"I have an idea but no website" used to mean a week of research, wireframes, and staring at a blank theme. With AI builders you can go from one sentence to a live URL in a single session — if you prompt in the right order. Paste a vague request into v0 or Bolt and you'll get a plausible page that converts no one. Follow a sequence — IA first, then site generation, then section copy — and you get a site you'd actually share.

This is that sequence, using Promptigo's builder prompts as the checkpoints.

Step 1: Plan the IA before you generate anything

The biggest waste is generating pages you'll delete. Fix the structure in 2 minutes first with the Website Information Architecture Planner:

Plan the IA for [product] aimed at [audience] with goal [goal — e.g., capture emails, book demos]. Output: page list, navigation order, and what each page must prove to the visitor.

You're looking for what each page must prove, not just its name. "Pricing must prove you're cheaper than the pain of staying put." "Docs must prove I can integrate in 10 minutes." If a page can't answer its proof, cut it before you build it.

Output you want: 4–6 pages max. For a SaaS launch: Home → Features → Pricing → FAQ → Contact. For a content play: Home → Blog → About → Subscribe. Fewer pages that each have a job beats a "complete" site that's diffuse.

Step 2: Pick the stack you'll still want in a month

All our builder prompts offer stack as a variable. Choose like this:

  • Plain HTML/CSS/JS — fastest to host anywhere, zero build step. Pick for waitlists, resumes, or a restaurant site you won't update daily. Our Build a Waitlist Page and Build a Personal Resume Site shine here.
  • React + Tailwind or Next.js + Tailwind — pick if you'll add a blog, auth, search, or API later. Next.js gives you image optimization, routing, and SEO metadata free. React is the middle ground if you want components without server rendering.

Don't optimize for trend. Optimize for who will edit this next month — you, a designer, or no one. That decides the stack more than tech does.

Step 3: Generate the site in one shot — but brief it like a client

Now use the builder prompt that matches your type. For most ideas, that's Build a SaaS Landing Page — it's the most forgiving single to repurpose:

Build a [site_type] for [product] aimed at [audience]. Pages: [pages from IA]. Include sections: [hero, features, pricing, FAQ, CTA]. Stack: [stack]. Brand colors: [colors].

The variables that matter most:

  • audience with specificity: "Solo founders who ship with AI but lose leads because their site looks like a template" produces sharper copy than "founders."
  • content_sections as Jobs: "Hero that says what you do in 6 words, feature sections that each claim an outcome, pricing that makes the middle tier obvious" beats listing "features, pricing, FAQ."
  • brand_colors as constraint: Two colors plus a neutral. Constraints force taste; "any blue" produces corporate mush.

Run this in v0, Bolt, or Lovable and you have a working site — not a mockup, but rendered code you can iterate on. Treat it as the first draft, not the final.

Step 4: Rewrite the three sections that decide conversion

AI builders generate plausible copy. Plausible doesn't convert — specific copy does. Rewrite these three sections with dedicated prompts before you publish:

Hero — says what you do in one glance. Use Homepage Hero Section:

Write a homepage hero for [product] for [audience] where the primary benefit is [benefit]. Need headline, subhead, and CTA.

Test: if a stranger can't say what you do after 3 seconds on the hero, cut words until they can. The prompt forces that constraint — headline under 8 words, subhead as supporting proof.

Feature — outcomes, not mechanics. Use Feature Page Copy:

Explain [feature] for [product] in terms of the outcome it produces for [audience], not how it works.

"AI generates your pricing page" is mechanics. "Publish your pricing in 2 minutes instead of waiting on a designer for a week" is outcome. The prompt bans mechanics-by-default — that's its value.

Pricing — make comparison easy. Use Pricing Page Copy:

Write pricing tiers for [product] for [audience] with offer [offer]. Make the recommended plan obvious and every tier comparable on the same axes.

The killer detail is "comparable on the same axes." Without it, pricing pages list different features per tier and force mental gymnastics. With it, the visitor scans down a column and sees the jump.

Step 5: Remove doubt before you launch

Two more prompts prevent the bounces that analytics will later surface — fix them now:

IA stress test. Re-run the IA planner as a critic:

Critique this IA for [product]: [pages]. Which page would a skeptic visit first, and what question would make them bounce if unanswered?

If the answer is "Pricing — they'll wonder if there's a free tier," add that answer to pricing before you ship.

FAQ — objections, not trivia. Use Website FAQ Section:

Turn these real objections for [product] from [audience] into an FAQ that removes doubt.

Don't invent FAQs. Paste the last 5 questions you got about the product, or the comparison a visitor actually does ("How is this different from [competitor]?"). Real objections in, researched answers out.

Step 6: Ship — then iterate with the copy helpers

Publish the site on the stack's easiest host (Vercel for Next.js, Netlify for plain HTML) and let real visitors tell you what's next. When they do, the 17 website-* copy prompts are your iteration kit without regenerating the whole site:

  • website-trust-and-social-proof-section turns scattered testimonials into evidence
  • website-exit-intent-popup catches bounces without a desperate discount
  • website-seo-meta-description-batch writes batch meta for the pages you just made
  • website-404-page-copy and website-empty-state-copy fix the dead ends the builder left as placeholders
  • website-blog-cta-optimizer rewrites the CTA at the end of each blog post you add after launch so it matches the reader's intent

You don't need to memorize them — browse by problem. The pattern is: build once with a build-* prompt, polish forever with website-* prompts.

What to do when your idea changes (it will)

Your first site won't be your last. After a week of real traffic you'll know which page earns its keep and which is dead weight. Don't rebuild from scratch — ask the planner again with what you learned:

Based on this live site: [URL or paste hero + nav], what page should I add, remove, or rewrite first given that visitors mostly do [behavior from analytics]?

That keeps the site evolving with evidence, not hunches. One more generation with a tighter brief beats a month of tweaking fonts.

The workflow in one line

IA plan → pick stack → generate site → rewrite hero/feature/pricing → FAQ → ship → polish with website-*.

That order prevents the usual failure: a beautiful site with generic words that no one acts on. Start with the Website Information Architecture Planner and Build a SaaS Landing Page, then polish with Homepage Hero Section and Pricing Page Copy — all free to copy in the Promptigo library.