# SEO Audit > Use when evaluating search performance or diagnosing ranking issues. - Skill: `majesticlabs-dev/seo-audit` (Agent Skill, multi-file: 3 files) - Install (CLI): `npx skillmds@latest add majesticlabs-dev/seo-audit` - Raw SKILL.md: https://api.skillmd.com/api/skills/majesticlabs-dev/seo-audit/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Marketing & Growth - Author: majesticlabs-dev (https://skillmd.com/u/majesticlabs-dev) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/majesticlabs-dev/seo-audit --- # SEO Audit Skill A comprehensive SEO audit methodology that evaluates content for both traditional search engine optimization and modern AI/LLM visibility (GEO). Informed by Google's multi-stage ranking pipeline. ## When to Use - Auditing existing content for SEO performance - Reviewing new content before publication - Identifying optimization opportunities - Assessing AI citation readiness - Evaluating E-E-A-T signals - Checking technical SEO elements - Diagnosing why content isn't ranking - Auditing new domains (sandbox considerations) ## Google's Ranking Pipeline Content passes through sequential evaluation gates. Understanding this pipeline informs what to audit and why. ``` Mustang → Topicality (T*) → NavBoost → Twiddlers ↓ ↓ ↓ ↓ Initial Query-match User clicks Final scoring relevance (13 months) adjustments ``` **Key Implications:** - Content must pass Mustang's quality gate before relevance even matters - User behavior (NavBoost) accumulates over 13 months, patience required - Two foundational pillars: Q* (site quality) and P* (popularity via Chrome data) - Page potential is capped by domain's `siteAuthority` score See `assets/google-ranking-signals.yaml` for complete signal reference. ## Audit Framework ### Phase 0: Domain & History Context Before auditing content, assess domain-level factors that cap page potential: **Domain Authority Check:** - [ ] Established domain (>6 months) or new domain sandbox? - [ ] Site-wide quality signals (clutterScore, ad density) - [ ] Homepage authority inheritance for new pages - [ ] Brand entity recognition (`queriesForWhichOfficial`) **URL History Assessment:** - [ ] Is this an established URL with history? - [ ] URL changes reset accumulated trust (Google tracks last 20 versions) - [ ] Evergreen URL structure (no dates unless news content) - [ ] URL permanence strategy in place **Sandbox Awareness (New Domains/Pages):** | Factor | Status | Impact | |--------|--------|--------| | hostAge | New (<6mo) / Established | Domain-level sandbox | | documentHistory | New URL / Has history | Page-level demotion | | Graduation signals | Social, backlinks, engagement | Exit sandbox | **If new domain:** Set realistic expectations. Dual sandbox (host + document) requires consistent quality publishing, backlinks, and positive engagement over months. ### Phase 1: Technical SEO Check Evaluate foundational technical elements: **Page-Level Technical:** - [ ] URL structure (clean, descriptive, <60 chars, evergreen) - [ ] Title tag (50-60 chars, keyword in first 30) → affects `titlematchScore` - [ ] Meta description (150-160 chars, compelling CTA) - [ ] H1 tag (single, matches topic) - [ ] Header hierarchy (logical H1→H2→H3) → larger text weighted higher - [ ] Image alt text (descriptive, keyword-relevant) - [ ] Internal links (contextual, relevant) - [ ] External links (authoritative sources) **Site-Level Technical:** - [ ] HTTPS enabled - [ ] Mobile-friendly - [ ] Page speed (<1.8s mobile) - [ ] Schema markup present - [ ] XML sitemap inclusion - [ ] robots.txt accessibility **Entity & Trust Signals:** - [ ] Verifiable ownership/authorship (disconnected entity problem) - [ ] Contact information accessible - [ ] Organization schema linking to Knowledge Graph ### Phase 2: Content Quality Assessment Evaluate content depth and value. Google's `contentEffort` signal measures ML-assessed effort invested. **Content Completeness:** | Factor | Check | Score | Signal | |--------|-------|-------|--------| | Topic coverage | Comprehensive vs. shallow | /10 | contentEffort | | Unique value | Original insights vs. rehash | /10 | OriginalContentScore | | Accuracy | Factual, verifiable | /10 | Trust signals | | Freshness | Current data/sources | /10 | semanticDate | | Depth | Expert-level detail | /10 | contentEffort | **Token Truncation Awareness:** Google uses token limits. Long documents may be truncated. Optimize for this: - [ ] Critical information in first 20% of content (inverted pyramid) - [ ] Key facts lead each section - [ ] Paragraphs <120 words - [ ] Front-load value, don't bury it **Readability:** - Reading level (target: Grade 8-10) - Paragraph length (2-3 sentences ideal) - Sentence variety - Scannable formatting (bullets, headers) - Visual aids (images, tables, diagrams) **Freshness Audit:** Google evaluates freshness through three distinct methods: - [ ] Explicit byline date (visible, accurate) - [ ] URL date signals (avoid dates in evergreen URLs) - [ ] Semantic freshness (are facts/sources current, not just timestamp?) ⚠️ **Freshness spam detection:** Changing timestamps without updating substance is detected via `semanticDate` comparison. ### Phase 3: Keyword & Semantic Analysis Assess keyword optimization. Affects Topicality (T*) stage scoring. **Primary Keyword:** - Present in title tag (first 30 chars ideal) - In H1 and early H2s - Natural density (0.5-1.5%) - In meta description - In first 100 words **Semantic Coverage:** - LSI keywords present - Related entities covered (`webrefEntities` signal) - Topic cluster alignment - Question variations addressed - Search intent match **Entity Optimization:** - [ ] Content associates with Knowledge Graph entities - [ ] Clear entity relationships established - [ ] Supports topical authority building ### Phase 4: E-E-A-T Evaluation Assess Experience, Expertise, Authority, Trust signals. These map directly to technical signals. **Experience Signals:** - [ ] First-hand experience indicated - [ ] Case studies/examples included - [ ] Original data/research - [ ] Process documentation **Expertise Signals:** (→ `contentEffort`, `OriginalContentScore`) - [ ] Author credentials visible - [ ] Technical accuracy - [ ] Comprehensive coverage - [ ] Expert quotes/interviews **Authority Signals:** (→ `siteAuthority`, `author` attribute) - [ ] Authoritative external citations - [ ] Industry recognition - [ ] Brand mentions - [ ] Published research **Trust Signals:** - [ ] Contact information (disconnected entity fix) - [ ] Privacy policy - [ ] Editorial guidelines - [ ] Reviews/testimonials - [ ] Security indicators **Disconnected Entity Check:** Sites without verifiable ownership, author info, and contact transparency trigger algorithmic distrust, even with high content quality. ### Phase 5: AI/GEO Readiness Assess content for LLM visibility: **Extractability:** - [ ] TL;DR or summary present (first 100 words) - [ ] Paragraphs <120 words (token limit friendly) - [ ] Clear topic sentences (lead with facts) - [ ] Bullet lists for features - [ ] Tables for comparisons - [ ] FAQ section present **Fact-Density:** - [ ] Statistics with sources - [ ] Specific numbers/dates - [ ] Verifiable data points - [ ] Expert quotes with attribution **Structure for AI:** - [ ] Question-based headers - [ ] Direct answers near top (inverted pyramid) - [ ] Logical information hierarchy - [ ] Schema markup (FAQPage, HowTo) ### Phase 6: User Behavior Signals Google's NavBoost uses 13 months of aggregated click behavior. Audit for "good clicks" potential. **Click Quality Indicators:** - [ ] Title/meta accurately represent content (no clickbait) - [ ] Content matches search intent (dwell time) - [ ] Clear value proposition visible above fold - [ ] Low bounce risk (content delivers on promise) **Engagement Factors:** | Signal | Status | Improvement | |--------|--------|-------------| | Expected dwell time | Low/Med/High | Match intent better | | Bounce risk | Low/Med/High | Clearer value prop | | Return visit potential | Low/Med/High | Brand building | ## Detailed Reference Load [detailed-reference.md](references/detailed-reference.md) for extended examples, templates, and advanced patterns.