Most websites are built on a guess about what people search for. Ours aren't. We pull the real numbers out of Google's Keyword Planner first, and only then decide what pages exist — how they're grouped, what each one targets and how they link to each other. Sixty pages that each answer a search someone actually makes, beats six hundred that don't.
Most site builds start with a sitemap someone imagined. This one starts with demand data, and the sitemap falls out of it.
Pulled live from Google's Keyword Planner — search volume, competition, CPC and 12-month trend for your services and market. Not a tool's estimate. Google's own numbers.
Python dedupes, scores and ranks every term by real opportunity. Anything below the volume threshold is dropped before it can become a page nobody visits.
Claude groups the surviving keywords into categories and child services — one theme per page, no two pages competing for the same query. That grouping becomes your sitemap.
Home → category → service page, each level linking down and back up. Every page sits in a hierarchy Google can read, not a flat pile of pages with a nav bar.
LocalBusiness, Service, FAQPage, BreadcrumbList — valid JSON-LD written per page from that page's own data, so search engines and AI answers can read what you offer.
Static HTML — no WordPress, no plugins, nothing to hack or update. Deployed from GitHub to Cloudflare Pages on free lifetime hosting, on your own domain.
This is the actual build output, not a mockup. Every service page announces which volume-backed keyword it was given before a word of content is written.
A page that targets a phrase nobody types will rank — and bring nothing. That is the single most common way an SEO site build wastes months. The only defence is deciding what to build from demand data, before anything is written.
The mainstream AI website builders are good at what they do. What they do is not this.
| This build | Typical AI site builder | |
|---|---|---|
| Page count | Up to 100, driven by demand | 5–8 template pages |
| What decides the pages | Keyword Planner volume | A prompt |
| Structure | Home → category → service silo | Flat |
| Structured data | JSON-LD per page, from that page's data | Basic or none |
| Internal linking | Built from the hierarchy, verified | Nav bar only |
| FAQ content | From real search questions | Generated filler |
| Languages | English, Arabic, Urdu, Spanish, French, German, Turkish, Hindi | Add-on |
| Output | Static HTML you own | Locked to their platform |
| Hosting | Your domain, free lifetime hosting | Monthly subscription |
A handyman and home-maintenance company in Dubai. Every figure below is exactly what the pipeline produced in a single run — and the site is live, you can open it.
| Category | Primary Keyword | Child Pages | Schema |
|---|---|---|---|
| Electrical and AC Services | ac repair near me | 8 | JSON-LD |
| Plumbing Services | plumber | 12 | JSON-LD |
| Carpentry and Woodwork | carpentry | 9 | JSON-LD |
| Furniture Assembly and Repair | furniture assembly service | 9 | JSON-LD |
| Doors and Partitions | door lock replacement | 7 | JSON-LD |
The site goes live on your domain — and you get every file underneath it, including the research it was built from.
Every keyword with real search volume, competition, CPC and 12-month trend — the data the whole site was decided from.
Categories, child services, the primary keyword assigned to each page and the questions behind each FAQ block.
Every page, stylesheet, script and image in your own GitHub repo. No platform, no plugins, nothing to renew.
LocalBusiness, Service, FAQPage and BreadcrumbList on every page — validated, not decorative.
Clean URLs that match what the server actually serves — submit to Search Console the day it goes live.
Every internal link crawled and confirmed. The build above shipped 11,865 links with zero 404s.
Every trade searches differently, converts differently and competes differently. The research run is per-industry, so the structure is too.
This is the method above applied to a single trade. The keyword run returned 4,940 searches a month across roofing SEO; the pages below are the clusters that cleared the volume threshold, each targeting the term its own data named.
What to ask an agency before signing, how to read a proposal, and the reporting that tells you whether the work is landing.
Why roofing search behaves differently — storm-driven spikes, insurance-claim intent, and the service-area rules that decide who ranks.
What roofing SEO actually costs, what sits inside each tier, and where the money stops buying anything worth having.
Business Profile, service areas and proximity — the part of roofing search that on-page work alone will never reach.
When paid beats waiting for organic, and how the two run off one keyword set. Pairs with the Google Ads builder.
Your part is a brief and a domain. Everything after that happens on our side.
Most service businesses need both — leads now from paid, and a falling cost per lead over time as organic takes over. The two systems run off the same keyword research.
You send your services and market. The research decides how many pages the demand supports — never charged per page. You keep the files, the repo and the domain.
Multi-language builds, ongoing content, or white-label for agencies? Talk to us.
Tell us your services and market — we'll come back with the real search volumes and the site plan they justify.
Fill this in and we'll message you on WhatsApp to get going.
✓ No contracts · ✓ You own the files · ✓ You approve the plan first