CiteUnseen

Rubric

Structured data

Rubric v0.8 · 4 checks · 10 of 100 points

Does your markup say the same thing your page says?

Structured data is the one place you state your identity in a form nothing has to infer from prose. It is also the cheapest of these to fix.

Last updated · Written by Chris Brock, who wrote the rubric and signs the audits that use it.

Why this dimension exists

Structured data is the dimension where the work is clear and the payoff is indirect.

It does not make a site rank, and no engine is known to require it. It removes guesswork: markup that declares an organization, connects it to its offerings, names a page's author and agrees with the visible content lets a machine read instead of infer. Markup that contradicts the page is worse than none, which is what the fourth check looks for.

Ten points, four checks, all deterministic: a machine settles each from the response bytes. That makes it the cheapest dimension to fix. Sites often score zero here for a reason unrelated to their markup: a JSON-LD block injected by client-side JavaScript is invisible to a fetch that does not run scripts.

The short version, for somebody asking what schema markup AI search needs, is <a href="../../schema-markup/">on its own page</a>.

The guide for this dimension.

  • Schema markup: which types matter, and why markup that contradicts the page is worse than none.

The 4 checks

4 of these 4 are deterministic, which means the free scan settles them without anybody reading the site.

SD-01 — Valid core entities

3 points · Settled by machine, from the evidence

What is tested. Parse structured data and validate organization, website, person, product, service, software application, or local-business entities as applicable.

What is recorded as evidence. Raw markup, parser output, errors.

What fails it. Invalid syntax; irrelevant types; placeholder values; entities that contradict visible content.

How to fix it.

  1. Add JSON-LD for the entities the site is actually about, starting with Organization and WebSite.
  2. Use the required properties for each type, not just the ones that are easy.
  3. Make every value match what the page says. Structured data that disagrees with the page is worse than none.
  4. Validate before shipping, and again after.

How to confirm it is fixed. Run the page through the Schema Markup Validator and the Rich Results Test and fix every error.

Reference.

SD-02 — Entity relationships

3 points · Settled by machine, from the evidence

What is tested. Review stable identifiers and relationships among organization, people, products, articles, breadcrumbs, and profiles.

What is recorded as evidence. Identifier and relationship map.

What fails it. Duplicate identifiers; disconnected entities; references to nonexistent nodes.

How to fix it.

  1. Connect the entities: an Article to its author and publisher, a Product to its brand and offers, a Person to their organisation.
  2. Use @id so the same entity referenced from two pages is one entity, not two.
  3. Keep the graph honest. Do not assert a relationship that is not real.
  4. Remove orphan nodes that nothing points at.

How to confirm it is fixed. Validate a page and check the graph it produces has the links you intended.

Reference.

SD-03 — Page-specific markup

2 points · Settled by machine, from the evidence

What is tested. Determine whether sampled page types use relevant, supported markup that matches visible content.

What is recorded as evidence. Page type, declared type, matching visible elements.

What fails it. Review, FAQ, product, or article markup for content not visible on the page.

How to fix it.

  1. Give each page type the markup that fits it rather than repeating one site-wide block.
  2. Add Article on articles, Product or Service on those pages, FAQPage where there are real questions and answers.
  3. Add BreadcrumbList that matches the visible navigation.
  4. Only mark up content that is on the page and visible to the reader.

How to confirm it is fixed. Validate three different page types and confirm each carries markup specific to it.

Reference.

SD-04 — Metadata consistency

2 points · Settled by machine, from the evidence

What is tested. Compare titles, descriptions, social metadata, canonical URLs, language, and structured-data values.

What is recorded as evidence. Metadata comparison table.

What fails it. Conflicting titles; obsolete descriptions; incorrect social URLs; wrong language declarations.

How to fix it.

  1. Give every page a unique title and meta description that describe that page.
  2. Keep the title, the H1, the og:title and any structured data name in agreement.
  3. Keep the og:url and the canonical the same URL.
  4. Remove templated descriptions repeated across many pages.

How to confirm it is fixed. List every page's title and description and look for duplicates.

Reference.