Marcus Andrey Study
Ler em português Library

UX Research · Step 1 · 6 min read

What research is for

Research is how a team stops guessing about the decisions that matter. What it's for, the three jobs it does at different moments of a product, the difference between what people say and what they do, and what research is not.

A path comes in and forks three ways at a signpost, each branch ending at a low marker post. Point down a branch and its post rises and takes the bright mark, which until then sat on top of the signpost, where you stand and guess. That's this step in one gesture: research turns a guess into a decision. This step has four lessons: what research is for, research questions that lead to decisions, when not to do research, and research ethics and consent.Illustration: Marcus Andrey, drawn with Hairline (MIT, Lucas Marques).

Research is not about users. It's about a decision you're about to make, and how sure you need to be before you make it. Every product team decides things every week: what to build next, which flow to fix, what to call a button. Most of those decisions rest on beliefs nobody has checked. Research is the habit of checking the beliefs that would hurt the most if they were wrong, with the people who know: the ones who use the product.

What you'll be able to do after reading:

  • Explain what research is for, in one sentence, to a sceptical product manager
  • Tell generative, formative and summative research apart, and when each fits
  • Explain why what people say and what people do are different kinds of data
  • Spot research that's really theatre: studies that can't change anything

1. Research is for decisions

Erika Hall puts it simply: good research "is about asking more and better questions, and thinking critically about the answers" (A Book Apart, Just Enough Research). Questions and critical thinking, not a lab or a big budget.

The useful way to think about it: every study should be attached to a decision. "Should we fix onboarding or build recurring invoices first?" is a decision. "Let's learn about our users" is not, and studies that start like that tend to end in a nice deck that nobody uses.

That doesn't mean research only matters for big launches. It means you should be able to finish this sentence before you start: "We'll do X differently depending on what we learn." If you can't, you're not ready to study yet. The next lesson, Research questions that lead to decisions, is about getting there.


2. Three jobs, three moments

Research does different jobs at different points in a product's life. NN/g describes them by stage (Rohrer, When to Use Which User-Experience Research Methods, NN/g, 2022):

  1. GenerativeWhat should we make, and for whom?

    Find needs, problems and opportunities before there is a design. Field studies, diary studies, interviews.

    Feeds: the direction
  2. FormativeHow do we make this design better?

    Find what gets in people's way while the design can still change. Usability tests, card sorts, tree tests.

    Feeds: the next iteration
  3. SummativeHow well does it work now?

    Measure the product against an earlier version or a competitor. Benchmarks, A/B tests, analytics, surveys.

    Feeds: the score, and the next priority
Three jobs for research, one per moment of a product. Each answers a different question and feeds a different decision.Diagram: Marcus Andrey, CC BY 4.0. Stages and methods after Christian Rohrer, When to Use Which User-Experience Research Methods (NN/g, 2022).
  • Generative research comes first, when the question is strategic: which way should the product go? In NN/g's words, these methods "help generate ideas and answers about which way to go".
  • Formative research happens while you design. It informs "how you can improve the design".
  • Summative research comes when there's a product to measure, "against earlier versions of itself or competitors".

A team that only does summative research learns, very precisely, that something is not working, too late to change it cheaply. A team that only does generative research understands people beautifully and never checks whether its design works for them. You need all three, at the right time.


3. What people say, and what people do

The second idea that organises the whole field is the difference between attitudinal and behavioural data. NN/g sums it up as "what people say" versus "what people do", and adds that "very often the two are quite different" (Rohrer, NN/g, 2022).

  • What people say (interviews, surveys) tells you about beliefs, motives, language and needs. It's limited, says NN/g, "by what people are aware of and willing to report".
  • What people do (usability tests, field visits, analytics) tells you what actually happens. It can't always tell you why.

There is a third distinction: qualitative methods are better at "why or how to fix a problem", quantitative ones at "how many and how much" (same source). One study rarely covers all of it. Choosing a research method turns these axes into a map.


4. What research is not

  • Not validation. "Let's validate the design" usually means "let's confirm it". A study that can only say yes isn't research. Go in ready to be wrong, and write down beforehand what would prove you wrong.
  • Not asking people what to build. People are experts in their own lives and problems, not in your solutions. Their needs are the data; the design is your job (see User needs and jobs to be done).
  • Not a phase that ends. NN/g's advice on when to research is to do it "at whatever stage you're in right now", and at every stage, because "the earlier the research, the more impact the findings will have" (Farrell, UX Research Cheat Sheet, NN/g, 2017).
  • Not only for researchers. Designers, product managers and developers can all do good research with some care. Your company may have specialists; most teams don't. This course is for the people who do it anyway.

5. Going deeper than a first study

If you've done Short user interviews and Usability testing with five people in UX Design Foundations, you've done research. This course goes further: how to frame the question so the study decides something, how to choose and combine methods, how to recruit and treat people properly, how to analyse without fooling yourself, and how to make the findings travel. The rest of Step 1 covers the framing: good questions, when not to study, and ethics.


6. Common mistakes

  • Starting with a method ("let's do a survey") instead of a decision.
  • Treating research as proof that the team was right.
  • Asking people what they'd want, then building the list.
  • Doing all the research before the design and none after.
  • Trusting only what people say, or only what the numbers show.

7. Vocabulary

Term Meaning
Generative research Research that finds needs and opportunities before there's a design
Formative research Research that improves a design while it can still change
Summative research Research that measures how well a finished design performs
Attitudinal Data about what people say: beliefs, opinions, reported behaviour
Behavioural Data about what people do, observed or logged
Qualitative Data gathered directly, good at answering why
Quantitative Data gathered through an instrument, good at answering how many

8. Try it (no tools needed)

  1. Write down three decisions your team made in the last month: anything, from a feature to a label.
  2. For each one, ask: what belief was it based on? Who checked it, and how?
  3. Pick the one that would hurt most if the belief were wrong. Which job would research do there: generative, formative or summative?

Reflection questions: Which of your team's beliefs about users has never been checked? Who would be upset if it turned out to be wrong, and is that a reason to check it or a reason people avoid it?


9. If you already practise

  • Look at the last three studies you ran. Can you name the decision each one changed? If not, the problem probably started before the first session.
  • Balance your portfolio: count generative, formative and summative studies this year. Most teams lean hard to one side.
  • When you present findings, open with the decision, not with the method.

Checklist

Learn more

Books, courses and videos to go deeper

Book links go to Amazon. As an Amazon Associate I earn from qualifying purchases, at no extra cost to you.

Research questions that lead to decisions · When not to do research · Choosing a research method · Short user interviews · Usability testing with five people · AI in UX research

Saved only on this device.

↑ ↓ to move · Enter to open · Esc to close