Structured Data Graph skill
The playbook Moonrank's SEO Agent uses to turn scattered schema blocks into a single consolidated entity — the thing answer engines actually ground on when they decide who you are.
Download the methodology, see the derivation table, or run it live inside the SEO Agent.
- Category
- SEO
- Agent
- SEO Agent
- Format
- structured-data-graph.md
- Price
- Free to download
The skill file
What is the Structured Data Graph skill?
The Structured Data Graph skill is how Moonrank's SEO Agent builds structured data as a system rather than decoration. One module owns all the JSON-LD. The site-wide entity — Organization, WebSite, Person — is emitted once and referenced everywhere by @id, so search engines consolidate a single entity instead of merging forty near-identical copies of it.
The second half is derivation. Every page-level schema is computed from the page's own content: an FAQPage only where a real FAQ block exists, a HowTo only where ordered steps exist, a word count calculated rather than typed. A field that has to be typed twice is already out of date.
It's written as a SKILL.md, a plain-language instruction format that people and AI agents can both follow. Read it to rebuild your own markup, hand it to a developer as a brief, or let the SEO Agent audit what you have.
See it in action
“We have schema on every page but we never get rich results. Why?”
Every page re-declares its own Organization and Person block with slightly different values — three logo URLs, two spellings of the company name. Search engines are merging four competing entities rather than consolidating one. Emit the graph once in the root layout, reference it by @id from each page, and delete the per-page copies. Separately, eleven pages emit FAQPage with no visible FAQ on them, which is a penalty risk, not a missing feature.
From input to fix list
- 1
You describe the stack and content model
The framework and version, and where articles, FAQs and videos actually live.
- 2
Moonrank designs the entity graph
Organization, WebSite and Person as one set of nodes with stable @id values.
- 3
It derives the page schema
Article, breadcrumb, FAQ, video and HowTo emitted only where the content proves them.
- 4
It validates against the checklist
Zero errors, zero invented fields, and no duplicated entity nodes anywhere.
Everything it checks
Single-source module design
All JSON-LD from one place, so a change happens once rather than on every template.
@id entity graph
Organization, WebSite and Person emitted once and referenced, never repeated.
Content-derived schema
A derivation table mapping each content signal to the schema it justifies.
Conditional FAQ, HowTo and video
Emitted only where a real block exists — no markup that lies about the page.
Deduplication pass
Every re-declared Organization or Person block found and removed.
Validation checklist
Eight checks to run until Rich Results and the schema.org validator are both clean.
When to reach for it
- Fix a site that has schema everywhere and rich results nowhere.
- Give answer engines one consolidated entity to ground on.
- Remove FAQ and HowTo markup that no visible content supports.
- Brief a developer with a derivation table instead of a wish list.
- Rebuild markup after years of per-page copy-paste.