What It Takes to Be a Good Designer (Before You Open Figma)

Anyone can now type a prompt and get a decent-looking layout back in seconds. So what, exactly, is a designer for?

It's a fair question, and a slightly uncomfortable one. The honest answer is that the part of the job a prompt can do was never the whole job. The rest is harder to see in a portfolio, and it's what employers keep saying they can't find.

Craft is more than polish

Figma's State of the Designer 2026 report, based on a survey of 906 digital designers across several regions, asked designers what "craft" means to them. 58% defined it as visual polish. 47% connected it to thoughtful problem solving.

Both answers are right, but the second group is describing the part of the work that gets more valuable as generating visuals gets cheaper. Polish can be produced faster than ever. Deciding what should exist, for whom, and why still takes a person who understands the problem.

Don Norman put the distinction bluntly in a 2023 interview with McKinsey:

"In the past, designers trained to be crafts people. They would make beautiful, wonderful things, and it was often thought of as a part of art, but design is not art. Design is building things to help society, to help other people."

You don't have to agree that design is never art to take the point. If your work is meant to help someone do something, the quality of the work depends on how well you understood that someone and that something.

What employers say is missing

The frustration is remarkably consistent across fields, not only design. Recent graduates arrive with technical and theoretical knowledge, but many struggle to articulate the skills they have, to solve problems they haven't seen before, or to show how their work creates value for the organization.

The data backs this up. In the QS Global Employer Survey 2024, the largest gaps between how much employers value a skill and how well graduates perform it were in problem solving, communication, and resilience and flexibility. The World Economic Forum's Future of Jobs Report 2025, which surveyed more than 1,000 large employers, found that analytical thinking is the top core skill, with seven out of ten companies considering it essential, followed by resilience, flexibility and agility, and then leadership and social influence.

None of those are software skills. All of them show up in how you approach a brief before you draw anything.

Good design starts with knowing what you know

To be a good designer you have to do good design. To do good design you have to understand what you're doing and why. And before you open any software, two questions need answers:

What information do we already have?
What information do we still need to know?

Luckily, there's an entire field devoted to questions like these. Epistemology is the study of knowledge, from the Greek epistēmē (knowledge) and logos (study, reasoning). You don't need to read philosophy to use it. You need the habit of asking how you know what you think you know.

Two older words help here too. Quantitative comes from the Latin quantus, "how much". Qualitative comes from qualis, "of what kind". They describe two different kinds of information, and designers need both.

Questions: Quantitative: How many? How often? To what extent?; Qualitative: Why? How does it feel? What does it mean?
Results: Quantitative: Numbers, statistics, patterns; Qualitative: Words, stories, context
Tools: Quantitative: Surveys, metrics, A/B tests; Qualitative: Interviews, observation, ethnography
Good for: Quantitative: Measuring and comparing; Qualitative: Understanding and exploring

Asking well: the survey as a design object

The survey is the most common way design students gather quantitative data, and it's a good test of this whole idea, because a badly designed survey can send a project in the wrong direction before any design starts. Designing one well is a design problem in itself.

Know who's involved

A survey has several roles, even if one person plays most of them: the client, who sets the goals and wants something to change; the questionnaire designer, who turns those goals into valid, unbiased questions; the people who administer it; the respondents, who provide the data; and the analyst, who interprets results and looks for connections.

Know what you're measuring

Surveys can measure behavior patterns, attitudes and opinions, satisfaction, and reactions to a concept or prototype. In brand work, two things get confused all the time. Brand awareness measures whether people know or remember a brand. Brand image measures how they perceive it. Different questions, different instruments.

Order is part of the design

The order of questions isn't neutral; it affects the answers. A reliable structure:

general before specific
behavior before attitudes
non-sensitive before sensitive

In plain terms: first ask what people do, then what they think or feel about it, and leave anything personal for the end. Group questions by theme so respondents don't jump between topics, and sketch the flow before you build the form.

Avoid questions that break your data

Leading: Broken example: What problems did you have with the launch?; Why it fails: Assumes there were problems; Better version: How would you describe your experience with the launch?
Vague: Broken example: How do we compare with the competition?; Why it fails: In what way? For whom?; Better version: In which area do you think [brand] does better than [competitor]?
Double-barreled: Broken example: Would you like it to be cheaper and better quality?; Why it fails: Two questions in one; Better version: Split it into two questions
Double negative: Broken example: Do you oppose not allowing pet adoption?; Why it fails: Nobody knows what to answer; Better version: Are you for or against allowing pet adoption?

Know what your scale lets you conclude

Every question implies a level of measurement, and that level decides what you can do with the data.

Nominal: categories with no order (city, product type). Options must cover every case and not overlap.
Ordinal: a ranking where the gaps between positions aren't equal (rank these four stores from best to worst).
Interval-style scales: the most familiar is the Likert scale, named after psychologist Rensis Likert. Use five or seven points, label every point, include a neutral option for opinions, and mix positive and negative statements so people don't agree with everything out of inertia. Strictly speaking, Likert responses are ordinal, so be careful about treating averages as precise.
Ratio: real quantities with a true zero (minutes, visits per month, units bought).

Sample, then pilot

You can't ask everyone in your population (the universe), so you design a sample that represents it. Free sample size calculators will tell you roughly how many responses you need.

Then test the survey with five to ten people before launching it. Check how long it really takes (aim for ten minutes or less), which questions cause confusion, which ones everybody skips, and whether the answers make sense together. No prototype survives its first user test untouched, and surveys are no different.

The designer is the one who asks well

The next time you open Figma, the most important work should already be done. Before the first frame, run through these:

What do I need to know?
How am I going to get that information?
Are my questions free of bias?
Does the scale I'm using let me draw valid conclusions?
Did I test the instrument before using it?

A designer who knows how to ask knows how to design. Pick the brief you're working on right now and write down, in one line each, what you know and what you still need to find out. The second list is your real starting point.