MKEO / Framework
Each layer answers one question a machine asks about you — and has a deliverable and a verification, so the framework is a work plan, not a poster.
Everything you want cited must be machine-readable without a browser: semantic HTML, server-rendered text, structured data (schema.org), llms.txt at the root, and no content that only exists behind JavaScript.
Deliverables: clean page templates · JSON-LD (Organization, DefinedTerm, FAQPage) · llms.txt · sitemap.xml
Verified by: site audit — every key page fetchable and parseable, zero JS-only content
Machines resolve claims to entities. One name, one identity, everywhere: the site's Organization schema links out via sameAs to every profile that represents the same entity, claims are sourced, and honest statements (like "this is a coined term") beat inflated ones — engines discount unverifiable superlatives.
Deliverables: sameAs graph (target ≥4 platforms) · platform profiles live · sourced claims
Verified by: entity audit — sameAs count, profiles live, third-party mentions
Content must exist where the questions are: pillar pages per product line, cluster articles per question, all cross-linked so nothing is an island. New pages need at least two inbound internal links — orphan pages are invisible to crawlers and unusable to answer engines.
Deliverables: pillar-cluster content plan · published pages with internal links · watchlist of external mentions
Verified by: orphan-page count (target 0) · published-vs-planned ratio · search coverage per keyword
Everything above is monitored on a schedule: AI recognition and recommendation probed across engines, search visibility per keyword and region, site audits gated by thresholds. A baseline is taken before any work starts; trends are reported against it. See metrics.
Deliverables: monitoring schedule · baseline report · periodic trend reports
Verified by: reproduced numbers from the open-source reference implementation