AI Visibility Readiness methodology

You submit a public URL (typically homepage or a key service page). The system fetches HTML with SSRF protections and evaluates on-page signals associated with entity clarity, extractability, trust, and crawl eligibility proxies.

• In scope: visible entity/offer language, structured data when present, proof and comparison cues, FAQ-like extractable blocks, basic robots/meta indexability signals in HTML.
• Out of scope for this run: live prompts against ChatGPT, Gemini, Perplexity, or other answer engines; private brand monitoring datasets.

1. Instant readiness (this tool): entity, offer, extractability, evidence, trust, comparison, schema, crawl eligibility.
2. Actual visibility (manual prompt tests / ongoing monitoring): defined prompts, platform, date, mentions, citations, competitors.

Layer 1 is a prerequisite checklist. Layer 2 is empirical. High readiness without prompt testing is incomplete; low readiness often predicts weak extractability even when brand searches still work in classic SEO.

Weights encode product judgment about what public HTML can support for answer-style discovery. They are not a claim that any single model uses these weights internally.

Example: a professional-services homepage scores 49. Entity name is clear, but offer categories are vague, case studies are only linked as “Clients,” no FAQ or definition blocks exist, and schema is limited to a bare WebSite node. Crawl eligibility looks fine (no noindex).

• Prioritize extractable facts: add plain-language “who we serve / what we do / proof” blocks with concrete nouns, not only brand adjectives.
• Entity + offer coupling: pair organization name with service lines and geography or ICP so answers can attribute correctly.
• Schema is supporting evidence: Organization/Service JSON-LD helps machines parse what humans should already see. Schema alone will not rescue a vague page.
• Next measurement step: run a fixed prompt set monthly (category + competitor + “best for X”) and log citations separately this tool does not replace that log.