Start Here.
Before you open Claude Code or look at a single design, you need real input from your client. Skip this and you'll spend three weeks building something they don't want.
Why this chapter exists
Most people skip discovery and jump straight to "let me build something cool." Then halfway through, the client says "actually we hate orange" or "we need a Spanish version" or "where's the donate button?" — and you're rebuilding from scratch.
Discovery isn't busywork. It's the foundation that makes every later decision easier. Brand colors, typography, voice, the whole sitemap, the design references you'll go shopping with — all of it falls out of one solid intake conversation.
What you need before you build
One thing: a completed intake form. Everything else you'd want — brand basics, reference sites, current-site audit, goals, content, technical specs — is inside it.
I send every client the same Google Form before we start. About 80 questions across 8 sections. It takes them 30 to 60 minutes. Some clients groan; every single one tells me afterwards it was useful.
Preview the form ↗ Duplicate to your Drive ↗
What's in the form, and why each section matters
-
Business basics
Name, contact info, social handles, domain, what they actually sell or do. The plumbing — gets the obvious stuff out of the way so the deeper questions have room.
-
Brand
Logo files, hex codes, fonts, voice and tone, taglines, do's and don'ts. Without this nailed down, every design decision is a guess. If they don't have any of it, you're also doing brand work — price accordingly.
-
Audience
Who the site is for, what they care about, their tech comfort level. Drives copy length, tone, navigation patterns, what content gets prioritized on the home page.
-
Current site, if any
What works, what's broken, what they're embarrassed by. Their answer is usually different from yours — and saves you from "improving" things that were fine.
-
Goals
"Get more donors." "Sell tickets." "Stop people from calling to ask about hours." Specific outcomes, not "look professional." Drives every CTA on the site.
-
Reference sites they love (and why)
3 to 5 sites with a sentence on why for each. The "why" matters more than the "what" — it's what you'll feed Claude when you start building.
-
Content
What they already have written. Photos and videos. What they need help producing. Catches the "we'll just write the copy in Squarespace" delusion before launch week.
-
Design preferences
Words that describe the feel they want. Sites they actively hate. Colors and styles to avoid. The negative space matters as much as the positive.
-
Technical
Hosting, analytics, integrations, payment processors, CRM, email tools. Catches platform constraints early so you don't design something you can't ship.
-
Timeline and budget
When does this need to launch, what's the realistic budget. Reality check. If a 3-week timeline meets a 12-page sitemap meets a $1,000 budget, that's a conversation, not a build.
Why the "why" matters
When a client says "I love this site," that's barely useful. When they say "I love this site because the typography feels confident without being cold, and the spacing makes me feel like the brand has its act together" — now you have something to build with.
Push them. If they can't articulate why they like a reference, sit with them and figure it out together. Is it the color palette? The way headlines and images interact? The pace of scroll? The voice in the copy? You're not just building what they like — you're building something that captures the same effect.
"Every reference site is data. Every reason they like it is the algorithm."
How Nicole and Cordis approached discovery
When Advancing Communities Forward came to me, they had a logo concept and a mission, but no brand guide and no existing website. They filled out the intake form over a weekend.
Their answers shaped everything. They wanted the site to feel "warm, energetic, trustworthy, hopeful — not pity-driven." They sent me three reference sites: a children's hospital foundation (for warmth), a youth empowerment org in Oakland (for energy), and a smaller transitional housing nonprofit (for tone). The "why" mattered: they all avoided trauma photography and led with outcomes, not problems.
That single insight — lead with outcomes, not problems — became the editorial rule for the whole site. Every photo, every headline, every page. You'll see how it cascaded into the sitemap (Chapter 02), the shopping list (Chapter 03), and the final build.
What to do with the answers
After the form comes back, write a short strategy doc. One to two pages. Synthesize their answers into a north star: who is this site for, what's it supposed to do, what's the voice, what are the must-haves and the do-not-include items. This is the doc you'll come back to every time you're stuck.
For ACF, that doc became the project brief I gave Claude when we started building — colors, typography, tone, pages, integrations, all of it on one screen. You can see the actual brand guide here.
Common mistakes
- Skipping the form because the client "doesn't have time." Push back. The form saves you both weeks of revisions.
- Letting them list reference sites without asking why. Lists of links are useless. Reasons are gold.
- Designing before you have brand basics nailed down. Picking colors during the build means the build never converges.
- Building for the boss, not the audience. The site is for visitors. Hold the line.
What's next
Once you have the form back, the strategy doc written, and a clear sense of who/what/why, you're ready to plan the actual site. Next chapter: deciding what pages exist and what goes on each one.