Pillar/Cluster Model
- Pillar page: broad topic, targets head keyword (e.g. "SEO Guide")
- Cluster pages: subtopics, target long-tail (e.g. "SEO meta tags", "SEO schema")
- Cluster pages link up to pillar
- Pillar links down to all cluster pages
- Cluster pages cross-link contextually
Anchor Text Rules
- ✅ Descriptive, varied, natural ("learn how to validate JSON-LD")
- ❌ Generic ("click here", "read more")
- ❌ Over-optimized exact match (looks spammy, can trigger penalties)
Local vs Global Architecture (2026)
URL structure (placeholders, no real city):
/services/[service]/ → global pillar — primary "[service]", NO city in title/H1
/zones-intervention/[region]/ → regional hub — groups the local pages, owns "[service] [region]"
/zones-intervention/[city]/ → local page — Map Pack target, owns "[service] [city]"
Link mesh:
- Global pillar
/services/[service]/→ down to each local page with a geo-specific anchor ("[service] in [city]"), and each local page → back up to the pillar. - Regional hub
/zones-intervention/[region]/→ down to its city pages, and back up. - Keep navigation/footer lean: link the hub, not a grid of every city (a city footer grid dilutes PageRank and triggers near-duplicate suppression).
Anti-cannibalization (local/global): the primary keyword is exclusive per page — pillar = [service] (no city), local = [service] [city]. See seo-cluster for the one-intent-one-URL rule and when to split.
Doorway pages — quality, not a number
Doorway risk is a quality judgement, not an absolute page-count threshold. One local page per documented presence/service area, each differentiable on >= 3-4 dimensions (local testimonials, local FAQ, geo context, area-specific services/hours). Risk signal: 70-90% identical content across local pages (templated city-name swap) → near-duplicate suppression.
Audits
- Orphan pages: pages with 0 internal incoming links
- Click depth: every page reachable in ≤ 3 clicks from homepage
- Link distribution: no page concentrates >10% of internal links
- Broken internal links: 404s within site
Workflow
- Crawl site (cheerio + breadth-first)
- Build link graph (page → outgoing links)
- Compute click depth from homepage
- Detect orphans and dead ends
- Suggest contextual link insertions based on topic clusters