Hub And Spoke Discovery And Recovery

Hub & Spoke Discovery & Recovery

Overview

Hub & Spoke is the content architecture — one hub owns a topic cluster; spokes answer sub-intents and link up. Cornerstone is the quality bar for those pages (depth, intent, linking, fan-out, citability). This playbook is the Discovery & Recovery path for sites where reality does not match intent.

It evolves the discovery-first audit pattern: triangulate first-party signals (GSC impressions + GA4 business value) with DataForSEO organic architecture and page inventory, then audit and plan recovery without assuming which pages are hubs today.

Not this playbook: Top-down health check when cornerstones are already known → [Cornerstone Content Audit](../cornerstone-content-audit/SKILL.md).

Packaging: Tier B playbook per [write-a-skill](../../../utilities/write-a-skill/SKILL.md) — orchestrates children; triangulation rules in REFERENCE.md.

Architecture

| Phase | Step | Skill / composite | Path | |-------|------|-------------------|------| | 1a | Performance matrix | Analytics and Search Console Performance Audit | ../../skills/analytics-and-searchconsole-performance-audit/SKILL.md | | 1b | Catalog + organic signals | Site Content Catalog + DataForSEO domain keywords | ../../skills/site-content-catalog/SKILL.md | | 1c | Triangulation merge | Playbook merge (this doc + REFERENCE.md) | — | | 3b | Fan-out | Fan-Out Coverage Analysis | ../../skills/fan-out-coverage-analysis/SKILL.md | | 3c | Information gain | Information Gain Evaluator | ../../skills/information-gain-evaluator/SKILL.md | | 4a | Cannibalization | SEO Cannibalization Audit | ../../skills/seo-cannibalization-audit/SKILL.md | | 5h | Crawler policy | robots.txt Audit | ../../skills/robots-txt-audit/SKILL.md |

Read each linked SKILL.md before calling it. Follow its workflow exactly. Do not re-implement child scoring inline.

Input policy

Collect from the user:

1. Domain — required
2. Scope — discovery-only (Phases 1–2) or full audit (Phases 1–6)
3. First-party data — at least one of:

• GSC page×query export (CSV) or GSC MCP when available
• GA4 landing-page export (CSV) or GA4 MCP when available

1. Conversion events — 1–3 GA4 key event names when GA4 is provided and analysis mode is discovery_plus_conversions
2. Analysis mode — discovery_only or discovery_plus_conversions (ask once; see analytics skill Step 0.5)

Optional:

• intended_hubs[] — pages the business wants to rank (commercial hub + service LPs)
• protected_pages[] — never auto-recommend delete
• Location/language for DataForSEO (default US English)

Degraded mode: When GSC/GA4 is missing, Phase 1a runs partial matrix with limitations[]. When both first-party sources are missing, skip Phase 1a and run Phase 1b only (DataForSEO-only discovery — existing behavior preserved). Document all gaps in report limitations[].

If domain is missing, ask once. Never invent URLs, metrics, or recommendations before tool results exist.

First response rules

On the first user-facing response:

1. Name scope (discovery-only | full audit).
2. List which Phase 1 inputs are present vs missing.
3. Name child skills for that scope.
4. Preview deliverable shape: Performance Matrix → Topic Cluster Map → human checkpoint.
5. State Phase 2 is advisory — no auto-delete or auto-redirect execution.

Use AskQuestion for Phase 2 approval when available; otherwise wait for explicit confirmation.

---

Phase 1: Triangulated discovery

Phase 1a — Performance matrix (first-party)

Call: [Analytics and Search Console Performance Audit](../../skills/analytics-and-searchconsole-performance-audit/SKILL.md)

Pass:

• domain
• GSC page×query export (or MCP equivalent)
• GA4 landing-page export when available
• conversion_events when discovery_plus_conversions
• intended_hubs[], protected_pages[] when provided

Receive handoff JSON v1.0 (REFERENCE.md in analytics skill).

Completion criterion: Valid handoff with pages[], summary.quadrant_counts, and limitations[] reflecting data mode (full | gsc_only | ga4_only).

If skipped (no first-party data): Set matrix_handoff: null and note in limitations[] — proceed to Phase 1b.

---

Phase 1b — Catalog + organic architecture (DataForSEO)

Two calls — inventory first, then domain keyword universe.

1b-i — Site content catalog

Call: [Site Content Catalog](../../skills/site-content-catalog/SKILL.md)

Pass sitemap URL or domain. Prefer sitemap_enriched when DataForSEO MCP is available.

Receive handoff JSON v1.0 (REFERENCE.md in catalog skill).

If catalog returns >200 pages, scope down with the user before deep enrichment.

1b-ii — Organic hub scoring

Pull domain-wide ranked keywords and derive de facto hub structure. Follow hub-scoring heuristics in REFERENCE.md § Organic hub scoring.

Call: dataforseo_labs_google_keywords_for_site

• target: domain
• limit: 1000, order_by: search volume desc
• Filter: rank position ≤ 100

For each URL: aggregate keyword count, total ETV, average rank, topic cluster assignment. Compute hub_score and classify: de_facto_pillar | strong_spoke | weak_spoke | orphan.

Completion criterion: Topic cluster map with de facto cornerstone per cluster + hub scores for top organic pages.

---

Phase 1c — Triangulation merge

Join Phase 1a matrix pages, Phase 1b catalog pages, and Phase 1b hub scores on canonical normalized URL (REFERENCE.md § Join keys).

Per URL emit triangulation row:

| Field | Source | |-------|--------| | matrix_quadrant | Phase 1a pages[].quadrant (null if 1a skipped) | | hub_score | Phase 1b organic scoring | | hub_type | Phase 1a hub_type_hint or catalog page_type | | intended_match | URL in intended_hubs[] | | triangulated_role | Merged role per REFERENCE.md § Role assignment | | signal_conflicts[] | Explicit mismatches (e.g. high ETV + low GSC clicks) |

Triangulated role enum: intentional_hub · accidental_hub · hidden_gem · strong_spoke · orphan · unicorn · protected_review

Completion criterion: Every URL in the union of catalog + matrix + top-50 organic pages has a triangulation row OR explicit out_of_scope note in limitations[].

Output: Unified cluster map ready for Phase 2 human review.

---

Phase 2: Confirm & decide

Human checkpoint

Present in this order:

1. Performance Matrix (when Phase 1a ran) — quadrant summary + top URLs per quadrant
2. Topic cluster map — de facto pillars, hub scores, keyword counts
3. Intended vs de facto — gap table when intended_hubs[] provided
4. Triangulation highlights — accidental hubs, hidden gems, signal conflicts

## Performance Matrix — {domain}

| Quadrant | Count | Top URLs |
|----------|-------|----------|
| Unicorn | {n} | ... |
| Accidental hub | {n} | ... |
| Hidden gem | {n} | ... |
| Prune candidate | {n} | ... |

## Topic clusters

| Cluster | De facto hub | Hub score | Matrix quadrant | Triangulated role |
|---------|--------------|-----------|-----------------|-------------------|
| ...     | ...          | ...       | ...             | ...               |

## Intended vs de facto

| Intended hub | De facto match? | Triangulated role | Recommended review |
|--------------|-----------------|-------------------|--------------------|
| ...          | ...             | ...               | ...                |

**Signal conflicts**
- {conflict 1 — e.g. high DataForSEO ETV but low GSC clicks on /pricing/}
- {conflict 2 — e.g. high GSC impressions, zero GA4 conversions on /blog/...}

Which pages should I audit in depth?
A) De facto pillars (what ranks today)
B) Intended hubs (what you're building toward)
C) Both — compare clusters
D) Let me pick specific URLs

Do not proceed to Phase 3 until the user confirms. Matrix and triangulation are advisory — not auto-execution.

Output: Audit target list for Phases 3–5.

---

Phase 3: Health audit (per target)

Run for each page in the audit target list (parallel OK).

3a: Traditional SEO scorecard

| Dimension | Source | |-----------|--------| | Rankings | Phase 1b keyword data or dataforseo_labs_google_ranked_keywords | | Backlinks | backlinks_summary | | Content | on_page_content_parsing or WebFetch | | SERP features | serp_organic_live_advanced |

3b: Fan-out coverage

Call: [Fan-Out Coverage Analysis](../../skills/fan-out-coverage-analysis/SKILL.md)

Pass: keyword, domain, anchor_url.

3c: Information gain

Call: [Information Gain Evaluator](../../skills/information-gain-evaluator/SKILL.md)

Pass: target_url, primary_keyword.

3d: AI citability

1. ai_opt_llm_ment_search — domain mentions for primary keyword
2. ai_opt_llm_ment_top_pages — pages AI cites
3. Structure check (WebFetch): H2/H3 hierarchy, tables, FAQ, extractable passages
4. Lightweight robots check if already fetched — full policy in Phase 5h

3e: Scorecard template

Same format as [Cornerstone Content Audit](../cornerstone-content-audit/SKILL.md) Phase 3e.

---

Phase 4: Cannibalization & pruning

4a: Cannibalization check

Call: [SEO Cannibalization Audit](../../skills/seo-cannibalization-audit/SKILL.md)

Focus on de facto pillars competing with intended hubs and blog posts outranking landing pages.

4b: Pruning candidates

From triangulation: orphan, prune_candidate (matrix), weak_spoke. Classify: Consolidate | Improve | Deindex | Delete.

Never recommend delete from matrix quadrant alone — require redirect target and human approval.

4c: Architecture conflicts

• Accidental pillar blocking intended hub
• Topic cluster with no pillar
• Pillar without spokes
• Spoke without pillar

---

Phase 5: Enhancement & recovery plan

Map triangulation roles to existing recovery templates (no new phases):

| Triangulation / matrix signal | Recovery template | Action | |------------------------------|-------------------|--------| | Unicorn | 5a Content improvements | Protect + replicate pattern on adjacent clusters | | Hidden gem | 5b Internal linking + 5e Fan-out gap content | Spoke creation, inbound link map toward hub | | Accidental hub | 5f Transition decision tree | Spoke extraction, hub retrofit, or demote-to-spoke | | Prune candidate | 5g Cannibalization + 5h Pruning | Redirect map only — no delete queue |

5f: Transition decision tree (accidental hub)

• De facto pillar is the right page → promote officially, strengthen
• De facto pillar should be replaced → content migration, link rerouting, eventual 301 when intended hub is strong enough
• No pillar for cluster → build intended hub, create spokes, establish links
• Accidental pillar has value → keep as spoke or consolidate into intended hub

5h: robots.txt and AI crawler policy

Call: [robots.txt Audit](../../skills/robots-txt-audit/SKILL.md)

Pass: domain, key_pages (audit targets), crawl_policy default max_discovery.

Prioritization

• Quick wins: internal links to hidden gems, title/meta on accidental hubs, schema, robots fixes
• Medium effort: content improvements, information gain, begin intended hub builds, clear prune/redirect candidates
• Strategic: full architecture transition, new cluster content, major consolidations

---

Phase 6: Report

# Hub & Spoke Discovery & Recovery — {domain} — {date}

## Executive Summary
- {X} topic clusters, {Y} de facto hubs, {Z} intended-hub gaps
- Performance matrix: {quadrant counts when 1a ran}
- {N} quick wins, {M} strategic projects
- Data limitations: {from limitations[]}

## Performance Matrix
{From Phase 1a — before topic cluster map}
{Quadrant table + top 10 per quadrant}

## Topic Cluster Map
{From Phase 1b + 1c triangulation}

## Intended vs De Facto Architecture
{Gap table + triangulated roles}

## Signal Conflicts
{Explicit conflicts from Phase 1c}

## Cornerstone Scorecards
{Per-page from Phase 3}

## Cannibalization & Architecture Conflicts
{From Phase 4}

## Enhancement & Recovery Plan
### Quick Wins / Medium Effort / Strategic Projects
{Mapped from Phase 5 recovery table}

## New Content Opportunities
| Topic | Target Keyword | Volume | Parent Hub | Format |

## Pruning & Redirect Candidates
| URL | Triangulated role | Recommended action | Target URL | Reason |

## robots.txt & AI Crawler Policy
{From Phase 5h}

## Appendix: Triangulation table
| URL | Matrix quadrant | Hub score | Triangulated role | Flags |

---

When to use which playbook

| Situation | Playbook | |-----------|----------| | Known cornerstones, health check only | [Cornerstone Content Audit](../cornerstone-content-audit/SKILL.md) | | Messy site, accidental hubs, new client | This playbook | | Have GSC/GA4 exports, need triage fast | This playbook Phases 1a–2 only | | Single keyword / fan-out gap | [Fan-Out Coverage Analysis](../../skills/fan-out-coverage-analysis/SKILL.md) only |

---

DataForSEO tools used

dataforseo_labs_google_keywords_for_site, dataforseo_labs_google_ranked_keywords, dataforseo_labs_google_related_keywords, dataforseo_labs_google_keyword_suggestions, dataforseo_labs_google_keyword_ideas, backlinks_summary, serp_organic_live_advanced, on_page_content_parsing, ai_opt_llm_ment_search, ai_opt_llm_ment_top_pages

Requirements: REQUIREMENTS.md · Triangulation SSOT: REFERENCE.md · Future work: FUTURE-WORK.md

Done definition

• [ ] Domain and scope confirmed (discovery-only or full audit)
• [ ] Phase 1 inputs inventoried; degraded paths documented in limitations[]
• [ ] Phase 1a handoff JSON emitted when GSC and/or GA4 provided (analytics skill)
• [ ] Phase 1b catalog + organic hub scores complete (or limitations recorded)
• [ ] Phase 1c triangulation table produced with triangulated_role per URL
• [ ] Phase 2 human checkpoint presented — user approved targets before Phase 3+ (when scope includes Phase 3+)
• [ ] Report follows template order: Performance Matrix → Topic Cluster Map → Intended vs De Facto
• [ ] No auto-delete or auto-redirect execution — recommendations only, with target URLs where applicable
• [ ] Full audit scope: fan-out, information gain, cannibalization, and robots.txt phases called per architecture table

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: Roots-and-Fruit
• Source: Roots-and-Fruit/skills
• License: MIT

Install and usage instructions live in the source repository linked above.