Blue Diamond Medical: a bilingual healthcare platform built for search and AI discovery

GEO architecture

Generative engines summarise sources they can parse and trust. The platform was structured so that each service, condition and doctor is a distinct, citable page.

• One page per medical intent: treatments describe procedures, concern pages describe problems, technology pages describe named devices — so an AI answer can cite the page that matches the question.
• Every page states who the clinic is, where it is (West Springs, Calgary) and what the page covers in its opening lines, rather than burying facts below marketing copy.
• Arabic pages are written for Arabic-speaking patients, not machine-translated duplicates, and carry right-to-left layout.

SEO architecture

• A single metadata source of truth assigns each page its title, description, H1 and primary search intent; a live re-crawl after release found 144 of 144 pages with no duplicate title, description or H1.
• 144 sitemap URLs at the time of verification (85 English, 57 Arabic), plus separate English and Arabic blog sitemaps.
• A registry returns 410 Gone for retired product URLs instead of soft-redirecting them, and a 501-row URL regression dataset built from Search Console history guards every release.

AEO implementation

• New-patient, walk-in and treatment questions are answered directly at the top of the relevant hub pages.
• FAQ content is visible on the page; any FAQ markup mirrors only what a visitor can read.

Entity work

• The clinic, its address, opening hours and services are expressed consistently across pages, schema and llms.txt.
• Physician and service pages connect back to the clinic entity instead of standing alone.

Schema

• MedicalClinic with PostalAddress, OpeningHoursSpecification and ContactPoint.
• MedicalProcedure for treatments, WebSite, BreadcrumbList and FAQPage where the questions are visible.

Content architecture

• Hubs for medical care and medical aesthetics, with treatment, concern and technology pages beneath them.
• A bilingual blog published through FeelStack with its own sitemaps.

Technical implementation

• Server-rendered Next.js so the full text is in the HTML that crawlers fetch, without depending on client-side JavaScript.
• Mobile critical-path work: self-hosted Arabic font loaded only on Arabic pages, lighter navigation code and inlined critical CSS.
• Blue/green releases with a retained rollback slot.

AI crawler readiness

• robots.txt allows public pages and lists the sitemaps.
• A curated /llms.txt summarises the clinic and links its key pages in Markdown.

Search and AI measurement method

• GA4 event model for lead actions (form, phone, booking and directions clicks) with a host guard so QA and non-production traffic are excluded.
• Google Search Console for organic search; AI visibility, when measured, will be tracked as a separate prompt-based dataset and never merged into ranking data.

What we built

• The complete bilingual website and its CMS integration.
• The SEO, schema, llms.txt and analytics layers described above.
• The release pipeline and rollback process the clinic's site runs on.