arwp-search-maturity

ARWP Search Maturity

Use this specialist when the Growth Loop needs a reference cohort, not another generic checklist.

Use docs/ANTI-PATTERNS.md to review fabricated maturity, weak evidence and measurement errors; preserve unknown states and legitimate countercases. Record findings with templates/growth/anti-pattern-review.md.

Goal

Turn volatile Search/AI observations into an explainable comparison:

intent → timestamped visible references → observable evidence vector → cohort pattern → target gap → experiment → verify → measure

The output is a hypothesis generator and prioritization input. It is not a list of secret ranking factors.

Sources of truth

Read:

• docs/SEARCH-MATURITY-BENCHMARK.md;
• docs/INTERNAL-DISCOVERY-EVIDENCE.md when reviewing internalTopicalGraph;
• docs/INTENT-OWNERSHIP.md when owner query/grounding evidence may imply a new or competing canonical page;
• schema/search-maturity-corpus.schema.json;
• the relevant reviewed corpus under benchmarks/search-maturity/;
• current primary platform guidance;
• the target site's actual content, entity, Search Surface and owner evidence where available.

Workflow

1. Define the intent family first. Do not benchmark unrelated high-authority sites merely because they rank for something.
2. Resolve intent ownership before creating URLs. When owner Search/AI evidence exists, map the intent family to a reviewed disposition and zero/one/multiple canonical owners with arwp-intent-ownership. Strengthen an existing owner first; query or fan-out variation is not a page-creation instruction. Deliberately decline query mismatches instead of turning them into owner gaps.
3. Capture the visibility observation exactly. Preserve provider/surface, query, locale/device when known and observation time. Store a numeric rank only when explicit rankEvidence independently supports that exact rank.
4. Separate ownership. Independent references can inform the external cohort. ARWP-owned/project-reference sites remain implementation/dogfood evidence.
5. Review only observable dimensions. Use the 16 Search Maturity dimensions. Omit anything not actually checked; omission is unknown, never failure.
6. Classify evidence. Keep documented-platform, observed-correlation, experiment and unknown distinct.
7. Compile the cohort vector. Prefer repeated patterns and state distributions over a universal score.
8. Compare the target. A target gap is interesting only when the reference pattern is sufficiently repeated and the target evidence is real.
9. Perform applicability review. Ask whether closing the gap improves the actual user/evidence artifact. Reject cosmetic copying and cargo cult.
10. Route accepted work through existing ARWP machinery. Growth hypothesis → Adaptive Upgrade → safe transformation/manual edit → verification receipt.
11. Probe owner production surfaces after source changes. Capture the bounded HTTP/robots/noindex/canonical observation in a derived ledger. A successful probe means evidence was collected, not that production is ready.
12. Gate served intent owners separately. Run the deterministic ownership gate over the observed ledger. A served family must have exactly one currently indexable owner; preserve the proof artifact on failure. Reviewed decline families are intentionally excluded.
13. Verify source→production parity separately. Technical owner eligibility does not identify the deployed commit. Use deployment/change receipts when that distinction matters.
14. Measure external outcomes separately. Search visibility, AI retrieval/citation, referral traffic and business outcomes are different signals.
15. Preserve negative results. A change with no effect is useful evidence and should influence future priors.