Case Study: How We Built Simeon Creatives as a Site, Search, and Intake System
A first-party look at our own studio stack: Cloudflare-hosted knowledge center, Search Console demand on topics we actually ship, and a contact path that logs every brief. No invented revenue percentages.
Agencies love case studies about other people. This one is us. Simeon Creatives had the same problem we diagnose in clients: a studio that can ship brand and sites, but whose own digital presence is either a brochure or a pile of disconnected tactics.
We treated simeoncreatives.com as a growth system: one offer, a fast site, a knowledge center that search and AI can read, and an intake path that does not lose briefs. The commercial frame is in the growth cornerstone. This is what that looks like when we run it on ourselves.
Executive snapshot
| Layer | What we run | Why it exists |
|---|---|---|
| Site | Vite and React on Cloudflare Pages, images on R2 | Fast templates, no plugin content management system (CMS) on the public origin |
| Search | Hub-and-spoke blog, sitemap from source, SEO HTML shells, Markdown twins, llms.txt | Earn queries we can actually deliver, and stay citable |
| Intake | Contact panel → Apps Script → Sheet, honeypot, required fields | Every brief is a row. Reply is a written procedure, not a memory |
flowchart LR
Offer["One offer"] --> Site["simeoncreatives.com"]
Site --> Search["Hubs · Console · Markdown"]
Site --> Intake["Form → Sheet"]
Search --> Talk["Conversation"]
Intake --> TalkThe problem we were solving on our own domain
- Offer drift: branding, websites, search, and automation are four services. Without one commercial story they look like four agencies.
- No place to teach: a work grid is proof. It is not how a stranger learns whether we are the right fit.
- Intake as chat: briefs in DMs do not become a pipeline. They become forgotten screenshots.
1. Site: an edge knowledge center, not a CMS theme
The public site is a Vite and React app on Cloudflare Pages. Media goes through Cloudflare R2 so the origin stays thin. We generate a sitemap from source, and at build time we stamp search-engine HTML shells so crawlers are not guessing a blank JavaScript app. The architecture class is the same one in monolithic CMS vs serverless edge.
Content is a knowledge center, not a dump. Five hubs. Each hub has a cornerstone and spokes. Branding, websites, search, automation, and growth. That is the offer, taught in public, with service pages that still say the same thing.
For machines we also ship /for-robots, llms.txt, and Markdown at the same path as the article. HTML stays canonical. Markdown is the export. That is Generative Engine Optimization (GEO) as a file, not a slogan.
2. Search: we publish what we can deliver, then we watch Search Console
We do not invent a keyword we cannot fulfil. AI automation articles exist because we ship that work. Brand-system articles exist because we ship brand books. Google Search Console (GSC) is how we check whether the index noticed.
The extract below covers 13 July to 15 August 2026: 890 impressions, 18 clicks, a 2% click-through rate (CTR), and an average position of 50.7. Impressions climbed in August. Clicks did not keep pace. That is early visibility, not a ranking win.
The queries Google is showing us cluster on work we actually publish. AI business process automation at 272 impressions. AI business automation at 178. The n8n versus Zapier versus Make comparison is in the list. Those rows still show zero clicks. We are on the results page. We are not the click yet.
Search Console Insights logged a 15-click milestone on 11 August 2026, and 13 of 20 toward the next 28-day click goal. We keep that as an ops cue, not a vanity number.
The playbook for originality versus generic rewrites is in the information gain framework. The playbook for the hosting and information architecture (IA) rebuild is in the site architecture organic-growth case study. We are not going to paste that 250% figure here as if it were a second, separate miracle.
3. Intake: a form that becomes a row
The contact panel collects name, email, phone, services, budget, timeline, and a message. A honeypot field stays empty for humans. The payload posts to a Google Apps Script web app and lands in a Sheet. That is the store we actually run on this domain today.
- Standard operating procedure (SOP): every submission is a row. Required fields match what we need to reply. Owner is Simeon. Done means logged and a same-day response when we are working.
- Why a sheet is allowed: studio volume. One operator. The process is written. When volume or privacy needs outgrow a spreadsheet, the next store is PostgreSQL plus a worker, as in our forms and n8n articles.
- What we did not do: pretend a chatbot is intake. Pretend n8n is already on this origin. The public repo would contradict that.
The volume pattern (edge validation, Row Level Security (RLS), async n8n) is documented for when you need it: dynamic forms and lead routing with n8n and Supabase.
What we will not claim
- No fake revenue: this article does not invent a close-rate or a shilling figure. Conversations and shipped work are the commercial proof. Search Console is the visibility proof.
- No tool cosplay: we describe the stack in the repo. Sheets today. n8n when the SOP outgrows it.
- No “set and forget”: hubs need new spokes. GSC needs a human. Intake still needs a reply.
What you can copy this quarter
- Write the four-line offer. Put it on the homepage and in outreach. See the channels spoke.
- Pick one hub you can actually deliver. A cornerstone plus three honest articles beat twenty generic posts.
- Make intake a row. Sheet or database. Same SOP. Same-day reply.
- Watch GSC weekly. Impressions tell you which spoke to write next. They do not replace sales.
If you want this system on your domain, start at /contact. The work examples that sit beside this site are on /work.
Frequently asked questions
Is this the same as the 250% impressions architecture case study?
That piece is the search-architecture playbook. This one is the commercial system around our own domain: offer, knowledge center, Search Console, and intake. We link it instead of restating the hosting migration as a second miracle number.
Do you run n8n on simeoncreatives.com today?
The public site writes contact briefs to a Google Sheet through Apps Script, with a honeypot and a required-field SOP. When volume or access control outgrows a sheet, we use the n8n and Supabase pattern documented in our automation case study. The SOP stays. The store changes.
Why publish Search Console screenshots of your own site?
Because impressions on queries we actually write about are evidence. They are not revenue. The current window shows 890 impressions and 18 clicks at an average position of 50.7. We show Google Search Console (GSC) so readers can see visibility arriving, then we say what we do not claim.
Can a smaller studio copy this stack?
Yes at a smaller scale: one offer, a fast site, a few honest articles, a form that logs, a same-day reply rule. You do not need five hubs on day one. You do need the four-line offer and an intake SOP.