The Two Layers
Every Framix engagement scores two distinct layers separately.
| Layer A — Agent Readiness | Layer B — GEO Visibility | |
|---|---|---|
| The question | Can agents technically discover, fetch, parse, and act on the site? | Do AI systems mention, cite, and recommend the brand? |
| Reference tool | IsItAgentReady (Cloudflare); Cloudflare URL Scanner agentReadiness mode | Profound, Semrush AI Visibility, Scrunch, Adobe LLM Optimizer, AthenaHQ, Peec AI |
| Primary owner | Engineering | Content + SEO; engineering for instrumentation |
| How we win | Headers, well-known endpoints, semantic HTML, markdown alternates, schema | Entity clarity, E-E-A-T, answer-first structure, citation-friendly blocks, freshness, primary sources |
| Measurement | IsItAgentReady score and level + Framix scorecard | Citation Share of Voice, AI referrer traffic, mention rate |
| Failure mode | Agent ingests inflated tokens, misparses content, or skips the site | Agent quotes a competitor in the answer the brand should own |
Why the split matters
Section titled “Why the split matters”Scanners only see Layer A. A perfect IsItAgentReady score does not guarantee citations — Layer B drives those, and no scanner measures it. Conversely, strong brand authority can earn citations on a technically poor site, but the engineering layer determines how much of that authority survives ingestion.
The scoring rubric makes the split operational: Pillars 1, 2, 4, 5 measure Layer A; Pillar 3 (Answer-Engine Quality, 30 points) is Layer B. Pre-launch gates sign off the layers independently — Gates A and C cover Layer A, Gate B covers Layer B.