Why keep some pages noindex?
Because a truthful preview is better than an indexable page with unsupported proof or thin content.
Methodology before scale
The MVP ships a structured input system first. Templates, metadata, schema, linking, and proof handling all assemble from typed records rather than loose page drafts.
01
Templates
Each family has a fixed section contract. That lets the pages feel consistent while still leaving room for different examples, proof states, and cluster links.
02
Internal linking
Linking is modeled as data. Each programmatic page carries an explicit link graph with a reason for each link.
- Outcome pages link to two use cases plus proof
- Use-case pages link back to an outcome and proof
- Proof pages link back to the buying surface and both use cases
03
Quality and index gates
The build checks metadata completeness, schema coverage, proof status, citation count, and unsupported promise language before deciding whether the page can publish.
- Pages can exist in preview while still defaulting to noindex
- Proof pages remain honest when evidence is partial
- Thin-page avoidance is enforced centrally rather than ad hoc
04
Questions
What is the source of truth for content?
Structured page briefs and evidence records. Narrative is rendered from those inputs rather than authored freehand as the database.