually help us design better products and services for people?

00:09:53.785 --> 00:09:57.745
Let's bring it all together using our restaurant app example.

00:09:58.563 --> 00:10:02.583
You are designing an app to help waitstaff in a busy restaurant.

00:10:03.143 --> 00:10:08.203
The stakeholders, the manager and the owner with their competing ideas,

00:10:08.583 --> 00:10:13.103
are probably expecting you to use a deductive approach.

00:10:13.403 --> 00:10:17.683
People in this position are often convinced from the outset that one of their

00:10:17.683 --> 00:10:19.383
proposed options is correct.

00:10:19.583 --> 00:10:22.923
They're experts in the area and they've spent a long time thinking about it,

00:10:22.983 --> 00:10:25.303
but they're not designers and they're not researchers.

00:10:26.023 --> 00:10:31.883
So starting with a set of options, then testing to eliminate options, that's deductive.

00:10:32.023 --> 00:10:35.823
But sometimes it's better to take an inductive route.

00:10:36.363 --> 00:10:41.623
And that might mean, in this restaurant case, that you do something like show

00:10:41.623 --> 00:10:46.883
up at the restaurant and systematically observe the waitstaff as they work.

00:10:47.083 --> 00:10:52.503
And after gathering and organizing your data and spreading it all out on the

00:10:52.503 --> 00:10:59.323
table and taking a step back, you look at the data and you recognize the critical

00:10:59.323 --> 00:11:02.063
opportunities for your design.

00:11:02.323 --> 00:11:04.963
This way you aren't locked into

00:11:04.963 --> 00:11:10.443
a limited set and assuming you've already identified the right solution.

00:11:10.723 --> 00:11:18.903
With this approach, you realize that you might uncover something entirely new during this process.

00:11:18.903 --> 00:11:23.863
And this might be something that sets your design apart from the competition

00:11:23.863 --> 00:11:27.583
or fixes people's problems in a way that hasn't been done.

00:11:28.043 --> 00:11:35.883
So how do you select the right research method to get the best results for your design?

00:11:36.103 --> 00:11:41.103
It all starts with understanding where you are in the design process.

00:11:41.343 --> 00:11:46.423
At the very beginning of your project, when you're still trying to grasp the

00:11:46.423 --> 00:11:51.863
full scope of your user needs, inductive research is often your best friend.

00:11:52.083 --> 00:11:58.103
By shadowing waitstaff in our example and observing how they work in their typical

00:11:58.103 --> 00:12:00.923
environment, you gather unbiased data.

00:12:01.714 --> 00:12:07.374
This broad exploration is untethered to your pre-existing ideas.

00:12:07.374 --> 00:12:10.954
It allows you to build a complete picture.

00:12:11.374 --> 00:12:17.974
As you progress in your design, you start to develop a set of potential solutions.

00:12:18.894 --> 00:12:23.694
Then you might enter a testing phase. If you have a clear hypothesis,

00:12:23.974 --> 00:12:30.154
for example, you want to decide whether images of tables versus a list of guests

00:12:30.154 --> 00:12:31.754
is better for efficiency,

00:12:32.174 --> 00:12:36.054
then deductive research may be the direction you take.

00:12:36.374 --> 00:12:42.994
Deductive methods let you test these specific ideas and validate which option

00:12:42.994 --> 00:12:45.234
truly works best for your users.

00:12:45.914 --> 00:12:52.814
So now that we understand how deductive and inductive methods fit into the design

00:12:52.814 --> 00:12:55.714
process, let's touch on budget considerations.

00:12:56.114 --> 00:13:01.114
When funding is tight, deductive approaches are often more practical.

00:13:01.114 --> 00:13:06.834
They allow you to focus on testing specific promising ideas.

00:13:07.194 --> 00:13:13.474
These approaches can sometimes help you avoid the time-consuming process of

00:13:13.474 --> 00:13:19.014
collecting extensive data, which often comes along with inductive research.

00:13:19.214 --> 00:13:25.514
By using a deductive approach, you can often find the most effective solution

00:13:25.514 --> 00:13:27.954
without stretching your resources too thin.

00:13:28.234 --> 00:13:33.194
And this is all said, of course, with some risk of oversimplification here,

00:13:33.374 --> 00:13:38.254
but these are commonly observed and recognized realities in industry.

00:13:38.594 --> 00:13:45.114
So now we've looked a little deeper into this inductive versus deductive dimension

00:13:45.114 --> 00:13:48.134
of thinking about research methods.

00:13:48.514 --> 00:13:54.894
Inductive is the knowledge expansion we get from adding data versus deductive

00:13:54.894 --> 00:14:00.054
is the knowledge pruning we get from removing what doesn't work.

00:14:00.274 --> 00:14:06.574
This is like a push and pull of exploring our reality, and we can use our knowledge

00:14:06.574 --> 00:14:13.614
of inductive versus deductive to guide our design research to discovering the best ideas.

00:14:14.074 --> 00:14:20.094
So whether you're just beginning your journey or validating a hypothesis versus

00:14:20.094 --> 00:14:22.294
working within a limited budget,

00:14:22.554 --> 00:14:28.054
I hope this clarifies one way to think about picking the right research methodology

00:14:28.054 --> 00:14:29.674
for your design project.