SEO Page Schema and Intent-driven Content

1) The problem: “one page per keyword” is not an SEO architecture

Imagine three closely related commercial queries and three separate landing pages. The titles differ, but two pages repeat almost the same body content; one includes Product properties that users cannot see on the page; and all three lead to the same generic CTA. The problem is not a shortage of keywords. The problem is that intent, page identity, visible content, canonical signals, claims and conversion path are managed separately.

2) COM-009 scope: a Page-Intent Schema, not the claim registry or full content QA

This lesson owns SEO Page-Intent Schema v1. COM-008 owns the Claim/Source Registry and COM-010 owns the Content QA Rubric. COM-009 does not guarantee Search ranking, a rich result, indexing or conversion, and no Google tool becomes MAKNOW release authority.

3) Mental model: Intent → Purpose → Canonical Identity → Visible Contract → Markup/Claims → Links/CTA → Approval → Measurement

Before choosing a slug or schema type, define the audience problem and the task the page must complete. Then establish a stable page_id, canonical_url and visible_content_contract. Markup, claims, internal-link role and CTA must derive from that same contract rather than being decided independently in separate tools.

4) FACT: structured data describes a page; it does not promise an outcome

Google describes structured data as a standardized format for explicit information about page content. Correct implementation can make a page eligible for supported Search features, but eligibility does not guarantee a rich result, ranking improvement or commercial outcome. In this lesson, markup_validation_state=PASS is technical evidence, not the page's final state.

5) FACT: do not invent invisible facts only in markup

Google guidance says structured data should describe the page it appears on and should not mark up information that users cannot see there. visible_content_contract and required_properties must therefore be comparable. If a material property exists only in markup, the safe state is REVISE_MARKUP_VISIBILITY, not READY.

6) FACT: canonicalization is deduplication, not business taxonomy

Google selects a representative URL among duplicate or highly similar pages. Site signals such as rel=canonical influence that selection, but Google may choose a different canonical. We record canonical identity for consistency and evidence, but Google canonicalization does not decide whether two MAKNOW audience intents deserve one page or two; that is a local content/product decision.

7) The 21-field contract: every page gets an auditable decision record

A defensible page plan contains much more than a URL and keywords: page_id, audience, problem_intent, query cluster, purpose, canonical, page type, visible contract, structured-data type/properties/validation, claim references, internal-link role, CTA/destination, indexability, owner/approval, measurement and evidence all live in one record.

8) Intent is not the same thing as a query cluster

A query cluster is a set of expressions; intent is a hypothesis about the user's actual task. Different queries can share one task and one strong destination. Similar-looking queries can also represent different tasks and justify separate pages. Decide with problem_intent + content_purpose + visible_content_contract, not keyword count.

9) Do not solve near-duplication with keyword-only pages

If a proposed page is essentially a synonym for an existing destination, repeats the same body and leads to the same CTA, a new URL lacks strong evidence. CONSOLIDATE_DUPLICATE_INTENT means route the query cluster and link strategy to the existing canonical destination, or prove a materially distinct purpose/audience. A canonical tag on a weak duplicate is not a substitute for clear information architecture.

10) Recheck claims when intent changes

If problem_intent moves from education to purchase or consultation while the old claim set and evidence are copied forward, the page may make promises that no longer fit the new decision context. Use HOLD_INTENT_CLAIM_DRIFT until claim references are rechecked against the new intent and visible content.

11) CTA and conversion destination continue the same promise

primary_cta should be the natural next step of the page purpose, and conversion_destination must actually support that action. If the page promises “request a demo” but the CTA leads to a generic newsletter, strong keywords and valid markup do not fix the mismatch; use REVISE_CTA_INTENT.

12) The Rich Results Test is a test, not an oracle

A passing validator can show that markup has no detected critical issue for the tested feature, but Google explicitly does not guarantee rich-result display. If a team converts validation into “the rich result will appear” or “rankings will improve,” use HOLD_RICH_RESULT_OVERCLAIM and narrow the claim.

13) Missing evidence is not permission to infer

When evidence_ref, canonical purpose, approval or another critical dimension is missing, do not infer a positive decision from keyword volume, model confidence or competitors. HOLD_MISSING_EVIDENCE means the decision evidence is incomplete; that is different from proving that the page itself failed.

14) Major failure modes and typed states

• every critical dimension closed, visible/markup aligned, intent and CTA aligned → READY
• markup properties are unsupported by visible content → REVISE_MARKUP_VISIBILITY
• CTA or destination does not match intent → REVISE_CTA_INTENT
• new URL is a near-duplicate without a distinct task → CONSOLIDATE_DUPLICATE_INTENT
• decisive evidence is missing → HOLD_MISSING_EVIDENCE
• intent changed while old claims were retained → HOLD_INTENT_CLAIM_DRIFT
• validation is treated as a rich-result/ranking guarantee → HOLD_RICH_RESULT_OVERCLAIM
• canonical signals conflict with the page decision → HOLD_CANONICAL_CONFLICT

15) Lab and transfer: rebuild the plan from changed evidence

The offline lab contains eight fixtures and 48 deterministic assertions. In transfer, a new query cluster has stronger commercial intent, but its planned page duplicates an existing canonical destination and some proposed markup is not visible. The correct answer is not “make another page because there is another keyword.” Consolidate/reuse the existing destination or prove a genuinely different task, remove or surface unsupported markup, and recompute the state from the changed evidence.