Summarize with AI
A website design brief for a startup should open with the sentence a stranger needs in the first four seconds: what the product is, in the category word your buyer already uses, and who it is for. Next come the one action the page asks for, the question every sales call starts with, and the proof you can actually show. Reference sites go last, five at most, each with a reason attached.
The order matters because you know your product and nobody else does. A designer reads a brief from the top and starts building from the top. Lead with a mood and a stack of links and you get a look with nothing to say. Lead with the sentence and the look has a job to do.
Two briefs for the same company, written the same week
Picture a seven person startup that screens public tenders for civil engineering firms. This is an example, not a client. The product reads a tender pack, the long PDF and the drawings a council posts, and tells a small firm whether it qualifies and what a bid would need.
Brief one is a Notion page called New site, draft 2. The first line reads: we want it to feel premium, clean and modern, a bit like Linear. Under that sit twelve links. Linear, Vercel and Raycast come first, then nine more dark software sites. Then a page list: Home, Product, Use cases, Pricing, About, Blog, Careers, Contact. Then a deadline: live before the seed extension closes. The word tender appears nowhere on the page.
Brief two fits on one screen. Line one says Tender screening for civil engineering firms. Line two says the button reads Screen a tender, with no second button of the same weight. Line three gives the question every sales call opens with, can it read scanned drawings or only the text, and the two sentence answer. Then two pilot firms, one of which has agreed to be named. Then three references, each with a line on what to take and what to leave.
Both briefs describe the same company. Only the second could not belong to any other company.
Why most brief templates put the wrong thing first
Search for a website design brief template and you get a form. Company background, goals, target audience, competitors, design preferences, budget, timeline. Every field gets the same size box, and that is the flaw. The form treats the line a stranger needs as one field among eight, usually filled in last and in a hurry.
Company background is your story: when you started, why, the pivot. A visitor needs none of it in the first four seconds. Design preferences is where premium and modern end up, words about how you want to feel about the site rather than what it tells anyone. Most advice says a good brief is a complete brief. This post argues that a good brief is an ordered one. The top three lines decide what gets designed first, and what gets designed first is what survives the revision rounds.
Line one: the category word and who it is for
Write this line before anything else. Category word first, the one your buyer would type into a search box, then who it is for. Tender screening, not bid intelligence. Civil engineering firms, not AEC teams. If you would not say it to a site manager over the phone, it does not go in line one.
You cannot test this line on yourself. You know what the product does, so every vague word gets filled in by what you already know. Read it to someone who has never seen the product and ask what it does. If they name a category you did not mean, fix the line before the brief goes to anyone.
Then list your proof, honestly. Write what you have and whether you may show it: a pilot customer who agreed to be named, one who did not, a number you can stand behind, a screenshot of real output. Mark the gaps as gaps. A studio can design around missing proof. It cannot design around proof that turns out to be wishful a week before launch. If your deck already carries proof the site does not, close the gap between the deck and the site first, because investors will read both.
The Mi-VAD website shows the line and the proof working as one. Mi-VAD is building a heart pump about the size of a coin, and its whole pitch rests on being small. The site had to speak to investors and clinicians at the same time. On the site we designed and built in Webflow, the homepage makes the case for small before it says a word about specifications. The proof is the size, so the size comes first.
The question every sales call starts with
Open your notes from the last five sales calls and write down the first real question each prospect asked about the product. Skip the small talk. For most early companies it is the same question every time, and the founder answers it in two sentences without thinking.
That question goes in the brief, with your answer underneath. It is the nearest thing you have to a list of what visitors leave your site still wondering. In the tender example, nobody can tell from the homepage whether scanned drawings work. So the site manager books a call to ask, or more often closes the tab and keeps reading tenders by hand.
If the second and third questions repeat across calls, add them too. A designer who sees three real questions knows which sections the page needs and in what order. That is a better input than any sitemap you could write.
What a vague brief costs before and after launch
A vague brief does not save time. It moves the work somewhere else. The studio spends its first calls working out what the product is, and if you sent the brief to three studios, you explain it three times. The proposals come back with three mood boards, all dark, because twelve dark references average out to a dark site. Then the revision rounds turn into arguments about adjectives.
After launch the cost moves to your calendar. If the homepage never answers the opening question, demo slots fill with people who wanted something else, and each one takes a slot a real buyer could have had. That calendar is your runway. Measured on AI companies, the Series A bar is around 3.5 million in ARR, up from roughly one million three years earlier (Carta, Q1 2026). A site that sends you the wrong calls does not move you any closer to it.
The brief, in order: a template to fill in this week
Keep it to one page. Fill it in this order, and do not move a section up because it feels more exciting.
- The line. Category word and who it is for, in your buyer's words. One sentence.
- The one action. The exact button label and what happens after the click. If there is a second action, say that it should look smaller.
- The opening questions. The first one to three questions from real sales calls, each with your two sentence answer.
- The proof list. What you have, what you may name, and what you do not have yet.
- The moment. What the site has to be ready for: a raise, a launch, the first enterprise pilot. A date tied to an event tells a studio more than a date on its own.
- The decider. One name who signs off. Then who gets to comment, and at which stage.
- The constraints. Which platform, who edits pages after launch, and which current pages bring in search traffic and must keep their addresses.
- References, last. Five at most, each with one line on what to take and one on what to leave.
Most founders can write the first four in an afternoon. If line one takes longer than that, it is telling you something about the positioning, not about the brief.
References go last, each with a take and a leave
References help. They make a bad first page. A designer who opens twelve links before reading a word about your product starts designing for those companies, because they are the only thing on the page with any detail.
Write each one as two short lines. Linear: take the way a real product screen sits right under the headline. Leave the dark theme, because site managers open the link on a phone, outdoors. Vercel: take how one button carries the whole hero. Leave the developer vocabulary, because your buyer does not deploy anything. One reference from your buyer's own world is worth adding too, such as a trade body site they already trust, for how it lays out documents.
If Linear, Vercel and Raycast are all on your list, it helps to know what that look is made of before you ask for it. Most of what founders admire in those sites is a set of decisions, and the take line is where you name the decision.
What to leave out of a website brief
- Premium, modern, clean, sleek. Swap each word for the decision behind it. Premium usually means the site looks smaller than the company now is, so say who you sell to today.
- A sitemap. Give the questions and let the studio propose the pages. Then push back on the proposal.
- A competitor's homepage to copy. It answers their buyer's question, not yours.
- The full feature list. Name the one output a buyer pays for. The rest belongs on a product page later.
- Proof you hope to have by launch. List it under gaps. A studio can leave a slot for it.
A brief in this order is what any studio needs on day one, ours included. Studio Maydit designs websites for AI founders in the US, UK and Europe, and our website design work is built in Framer, Webflow or custom code, as a fixed scope of three to four weeks or a monthly retainer with no long lock-in. If you have line one and the opening question but not the rest, bring what you have and book a 30 minute call with us.
