An SEO, AEO and GEO program across six websites for a multi-state cash home buyer
A US cash home buying company operating across several states had a parent website and five state websites, with about six indexable pages between them. We delivered an SEO, AEO and GEO audit and growth roadmap on 2 September 2026, then started a location-by-location rollout that the founder's AI coding agent carries out and we review. No performance figures yet.
- Audit
- Roadmap
- Location plan
- Agent rollout
Snapshot
- Client type
- Cash home buying and wholesaling company
- Country
- United States, operating across several states
- Industry
- Real estate: cash home buying and wholesaling
- System
- SEO, AEO & GEO
- Use cases
- Search & AI visibility
- Stack
- Hand-built sites on Cloudflare Pages · the founder's Claude-based AI coding agent
- Timeline
- Audit and roadmap delivered 2 September 2026 · first location plan 14 September 2026 · in progress
- Headline numbers
- No performance figures yet · portfolio score 32/100 at audit, 2 September 2026
What did the client need?
Six well-designed websites that almost nobody could find, in a market where most searches name a service and a city.What did we build?
An audit and phased roadmap for all six sites, then a single written plan per location for an AI coding agent to carry out.How does it run day to day?
The agent works one phase at a time and reports back; we review every report, and people handle anything that needs a real address, photo, review or legal check.
What did the client need?
The parent site and five state sites had warm, specific copy and careful legal wording, but almost none of it could be found in search. The audit found:
- About six indexable pages across six domains. Each site was a single page plus privacy and terms pages, with no page for any city or seller situation.
- No local or off-site presence. No Google Business Profile, address, analytics, structured data or reviews, and no mentions of the domains elsewhere on the web.
- Trust and compliance defects. Placeholder text was live on the state sites, and testimonials were duplicated across sites under different names and cities.
- Heavy duplication between state sites, which risks them being treated as doorway pages.
The portfolio scored 32 out of 100 at audit, and the lowest site scored 26.
What did we build?
The audit and roadmap. We scored every site on crawlability, metadata, headings, content, internal linking, structured data, local SEO, off-site authority, trust, performance and AI visibility, then set out a phased plan.
| Phase | Timing | Focus |
|---|---|---|
| 0. Emergency fixes | Week 1 | Trust and compliance: remove placeholder text and fabricated social proof, fix broken links and headings |
| 1. Foundations | Weeks 1 to 4 | Analytics, Search Console, sitemaps, schema, title and meta rewrites, Google Business Profiles and citations, reviews, an AI visibility baseline |
| 2. Content & architecture | Weeks 5 to 12 | City pages, seller situation pages, hubs, FAQ, about and reviews pages, a multi-step offer form |
| 3. Authority | Months 4 to 6 | A net proceeds calculator, case studies, outreach to "best cash buyers" lists, a blog and comparison pages |
| 4. Compounding | Months 7 to 12 | Building on what works |
The roadmap sets targets, such as 60 or more indexable pages at 90 days and 150 or more at 12 months. They are targets, not results.
One plan per location. Rather than prompting the agent piece by piece, we write the whole plan for a location as one markdown file: a scope guard naming the one domain in scope, operating rules, and tasks with IDs, a type (agent, human or both), dependencies, instructions and testable acceptance criteria.
- Hard guardrails for the agent. Never fabricate testimonials, addresses, hours, legal deadlines or market data. Never add review schema without real reviews. Never build city pages by find-and-replace: each needs at least three facts true only of that city and less than about 30% overlap with the sibling sites. Never remove consent or legal wording. Stop and ask when a task needs a person.
- Local angles first. Each state's own foreclosure rules, which competitors explain poorly, go to the top of that location's plan.
- Built to be reused. Page templates, schema and URL patterns from the first location are documented so the next state can lift them.
- Organic and paid together. City and situation pages are planned to mirror the Google Ads campaign structure.
- AI search visibility. A set of 20 prompts to track whether AI assistants mention the brand, an `llms.txt` file, and outreach to the "best cash home buyers" articles that AI answers draw on.
How does it run day to day?
Location plan ─► AI coding agent (own branch) ─► report ─► our review ─► next prompt
└─► human-only tasks ─► owners and team- The agent already has access to the sites. It works through the plan phase by phase on its own branch, then reports back. We review each report and send the next prompt.
- Human-only tasks are flagged to the team. Addresses, Google Business Profile verification, real photos, real reviews and legal checks stay with people.
- The owners handle Google Business Profiles, because they need the registered address. The first profile is for the state where the business is registered, with the other state sites linked to it.
- One location at a time. The lowest-scoring site went first, then the other state sites follow, then the parent site, which mostly sends visitors on to the state sites.
What were the results?
There are no performance figures yet. We delivered the SEO audit and growth roadmap for all six websites on 2 September 2026 and the first location's execution plan on 14 September 2026. That location's homepage fixes are done and its later phases are in progress. The other state sites follow. The roadmap's numbers are targets, not results.
| Part | Status |
|---|---|
| SEO audit and growth roadmap (all six sites) | Done, delivered 2 September 2026 |
| First location's execution plan | Done, delivered 14 September 2026 |
| First location: phase 1 | Sent to the agent, report reviewed, next steps sent |
| First location: homepage fixes | Done |
| First location: later phases | In progress |
| Other state sites, then the parent site | Planned |
| Google Business Profiles | Planned, handled by the owners |
- 32/100Portfolio score at audit2 September 2026
- ~6Indexable pages across six domains, at audit2 September 2026
- 14 Sep 2026First location's execution plan deliveredDelivery date
What did we learn, and what are the limits?
- Pages first. Nothing else in SEO matters until there are pages to rank. Six one-page sites can't compete for city searches.
- Fix trust before growth. Placeholder text and duplicated testimonials were the cheapest problems to fix and the most damaging to leave.
- A full plan beats a string of prompts. One markdown file per location, with task IDs, guardrails and acceptance criteria, lets an AI agent do the bulk of the work while we review and steer.
- Write the guardrails down. An AI agent writing location pages will invent facts unless told not to. Clear rules on testimonials, addresses, legal claims and duplication keep the sites honest.
- Pilot on one location, then reuse. The lowest-scoring site has the most room to improve, and documenting its patterns makes each later state faster.