Which Qualitative Research Method to Use at Each Product Stage

- Reading
- 8 min
- Tables
- 1
A practical guide based on material from the Nielsen Norman Group for deciding between usability testing, interviews, field studies, and diary studies, with what changes when applying this with a real team in Brazil.
Every discussion about qualitative UX research starts the wrong way: from the technique, not the question. An article published on November 20, 2026 by the Nielsen Norman Group, written by Maria Rosala, organizes the four most commonly used qualitative methods in product (qualitative usability testing, user interviews, field studies with contextual inquiry, and diary studies) around a simple criterion: what you need to discover defines the method, not the other way around.
The distinction that NN/G reinforces is worth repeating because product teams frequently get it wrong. Qualitative research answers "why" and "how": it delivers observations and themes from a small number of participants. Quantitative research measures "how many," "how much," and "how often," with larger samples and statistical analysis. The same method, such as usability testing, can serve both purposes: what changes is the study design and sample size, not the technique itself.
The Question Comes Before the Method
If the question is evaluative (do people understand the shipping cost before checkout? why are they abandoning the signup form?), the path is qualitative usability testing. If the question is generative (what are the biggest frustrations when planning a trip? how do nurses use the medical record during a shift?), the path goes through interviews, field studies, or diary studies. Among these three, the choice depends on three questions the article itself suggests asking before designing the study:
- Do you need to see what people do, or is hearing enough? Interviews are quick, but depend on the respondent's memory.
- Does the environment matter? If tools, interruptions, or other people shape the experience, it's worth observing on site.
- Does the experience unfold over days or weeks? If so, no single session captures it as well as a diary study.
Qualitative Usability Testing: Finding the Problem and the Why
In qualitative usability testing, a facilitator asks the participant to perform realistic tasks while thinking aloud, narrating expectations and confusion as they happen. The focus is on observation, not metrics: where the person got stuck, what they expected to happen, why they chose the wrong link. This differs from quantitative testing, which measures success rate and task time with larger samples, around 40 participants.
The small sample isn't poorly done economizing: five participants per user group, per round, already reveal most of the common usability problems in a design. Testing with more people in the same round has diminishing returns; NN/G recommends spending the budget testing, fixing, and testing again, rather than inflating a single round. The method works from paper prototypes to production products, in person or remote, with or without a live facilitator.
The biggest risk is influencing the participant without realizing it. A poorly worded task can lead the person to search for a specific word or feature, and an inexperienced facilitator can change the observed behavior just by interrupting at the wrong moment. This makes facilitation a skill that is trained, not an improvisation by someone who "is good at conversation."
Interviews: Understanding Before Designing Anything
The user interview doesn't show a design: the participant talks about experiences, needs, and motivations, and the interviewer follows a script of open-ended questions. It's the right method for the discovery phase, when there's no solution hypothesis yet and the goal is to feed personas and journeys with real data instead of assumptions.
One practical recommendation from the article is worth citing because it changes the quality of the answer: asking about a specific event ("tell me about the last time you booked a vacation trip") produces richer, more precise details than asking about general habits ("how do you usually book your vacation trips?"). The second question invites a generic, rationalized answer; the first forces the person to reconstruct a concrete situation.
The limit of interviews is structural: they collect reported behavior, not observed behavior. People forget details, omit what they don't perceive as relevant, and sometimes describe themselves more favorably than they actually act. When the research question requires knowing what people actually do, not what they say they do, interviews alone won't solve it: you need to combine them with an observational method.
Field Studies and Contextual Inquiry: What No One Mentions in an Interview
Field study happens in the user's natural environment (home, office, store, workstation), not in a meeting room. That's where the sticky note taped to the monitor appears, the parallel spreadsheet the person keeps open next to your product, the colleague they turn to when stuck, and the constant interruption that an interview never captures. The article lists variations, from direct observation (without asking anything) to ethnography (prolonged immersion in context), including the customer-guided tour.
Contextual inquiry is the most useful format for product because it combines observation with targeted questions, following the logic of a master craftsperson teaching an apprentice: the participant works normally, the researcher observes and only interrupts when something isn't clear. The session follows four stages: warm-up and explanation of the process, an explicit transition from "conversation" to "observation," alternating between observing and discussing what was seen, and a wrap-up in which the researcher summarizes their interpretations for the participant to confirm or correct.
The example NN/G brings from a redesign project illustrates why the logistical effort of going into the field is worth it. Insurance specialists described, in an interview, a simple process of copying and pasting data between systems. Observing the actual work, steps emerged that no one had mentioned: checking a second form for missing information and repeatedly clicking Save even knowing the system saved automatically. Both discoveries only existed because someone observed, not because someone asked.
Contextual inquiry pays off for complex systems, long processes, and expert users; it's wasted effort for a simple newsletter signup screen. The most common trap is the participant slipping back into interview mode, narrating the work instead of doing it: a gentle reminder to "work as you normally would" resolves most cases.
Diary Study: Capturing What Changes Over Time
In a diary study, the participant records activities and experiences as they happen, over days, weeks, or even a month. It's the only one of the four methods designed for habit, long journeys, a system's learning curve, or changes in channel and device over time, situations where a single session simply has no way of surfacing.
Entries can be triggered by event (whenever something relevant happens), by interval (every night, for example), or by external signal (a text message reminder). Planning a diary study takes more work than the other methods: it requires choosing a logging tool, writing clear instructions and examples, defining a period long enough to capture the behavior, and following entries as they arrive to ask more while the memory is still fresh. It's common to close the study with a wrap-up interview, going deeper entry by entry.
Since the commitment lasts days or weeks, dropout is expected: the article recommends recruiting beyond what's needed and offering a larger incentive than for a single session. And, since all the data is self-reported, the diary study inherits part of the interview's bias, even while reducing the memory problem by recording close to when the event occurred.
Comparing the Four Methods
| Method | Answers | Cost | Type of Data | When to Avoid |
|---|---|---|---|---|
| Qualitative usability testing | Where and why a design fails | Low to medium | Behavioral | When the question is generative, not evaluative |
| Interview | Needs, motivations, experiences | Low | Attitudinal (what the person says) | When precision about actual behavior is essential |
| Field study / contextual inquiry | Real workflows, tools, and workarounds | Low to high | Behavioral and attitudinal | In simple interfaces, without process complexity |
| Diary study | Behavior change over time | Medium | Self-reported attitudinal | When the project timeline doesn't allow for weeks of collection |
Combining Methods Instead of Choosing Just One
The article suggests combinations that work well in practice: opening a usability test with a short interview about the participant's background (without revealing what's being tested); interviewing before or after a field observation; closing a diary study with a wrap-up interview; or following a qualitative study with a quantitative one to measure how common what was found actually is. None of these methods compete with each other: they answer different questions along the same discovery journey.
Regarding sample size, qualitative logic differs from quantitative logic. Instead of calculating statistical significance, researchers using generative methods (interview, field, diary) look for saturation: the point at which new sessions stop revealing new information. In practice, this means starting small, five or six participants, analyzing as the data comes in, and only adding people while the learning continues to be worth the effort.
What This Changes for Those Running Product in Brazil
Product teams in Brazil tend to jump straight to usability testing because it's the most familiar method and the cheapest to justify to leadership. The problem is using usability testing to answer a generative question: asking "why do people abandon the delivery app after their third order?" with a task test doesn't work, because the question calls for an interview or a diary study, not observation of a single task.
For those deciding on a tight budget, the article's own math helps with prioritization: a simple usability study fits into three days of work (planning, testing, analyzing), and a remote session cuts travel and room costs. Field studies cost more when they require traveling to the user; if an in-person visit doesn't fit the budget, a remote diary study captures part of the same context without the travel cost. In Brazil, where recruiting participants outside the Southeast hub tends to be the biggest logistical bottleneck, this is the practical difference between doing the right research and simply doing the research that fits the calendar.
It's also worth remembering that diary studies and contextual inquiry involve recording routines and, sometimes, sensitive third-party work data (medical records, a client's internal system). This requires a specific consent form and extra care with the LGPD (Brazil's data protection law) when storing and anonymizing entries, a step that the NN/G article, aimed at the American market, doesn't address, but that any research team in Brazil needs to formalize before recruiting the first participant.
Translated from the Brazilian Portuguese original · Read the original
Nielsen Norman Group study finds AI designs better with synthesized research than with raw data
A Nielsen Norman Group study compared AI-generated designs based on raw user comments with designs generated from the same research after it was synthesized into findings and recommendations. Synthesized research won in both tests, with an advantage of up to 11.4 percentage points.





