tl;dr
Defining qual as “why” and quant as “what” can lead you to choosing the wrong method for a problem. It’s more useful to consider qual as identifying the elements of interest and quant as establishing magnitude or prevalence of the elements of interest.

UX research as a domain has always included qualitative and quantitative practices, the role is naturally mixed-methods. As a field, we need to grapple with what distinction is between qualitative and quantitative approaches. It doesn’t just let us argue about which side is more valuable but better informs the method that is best suited for a given product problem. 

The distinction has been written about a number of times¹. Most authors primarily summarize it as  qualitative research is related to “why” and quantitative research is related to “what”.

These definitions never fit for me. Saying what questions the domains answer is a somewhat useful heuristic, but never quite seemed to be a “definition”. Beyond that, I can think of plenty of situations that directly counter answering “why” with qualitative research. 

What and why are poorly defined

What and Why are not able to fully explore all research questions along consistent methodology lines. These separators break down quickly when you go outside the bounds of basic descriptive surveys or generative interviews. The same component of a design or experience can be the why or the what, depending on your context.

Let’s start with a hypothetical to prove this point:

• Imagine you are working to redesign the navigation in a homepage. 
• You see quantitatively in the metrics people aren’t clicking on your button in a new design. That is your “what” that is happening quantitatively.
• You do a qualitative usability test to figure out “why” behind this behavior and discover that people had a hard time seeing the button. This is your “why” that causes the “what” of the button not being clicked.
• But now, let’s say we go a level deeper. Your “what” is that people have a hard time seeing the button and you want to figure out “why” that is the case. Our “why” of users not seeing a button is now our “what”.

The initial “why” can be the “what” in our next line of investigation. This spiral transition of “why” and “what” could go up or down forever. 

In the classic description of our methods through “what” and “why”, we’d reach for a qualitative method to investigate “why” users are not seeing the button. This means interviews or usability testing most often. However, either option in this case would be wrong.

Why can sometimes only be answered with quantitative research

Staying with our hypothetical, we know that users don’t press the new button. The button position in the page hasn’t changed but the design of the button has. Both its size and contrast have changed. A user cannot reflect on their cognitive processes to know which issue is causing them to miss seeing the button². Sure, if you ask someone, they may say “yeah, it’s quite small” or “I have a hard time seeing light green”, but is that accurate? It’s unlikely. Perhaps contrast is the true issue, but more users have a hard time with small buttons on other websites, so they’re biased to generate that as the reason for missing the button in our context. This is why a qualitative method would be the wrong approach to answer this particular “why”.

In this case, we’d need to use quantitative A/B testing or usability experimentation where we vary the conditions of size and contrast and evaluate the behavioral performance  to see which element truly affects the button’s salience. The “why” in this case cannot be answered with a qualitative evaluation that relies on a user’s introspection. So saying that qualitative research answers “why” does not effectively define when to appropriately use qualitative research. I say “appropriately” because this is one of the most common things I see in undertrained UX researchers, attempting to use a method that will yield data but is not truly suited to finding rigorous results.

A philosophical detour

What is the nature of this research question around button size/contrast that makes classic qualitative approaches insufficient to address it? We need to dig into the philosophy of science. You don’t absolutely need to know these philosophies to be a UX researcher. However, if you don’t understand these philosophical concepts, you essentially allow others to define your own philosophy in research and make choices for you. My goal in writing is to fully equip people interested in UX research to understand how deep the rabbit hole goes and not to overly simplify things for the sake of moving faster. But that said, let me do my best to break it down simply.

David Travis provides a nice UXR-focused breakdown of our key, conflicting roots: positivism and interpretivism. In a nutshell, positivism holds that there is an objective truth and we can measure it without bias through strict empirical research practices. Interpretivism holds that truth is individually and socially constructed, so there is no possibility of a singular, unbiased measure. 

One of the philosophies that emerged from the conflict between these traditions is critical realism. Critical realism holds that there is a singular, objective truth, but we cannot access it without bias. We can access parts of objective reality at varying levels of fidelity, acknowledging the ways that human construction will always warp some measure of reality. Critical realism is highly aligned with the philosophy of action research that I used to define rigor in UX research as a mixed-methods discipline. 

We need to distinguish between the reasons actors give for their behavior (subjective rationalization) and the causes that generate that behavior (generative mechanisms). While qualitative research excels at uncovering the former, it is structurally ill-equipped to isolate the latter, particularly when the cause is a mechanical variable (like low-level visual perception) rather than a conscious choice or socially-influenced domain³. 

This was a long way to say, some “why” questions cannot be answered through qualitative, interpretivist methods. 

Moving beyond “What” and “Why” as definitional

Up until now, all I have done is provide concrete and theoretical examples as to why the current dominant thoughts on qual vs. quant UX research are insufficient. Next, I’d like to focus on defining qual vs. quant methods in UX research.

Again, we could go quite deep in the explanation. If you have a lot more time than this post, I highly recommend Mixed Methods: A Short Guide to Applied Mixed Methods Research from the brilliant Sam Ladner. She gives a much more thorough accounting of history and differences in mixed-methods research. What I am attempting to add in the following sections is a (somewhat) novel and (somewhat) concise framework that will set most UX researchers on the right path to selecting the right methods for the right problem. 

So, what are the definitions?

These definitions (1) get to the point of the reason you’d want to employ each as a UX researcher without relying on just “what vs. why” and (2) skip beyond the deep philosophical discussion of “fundamentally different assumptions about what is even real” which differentiate them, as Ladner put it.

The value of better definitions

There is a tactical value in taking my definitions on and getting outside of the box of “interviews are qual” and “surveys are quant”. It’s not just philosophy for its own sake.  We can better understand the weaknesses or strengths of methods, as well as consider methods we may not have called “research” before.

Qualitative research without participants

The process of identifying elements of interest can take on so many forms beyond interviews or usability tests. For example, The Big 5 personality inventory is a gold standard for measuring human attitudes in its comprehensiveness and reliability. However, the way this inventory was created was through an old technique called “Lexical Hypothesis”. Essentially, the researchers went through an extremely manual process of extracting words from the entire English language, categorized them (like a card sort of thousands of words), and then reduced them into final latent constructs. This is how the elements of interest (personality traits) were identified. 

No interviews or behavioral observations were used in early Big 5 research, but this work is certainly qualitative and focused on identifying the elements of interest. UX research generally sees qualitative research as interviewing or observing users, but that is just one flavor of an analytical process to identify what we care about measuring or defining further.

Of course the quantitative work of scale development, refinement, and measurement in the Big 5 looks much more like a classic, quantitative psychometric survey process. This was done once the elements of interest were established, and many “why” questions have been answered through quantitative investigations using the Big 5.

AI interviews aren’t that novel

Jumping ahead 60 years from the Big 5’s roots, the lines between qual and quant seem to be getting blurrier as LLM technology rapidly evolves. Anthropic made a splash with the  80,000 unmoderated AI-interview project earlier in 2026. This ignited a discussion about this “new” method, but in my view, it’s really just a poor mash-up of existing methods. The results show how the method integrated both the identification of elements of interest and their prevalence in the same moment. 

AI-moderated interviews are “quanting the qual”, but done at a scale no human could hope to approach. We’re not in a better position for that, and our new methodology definitions help explain why this is a problem. It’s not just “why” at scale, but rather attempting to count things that haven’t yet been identified as worth counting.

“Objectivity” in data science

UX research often struggles in a battle of prestige with data science (DS). DS is seen as objective, using log data at a great scale to understand population behaviors and characteristics. Our qualitative/quantitative definitions uncover a subtle aspect of DS work that is often glossed over. Any DS output is still subjective because “the existence of data demands that someone has discovered, gathered and to some extent processed and presented those data.”  In other words, some humans are identifying the elements of interest before they measure anything. 

This sneaky aspect of data work is not intentionally sneaky, but more likely because many are not trained or incentivized in DS to consider what this bias is, even though it’s inescapable. In UX research, the training to reduce this kind of bias is a significant element of robust qualitative research education, whether in human factors or anthropology. By understanding the real nature of qualitative research, it’s easy to see where DS needs to spend more time identifying elements of interest. The purpose of saying all this is not simply to be more right than our friends in DS. I’ve done collaborative work with DS in the past to identify what really matters ahead of creating the log measures. In my experience, these human-grounded metrics matter more to product teams and show more value in product strategy because they’re tied to the reality of our users.

Wrap up

We need to throw out the idea that qual is “why” and quant is “what” or “how much”.  It’s not necessarily that it’s too simple, but the categorization is plainly wrong in certain situations. The idea of counting for quantitative work is so ingrained from our positivist-biased society that we don’t need to retread it. One pithy way to integrate qualitative work in that phrasing is then “to think of qualitative methods as procedures for counting to one”.

When it comes to practicing as a researcher in any insights function, whether UX research or data science, it’s best done through an iterative process. We must define our elements of interest (qual) and then discover their prevalence or magnitude (quant) to guide the organization in the right direction.

Appendix

1. I combed through what websites I could find and here is the table of sources. ↩︎
2. Dig deeper here on Telling More Than We Can Know, a seminal paper about what we can and cannot cognitively reflect on. Simply put, we can’t know all of our processes and causes consciously. Metaphorically, an eye cannot see itself and a camera cannot record an image of its own lens as the subject.  ↩︎
3. Critical naturalism is another term that sums this up if you want to dig deeper on the topic. Essentially, human social structures defy the same types of investigation as the physical world. You could think of it as a spectrum from physical to social: human relationship -> human perception/cognition -> neurology -> biology -> chemistry -> physics. Each of these domains are increasingly better suited to study through positivism. ↩︎