How to brief a web designer so they actually build the right thing the first time.

The hook

A 2024 ResearchGate survey of 60+ PMs found that 92% of failed projects were attributed to scope creep, and the root cause, almost without exception, was a vague brief at the start. Every change order, every "can we just add" request, every 4-week project that became 12, traces back to the kickoff document.

A good brief takes a founder 60–90 minutes to write. It saves both sides weeks of confusion. Below is the exact template Plinth uses with every new client, with examples of what good and bad answers look like, plus the questions designers wish you'd answer that you usually don't.

The brief structure

The Plinth brief has nine sections. None are optional. The whole thing should fit on two pages. If your answer to any section runs longer than a paragraph, you probably need to think harder, not write more.

1. The business in one sentence

The question: What does your company do, for whom, in one sentence?

Good answer: "We make AI-powered legal contract review software for in-house counsel at companies with 50–500 employees."

Bad answer: "We're a next-generation legal tech platform leveraging AI to transform how legal teams work."

The bad answer tells the designer nothing. The good answer tells them the product (contract review), the user (in-house counsel), the company size (50–500), and the tech (AI). A designer with that one sentence can already make good calls about tone, visual references, and content priorities.

2. The primary visitor

The question: When this site is live, who is the single most important person visiting it? What are they doing right before they land? What do they need to learn in their first 10 seconds?

Good answer: "A General Counsel at a mid-market SaaS company who got our name from a peer in a Slack community. They're skimming our site on their phone between meetings. In 10 seconds they need to learn: we're real, we're enterprise-grade, and we're built for legal, not a generic AI tool."

Bad answer: "Our target customers are legal professionals at growth-stage companies."

The bad answer is demographic. The good answer is behavioral, it tells the designer what state the visitor is in, what context they're carrying, and what objections to handle. Design follows behavior.

3. What you want them to do

The question: What is the single action you want a visitor to take? If you list more than two, pick again.

Good answer: "Book a 30-minute demo with our sales team. Secondary: download a one-page security overview for IT review."

Bad answer: "Sign up for a trial, book a demo, download our whitepapers, follow us on LinkedIn, subscribe to our newsletter, and read the blog."

CTA discipline starts in the brief. If you can't pick one action in the brief, your homepage will end up with six CTAs competing with each other.

4. The current site's biggest problem

The question: In one paragraph, what's wrong with your current site? Why are you redoing it?

Good answer: "Our current site reads as a developer tool, but we just pivoted to selling to legal teams. The hero says 'AI infrastructure' and the screenshots show terminal interfaces. We're losing legal buyers in the first 5 seconds because they don't recognize themselves on the page."

Bad answer: "It looks dated and we want something more modern."

A specific problem statement points the designer at the right fix. "More modern" tells them nothing, modern in what sense, fixing what failure mode? The good answer makes the project obvious: reposition the site for the new ICP.

5. Three sites you love (and why)

The question: Three websites whose work you respect. For each, one sentence on the specific thing you love.

Good answer:

• "Linear (linear.app), the way they made the homepage feel like the product is actually running on it."
• "Vercel (vercel.com), total grayscale discipline, color reserved for product output only."
• "Stripe (stripe.com), italic-only emphasis system, looks expensive without trying."

Bad answer: "I love Apple's design."

Two problems with the bad answer. Apple has thousands of designers and a $3T market cap, the reference is unhelpful. Worse, "Apple's design" isn't specific to any quality. The good answer names specific decisions you respect, which gives the designer permission to make similar decisions.

6. Three sites you hate (and why)

The question: Three websites whose work you actively dislike. For each, one sentence on what's wrong.

This question matters more than the previous one. Knowing what to avoid is easier and more useful than knowing what to copy.

Good answer:

• "[Competitor], every page looks like a SaaS template, no point of view."
• "[Other Competitor], hero is a stock illustration of people at laptops."
• "[Generic enterprise site], copy is so abstract I can't tell what they actually sell."

Bad answer: Skipping this question. Most clients do. Then they reject design directions in week 2 and the team has to start over.

7. The constraint list

The question: What constraints do we have to honor? Brand colors, fonts, copy, content, integrations, timeline?

Good answer:

• "Brand: primary color is #0A47C5, logo is in the brand kit (link), no font preference yet."
• "Content: we have copy for the homepage and pricing page. Need help with blog + about."
• "Tech: must integrate with HubSpot CRM, Pylon for support, and Resend for transactional email."
• "Timeline: live by August 15 for our conference launch."

Bad answer: "We're flexible on everything, just make it good."

"Flexible on everything" is a lie. Every client has constraints. The good answer surfaces them in week 1. The bad answer surfaces them in week 4 when the design is half-done.

8. Success in 6 months

The question: If this project is a clear success when we look back in 6 months, what specifically will be true?

Good answer: "We'll have doubled demo requests from organic traffic. Sales will stop complaining that the site isn't qualifying prospects. We'll be running content marketing weekly without engineering tickets."

Bad answer: "The site will look great and people will love it."

A measurable success criterion shapes every design decision. If demo requests are the goal, the demo CTA matters more than the team page. If content velocity is the goal, the CMS choice matters more than the homepage animation.

9. Who decides

The question: Who is the single person with final design approval authority? Who else gets input, and by when?

Good answer: "Fred (CEO) has final approval. Sarah (Head of Marketing) gets advisory input on copy by end of week 1. No one else has design input."

Bad answer: "We'll discuss design as a team."

This is the question that determines whether the project ships in 5 weeks or 12. If it's the team, it's a committee. If it's a committee, every revision round is political and the design ends up averaged into mediocrity. One person decides. Everyone else gets advisory input by a fixed date, then sits down.

The decision-maker test

Look at your last big design project. Count the people who had veto authority over the final design. If the answer is more than one, that's why it took twice as long as planned. Single decision-makers ship better work faster, every time.

The questions designers wish you'd answer

The brief above covers the basics. A few more questions that separate good clients from great ones:

What's your reading-level vibe? "This should read like a Stripe homepage, short sentences, grade 6, no jargon" vs "This should read like a McKinsey white paper, denser, professional, citation-heavy." Tells the designer how to write.

What can we kill if we run out of time? Forces priority ordering. The careers page, the blog, the press page, what gets cut if week 4 is tight?

What's your appetite for animation? A simple scale: 1 = no motion, fully static; 5 = generous animation, things move on scroll, hero has motion graphics. Most clients say 4 then panic when they see it in design. Set the level upfront.

What's the writing voice? Three adjectives. "Confident, direct, slightly dry" tells a writer something. "Professional and approachable" tells them nothing.

What's the AI policy? Is it ok if we use AI to draft hero copy variations? To generate illustrations? To write the blog? Some clients are strict no-AI. Some are AI-first. Most haven't thought about it. Decide in the brief.

The 2-hour kickoff

Once the brief is written, a 90-minute kickoff call should cover:

1. Read through the brief together (30 min), designer asks clarifying questions
2. Look at the inspiration sites (15 min), what specifically do you want from each
3. Walk through your current site (15 min), what's broken, what to keep
4. Review the timeline + milestones (15 min), when designs happen, when builds happen, when you review
5. Confirm logistics (15 min), Slack channel, file sharing, payment schedule, change order policy

After that call, the designer should be able to start work without coming back with major questions for at least a week. If they can't, the brief wasn't tight enough.

Common brief mistakes

A few things I see go wrong over and over:

• Writing the brief as a sales pitch for your company. The designer needs to understand the business, not be convinced by it. Strip the marketing language out.
• Treating "the team will weigh in" as a decision-making structure. It isn't.
• Refusing to name competitors. Designers need to know your competitive frame. Pretending you don't have competitors helps nobody.
• Skipping the "sites you hate" question. This is the most useful question in the brief and the most-skipped.
• Sending design references that look nothing alike. "I love Apple, Nike, and Notion." These have nothing in common. Pick references with shared visual DNA.
• Adding requirements after the brief is signed. Use change orders for new scope. Don't let the brief drift mid-project.

Free template

A copy of this brief structure as a Google Doc template is available, request it here and I'll send it after our intro call. Or you can rebuild it from the structure above in 10 minutes. The template is less important than the discipline of answering the nine questions honestly.