The MX Readiness Model (0-5) — Not Ready, Basic, Structured, Signed, Registered, Audited. Each level shows publisher capability, agent outcome, and improvement steps.

Level 1 — Basic

Can they find you?

The machine does not know you exist. You are not in its knowledge base. At this level, AI systems are crawling the web — building training datasets and fetching pages in real time during user queries.

Requirements

  • robots.txt compliance for training-time crawlers
  • sitemap.xml for both training and inference mechanisms
  • Semantic HTML markup that serves training datasets and live parsing
  • Server-side rendering for JavaScript-heavy content
  • Crawlable structure — quality content that machines can discover and rank

Without Level 1: You do not exist

The machine recommends competitors because you are not in its knowledge base (training gap) and it cannot find current information when users ask (inference gap). You are invisible.

Side benefits: SEO improvement (organic search traffic) and WCAG compliance (semantic structure).

Level 2 — Structured

Do they reference you?

The machine is aware of your site and can recommend it. But awareness is not enough — it needs to extract facts that are accurate enough to cite.

Requirements

  • Fact-level clarity — each statistic, definition, concept needs standalone clarity
  • Schema.org JSON-LD structured data for AI platforms
  • Citation-worthy content architecture optimised for being featured in AI responses
  • Standalone definitions that can be extracted without losing meaning

Without Level 2: Hallucinated details

The machine knows you exist but cannot vouch for you. It fills the gaps with fabricated details, or moves on to a source it trusts more. Your brand is represented by fiction.

Side benefits: GEO (citations in AI-generated responses), SEO (rich snippets), and WCAG compliance (clear content structure).

Level 3 — Signed

Can they evaluate you?

The machine is building comparison lists, sorting by features, evaluating options. This is where purchase intent forms — and where many businesses become invisible.

Requirements

  • JSON-LD microdata at the feature and specification level
  • Explicit comparison attributes (product features, ratings, specifications)
  • Semantic HTML that machines can parse for feature extraction
  • Structured specifications in machine-readable format

Without Level 3: Skipped in comparisons

The machine cannot understand what you offer or how you compare. You are invisible in the shortlist. The customer never sees you as an option.

Side benefits: GEO (AI comparisons), SEO (structured data), and WCAG compliance (clear data presentation).

Level 4 — Registered

Do they parse costs?

The machine needs exact pricing to make recommendations. Ambiguous number formatting, missing currency codes, and unstructured price data cause errors measured in orders of magnitude.

Requirements

  • Schema.org types: Product, Offer, PriceSpecification
  • ISO 4217 currency codes (GBP, EUR, USD — not symbols)
  • Unambiguous pricing structure with explicit decimal formatting
  • Validation to prevent magnitude misinterpretation

Without Level 4: 100x price errors

A river cruise priced at £2,030 was reported by an AI agent as £203,000. European number formatting, no PriceSpecification. The couple never booked. The cruise company never knew why.

Side benefits: SEO (product rich results), GEO (pricing citations), and WCAG compliance (clear pricing).

Level 5 — Audited

Can they complete checkout?

The machine must complete the transaction with confidence. Hidden state in JavaScript, non-semantic buttons, and ephemeral feedback messages create silent gaps that no analytics tool can detect.

Requirements

  • No hidden state buried in JavaScript — state must be DOM-reflected
  • Explicit form semantics (<button> not <div class="btn">)
  • Persistent feedback (role="alert" for important messages)
  • data-state attributes for checkout progress tracking

Without Level 5: Cart abandoned silently

The machine cannot see what buttons do, cannot track checkout progress, and times out. Unlike a human who would phone support, the agent simply leaves. You never see it in your analytics.

Side benefits: WCAG (form accessibility) and user experience (faster checkouts for humans too).

The Cumulative Chain

Each level builds on the one below. Level 1 requires semantic HTML. Level 2 requires structured data. Level 3 requires JSON-LD. Level 4 requires Schema.org. Level 5 requires explicit state. Skip a level and everything above it is unreachable.

Sites that complete the full progression earn computational trust — machines return repeatedly through learned behaviour. Sites that stop short risk being bypassed. Unlike humans who persist through bad UX and can be won back with improvements, machines offer no second chance. You may never see them leave.

Human Users vs AI Agents

The fundamental difference that makes MX critical for commerce:

BehaviourHuman UsersAI Agents
PersistenceHigh — will retry, call support, or use workaroundsZero — encounters a gap once, times out, disappears
Gap visibilityVisible in analytics, heatmaps, support ticketsSilent — invisible in analytics, no trace, no second chance
Trust recoveryCan be won back with marketing, discounts, outreachLearned behaviour — agents favour "reliable" nodes and bypass others
ScaleHundreds or thousands per dayMillions of queries per day, growing exponentially

Without MX vs With MX

LevelWithout MXWith MX
1. BasicCrawls links, hopes for a sitemapStructured metadata, semantic HTML, declared in REGINALD
2. StructuredScrapes paragraphs, risks misattributionFact-level clarity in Schema.org JSON-LD, citation-ready architecture
3. SignedInfers attributes from unstructured textExplicit comparison attributes in microdata — price, features, ratings
4. RegisteredParses strings like "$9.99/mo" and guessesSchema.org Product/Offer with ISO 4217 currency codes
5. AuditedNavigates forms blind, gaps go undetectedExplicit DOM state, semantic forms, persistent confirmation feedback
The core principle: When machines must infer, they hallucinate. When the information is written down, they read it. MX writes it down.

One Implementation, Three Audiences

The same structural patterns that serve MX also serve two other audiences — with zero additional work:

AI Agents

The primary focus. Structured metadata, semantic HTML, Schema.org, and explicit state enable machines to find, understand, compare, price, and transact — without guessing.

Users with Disabilities

Side benefit. Semantic HTML, explicit form labels, role attributes, and persistent feedback are WCAG requirements. MX and accessibility converge on the same technical patterns.

Go beyond the starter kit

The MX Cogify plugins cover the foundational metadata layer — Stages 1 and 2 of the journey. The full Machine Experience encompasses all five stages, including content strategy, accessibility architecture, pricing structure, and commerce readiness across your entire digital estate.

To understand the full picture, read the books (MX: The Protocols and MX: The Handbook) or get in touch for a consultation or audit.