The MASTERY-AI Framework
An open methodology for AI search visibility. Version 3.2.0: 8 weighted pillars and 149 atomic factors, all of them published.
Most businesses have no idea how AI search evaluates their website. Traditional SEO metrics tell part of the story, but AI systems weigh different signals. Without a framework, optimisation is guesswork.
This page is the summary. The specification, the factor definitions and the scoring code all live in the repository, under the MIT licence, so you can check any claim on this page against the source.
github.com/TheWayWithin/mastery-ai-framework (opens in new tab)
8 Pillars of AI Visibility
Each pillar carries a weight, and the eight weights add up to 100 percent. The factor counts below are the v3.2.0 counts and total 149.
M — Machine Readability & Technical Infrastructure
22 factors, 15.0%Can AI systems reach and parse your content? Technical implementation, llms.txt and robots.txt.
A — Authority & Trust Signals
15 factors, 17.8%The credibility and trust indicators an AI system can actually verify.
S — Semantic Content Quality
22 factors, 13.8%Content depth and semantic richness: whether the page says something specific enough to quote.
T — Topical Expertise & Experience
14 factors, 8.9%Whether the site demonstrates real expertise in its subject rather than asserting it.
E — Engagement & User Experience
19 factors, 10.9%The user experience signals that carry over from ordinary web quality.
R — Reference Networks & Citations
19 factors, 5.9%External validation: who else refers to you, and how credibly.
Y — Yield Optimization & Freshness
15 factors, 4.0%Whether the site is maintained: freshness and continuous improvement over time.
AI — AI Response Optimization & Citation
23 factors, 23.7%The heaviest pillar. Optimisation aimed directly at how AI systems assemble an answer, including MCP integration.
Open Methodology
Every factor and every weight is published. There is no black box and no proprietary scoring to buy access to. You can read the framework, disagree with a weight, fork it and change one.
That is deliberate. A scoring system nobody can inspect is just an opinion with a number attached to it.
M — Machine Readability & Technical Infrastructure
Can AI crawlers reach your content? This is the foundation: if an AI system cannot read the site, nothing else on this page matters. The pillar covers robots.txt configuration, the presence and quality of an llms.txt file, crawl accessibility and the technical structure of the content.
Version 3.2.0 added AI bot access configuration here, which is the specific question of whether your robots.txt allows the crawlers that feed AI answers, such as OAI-SearchBot and GPTBot.
22 factors, weighted at 15.0%
A — Authority & Trust Signals
Does the site show credibility an AI system can check? This pillar measures trust indicators and the signals that separate a source worth citing from one worth skipping. AI systems weigh these differently from traditional search engines, which is why a site can rank well and still go uncited.
15 factors, weighted at 17.8%
S — Semantic Content Quality
Is the content deep and specific enough to be worth quoting? This pillar measures semantic richness and content depth: whether a page makes claims concrete enough for an AI system to lift and attribute, or whether it is generic enough that any competitor's page would do instead.
22 factors, weighted at 13.8%
T — Topical Expertise & Experience
Does the site demonstrate expertise in its subject, or only claim it? This pillar looks for the evidence of experience: depth across a topic rather than one thin page, and first-hand knowledge rather than a restatement of what everyone else has already written.
14 factors, weighted at 8.9%
E — Engagement & User Experience
The user experience signals that carry over from ordinary web quality: whether the page is usable, fast enough and readable by a human as well as a machine. AI systems inherit a good deal of their judgement from web quality signals, so this pillar rarely wins a citation on its own but regularly loses one.
19 factors, weighted at 10.9%
R — Reference Networks & Citations
Is the business referenced elsewhere? AI systems triangulate. When several credible sources say the same thing about you, a model treats the claim as safer to repeat. This pillar evaluates external references, the quality of those citations and the health of the link network around the site.
19 factors, weighted at 5.9%
Y — Yield Optimization & Freshness
Is the site maintained? This pillar measures freshness and continuous optimisation: whether content is kept current, and whether the site shows signs of being actively looked after rather than published once and abandoned.
15 factors, weighted at 4.0%
AI — AI Response Optimization & Citation
The heaviest pillar in the framework, and the one with the least crossover from traditional SEO. It covers optimisation aimed directly at how AI systems assemble an answer: content format, conversational relevance, direct answer provision, and Model Context Protocol integration.
It is also where most sites have done the least work, which is the whole reason the framework weights it first.
23 factors, weighted at 23.7%
MASTERY-AI Framework v3.2.0. Pillar names, weights and factor counts on this page are taken from the repository README (opens in new tab). Page last reviewed 13 September 2026.