Where to find realistic design briefs to practice with

Type "design brief generator" into Google and you will find tools that hand you a sentence: design a mobile app for a local bakery. That is a prompt, not a brief. It contains no business model, no constraint, no objective and no customer, which means every decision you make against it is arbitrary — and a decision that cannot be wrong cannot teach you anything.

The gap between a prompt and a brief is the difference between practice that compounds and practice that fills an afternoon.

What a real brief contains

A brief is worth working from when it gives you enough to be wrong. Six components do that.

Brand context. Who this company actually is: how old, how big, how funded, run by whom. A four-year-old label run by two ex-musicians funded by direct sales is a different design problem from a venture-backed marketplace, and the difference shows up in every decision from tone to how much friction you can afford.

The value proposition. What this business offers that its competitors do not, stated concretely enough to design around. "Quality products" is not one. "Early-fan tier unlocks private listening links 72 hours before public release" is.

The problem, with its context. Not "the page needs a refresh" but what is actually happening: where people arrive from, where they leave, what the team has already tried. A problem without context invites you to solve the one you find most interesting.

An objective, with sub-objectives. One measurable thing this work should move, broken into the specific changes that would move it. Without an objective there is no criterion for success, so the review collapses into taste.

The customer. Who this is for, specifically — what they already believe, what they are comparing against, what they are afraid of. Demographics are close to useless here; beliefs and alternatives are what you design against.

A constraint. The deadline, the thing you cannot change, the platform you are stuck with, the strong opinion held by someone senior. Constraints are what make the exercise resemble the job. A brief with no constraint produces a portfolio piece; a brief with one produces a designer.

A good brief also names the mechanisms a brief should let you choose between, rather than leaving them to taste. If you want to see all six assembled, here is a complete brief in the format used for timed sessions, and here is what one designer did with it in an hour.

Where to get them

Real briefs from real work

The best source, and the least convenient. If you freelance, keep the briefs you have been given and re-run them later with a different angle. If you work in-house, the ticket you were handed usually contains a hidden brief — reconstruct the business context nobody wrote down and you have a good exercise plus a better understanding of your own job.

Rebuild a brief from a company that exists

Pick a real small business you find interesting — a specialty roaster, an independent label, a boutique gym — and reconstruct its brief from what is publicly available. Read the about page, the pricing, the reviews, the social accounts. What are they actually selling? What is their model? Who are they competing with?

This takes about twenty minutes and produces something far better than any generator, because the constraints are real. The reviews in particular are a gift: complaints are unsolved design problems written by the people affected.

Structured challenge platforms

Several sources publish briefs designed for practice. The Daily UI series is the best known and useful for building interface fluency, though the prompts are component-level and deliberately contain no business context. Sharpen and similar generators produce more varied prompts but with the same limitation. Real product teardowns from Growth.Design are not briefs but reverse-engineer the reasoning behind shipped decisions, which is the closest free equivalent to reading someone else's brief.

Quest generates a full brief daily with all six components and a starting design to iterate on, scored against 27 behavioural skills — which is the approach this article is arguing for, so weigh that accordingly.

Client work that never happened

Look at pitches you lost or projects that were cancelled. You already have the context and you now have no obligation to please anyone, which makes them unusually good practice material.

Write your own in ten minutes

Pick a real business, then fill this in. Answer every line — the ones you find hard are the ones doing the work.

• Company. Name, age, size, how it makes money.
• What they sell that others do not. One concrete sentence. No adjectives.
• The surface. The specific page, flow or screen in scope. One, not the whole product.
• What is going wrong. Observable behaviour, not an opinion about the design.
• Objective. One metric or behaviour this should move, and roughly by how much.
• Customer. What they believe before arriving, and what they are comparing you against.
• Constraint. Something you are not allowed to change, plus a time limit.
• Skills targeted. Two or three behavioural mechanisms you intend to rely on, chosen before you start.

That last line is the one people skip and the one that changes the exercise most. Deciding in advance that you are going to lean on social proof and cost simplicity gives you a criterion for every decision that follows, and something specific to evaluate at the end.

The test for whether a brief is any good

Read it and ask: could two competent designers read this and disagree about what to do?

If the answer is no, the brief is too thin — it either tells you the answer or gives you nothing to reason from, and neither produces learning. If the answer is yes, and you could argue for your choice using something other than personal preference, you have a brief worth an hour of your life.

Then set a timer. Sixty minutes is enough, and the constraint is doing more work than the extra hours would. The brief is only half of it: the other half is what to actually build with it. And once the brief is written, the same brief works as an AI prompt with nothing changed, which is the fastest way to find out which of its fields you left vague.