The Readiness Signals

How this maps to your score: these are the underlying readiness signals this article explains. The served ClickRadius AI Readiness Score is composed of five weighted evidence families — Off-site Authority, On-page GEO, Schema Engine, Entity / E-E-A-T and Crawler Access. SEO Health and measured AI Presence are reported as separate scores and are never blended into it.

By ClickRadius · Published April 8, 2026

Our companion article, What Is an AI Readiness Score?, introduces the 0–100 readiness measure. This article documents the five evidence families that make up that score, their current weights, and what each family does and does not establish. AI Readiness is an input-oriented measurement; it is not a substitute for rank data or for observed AI Presence.

Why these signals, and why weighted

The served weights are Off-site Authority 28%, On-page GEO 26%, Schema Engine 18%, Entity / E-E-A-T 18%, and Crawler Access 10%. They are labeled starting estimates and have not been fitted to customer outcomes. The score therefore describes the currently measured readiness evidence; it does not forecast a citation, ranking, traffic, revenue, or score change.

Evidence coverage and age matter as well as the number. Missing evidence remains unknown rather than becoming zero or “fine,” and insufficient measured coverage can withhold the score instead of producing a confident number from a thin denominator.

Family 1: Off-site authority setup (28%)

The largest family asks whether the places AI engines can read about you are set up and consistent: directory listings, the review base, syndication and social footprint, links from other sites, and the entity record. This is evidence about third-party and profile surfaces, not a promise that an external engine has read or credited them.

Family 2: On-page GEO / content kernel (26%)

Whether your pages contain extractable answer passages, clear structure, specific claims, and content that is not purely promotional. This family evaluates on-page inputs; it does not establish that an engine retrieved, selected, or cited a passage.

Family 3: Schema markup, engine-weighted (18%)

Whether pages carry structured data describing the business, services, articles, and other supported page types in a machine-readable form. Schema can reduce ambiguity about what a page declares; it does not compel an engine to crawl, index, rank, recommend, or cite that page.

Family 4: Entity / E-E-A-T setup (18%)

Whether the available evidence describes a recognizable entity: a complete entity record, identifiable authors with supported credentials, and consistency between the site and independent sources. This is a readiness input, not proof of an external engine's entity resolution.

Family 5: AI crawler access + llms.txt (10%)

Whether the robots policy declares access for relevant AI crawlers and whether the site publishes an llms.txt. The boundary is important: this family reports the policy declared by the site. It does not establish that an engine reached, rendered, indexed, or cited a page.

Beside the score: SEO Health and AI Presence

Two things deliberately sit outside the AI Readiness Score rather than inside it. SEO Health covers conventional Google hygiene — meta and head signals, technical crawlability, security headers and page performance — and is reported as its own 0–100 score. AI Presence measures the output rather than an input: whether engines are actually mentioning or citing you, expressed as a percentage with a sample count. Neither is blended into AI Readiness, because a composite that mixes inputs with outputs hides which part is weak.

The order of operations

A review can proceed without pretending that one sequence fits every site:

1. Establish the evidence state: identify unknown, insufficient, stale, and currently measured inputs before interpreting a score.
2. Review access and raw-page availability: treat robots and raw HTML as site-controlled evidence, not proof of an external crawl.
3. Use the five-family breakdown: prioritize supported weaknesses without relabeling SEO Health or AI Presence as readiness components.
4. Prepare exact revisions: retain the proposed content or technical change for review with its evidence and scope.
5. Gate every site effect: application requires current authorization for the exact revision and a connection proven capable of that effect; otherwise the work remains for review or customer implementation.
6. Observe separately: reread supported changes at the independent public origin where possible, and keep monthly AI Presence observations separate from provider acceptance and readiness scoring.

Five families, one gauntlet: reach me, parse me, understand me, verify me, trust me. Clearing all five improves the inputs an engine can evaluate; the engine still decides whether and when to crawl, index, rank, or cite the site.

— ClickRadius Institute

Off-site Authority is already one of the five readiness families; it is not a sixth score or an add-on to the model. AI Presence is different: it is a separate observed output. ClickRadius measures the five-family score, prepares eligible corrections, supports entity and authority work, and records monthly observations across ChatGPT, Gemini, Perplexity, Claude, and Grok. Microsoft Copilot is not currently supported. Applying an exact supported revision requires current authorization and a connection proven capable of that effect; otherwise it remains for review or customer implementation. External engines decide their own outcomes.

How the evidence families relate

The family values are combined by the served model, but no one family turns another into proof of an external outcome. Three distinctions matter during review:

• Declared access is not observed access. A permissive robots policy and raw HTML are useful site-controlled inputs, but neither proves that an external engine reached the page.
• Schema and prose are different evidence. Structured data describes the page in machine-readable form; the visible content must still support the facts it declares.
• Age changes interpretation. A measured value without current observation context must not be presented as current merely because a row or score still exists.

No universal timetable

The work required depends on the evidence actually measured, the site's technology, the selected effect, current authorization, and demonstrated connector capability. A readiness change cannot honestly be assigned a universal number of days, and it cannot be converted into a promised external-engine result.

Frequently asked questions

Which AI Readiness family has the largest weight?

Off-site Authority is 28%, followed by On-page GEO at 26%, Schema Engine at 18%, Entity / E-E-A-T at 18%, and Crawler Access at 10%. These are current starting estimates, not weights fitted to customer outcomes.

Which family should I review first?

Start with missing or stale evidence and declared-access blockers, then review the largest weaknesses shown by the measured family breakdown. A crawler-access finding describes policy and raw-page availability; it does not prove that an external engine reached or used the page.

Can a site have a high AI Readiness Score and still have no measured citations?

Yes. AI Readiness measures weighted inputs. AI Presence separately records monthly observations across supported engines, and those external engines independently decide whether and when to crawl, index, rank, mention, recommend, or cite a site.

Review the measured evidence. Get your free AI Readiness Score across all five evidence families, or see plans for governed correction workflows and monthly five-engine AI Presence observations.