AI-Readiness Scoring Methodology
friendly4AI scans websites against 48 parameters across 7categories to produce an AI-readiness score (0–100). GEO and AEO are weighted views over different, partly overlapping groups of categories. They are reported alongside the score, not combined into it.
Composite Score Formula
The score is a weighted average of the parameters that apply to the site:
score = weighted average of the applicable, non-excluded parameters
GEO measures how discoverable and understandable the site is for generative AI systems. AEO measures how well the site surfaces as an authoritative answer source. GEO and AEO are weighted views over different, partly overlapping groups of categories. They are reported alongside the score, not combined into it.
Parameter Breakdown
Parameter Categories
Each category contributes a weighted share to the composite score. Category-level weights are published; per-parameter weights are not.
Crawlability & Access
crawlabilityDiscovery & Metadata
discoveryContent Structure
content-structureAI-Specific Signals
ai-signalsAuthority & Trust
authorityTechnical SEO
technical-seoEntity & Schema
entityKey takeaways
48 parameters across 7 categories, released 2026-09-06.
score = weighted average of the applicable, non-excluded parameters. GEO and AEO are weighted views over different, partly overlapping groups of categories. They are reported alongside the score, not combined into it.
47 scored parameters contribute to the 0–100 score; 1 are informational only.
Category weights are published; per-parameter weights are not.
Limitations
What the score proves. That, at the moment of the scan, the page published the signals this methodology checks for — and which of them were missing, with the specific fix for each. Every parameter carries an evidence class saying how well-founded that signal is, from a documented platform requirement down to a heuristic.
What it does not prove. It does not predict that an AI assistant will cite, recommend or rank the site, and it is not a traffic, conversion or revenue forecast. No public methodology can promise that: the assistants do not publish their retrieval or ranking rules, and their answers vary between runs for the same prompt. A high score removes known obstacles to being read and understood; it does not buy an outcome. AI Visibility is measured separately, by asking the models directly, and it is a sample of observed answers rather than a guarantee about the next one.
Comparing scores across versions. Scores are comparable within one methodology version. Version v4.7 can score a page differently from an earlier version with the page unchanged, because parameters are added, retired and reweighted — so a before/after comparison spanning a version change measures the methodology as much as the site. Each report records the version it was produced under, and the changelog records what moved and when.