Page Audit Procedure

Home / Content Procedures / Page Audit Procedure

Page Audit Procedure

Last Edited: 2026-06-29 01:32

Description

Detailed procedure for Editors to audit page drafts using standard prompt frameworks, evaluating SEO metrics, E-E-A-T signals, and AI pattern risks.

Goal

Ensure all content drafts undergo rigorous review before designer handoff to maintain high entity density, prevent AI footprint risks, and optimize local ranking strength.

Workflow

Step 1: Pre-Audit Check

Open the freelancer’s submitted Google Doc.
Verify the document is formatted correctly and variables are intact.
Copy the raw text and run it through ZeroGPT or equivalent tools to record AI percentage baselines.

Step 2: Run the Page Audit Prompt

Paste the draft text alongside the following standardized Page Audit Prompt into the evaluation model:

Please audit the provided page draft using the framework below.

### Audit Scope
1. SEO effectiveness (including Entity/Semantic density)
2. EEAT / trust
3. Content quality & Local relevance
4. Conversion strength & Likelihood to rank
5. Template / AI pattern risk
6. Comprehensiveness & AI citation/retrieval readiness

### Core Rules
1. Chain of Verification (Required)
   Use CoV before scoring. Verify all critique against the draft.
   - Negative Space Check: Before listing a section as "Missing," verify the topic isn't simply covered under a different heading or integrated into another paragraph.
   - Entity/Semantic Check: Evaluate for industry-specific technical terms (e.g., SEER ratings, specific parts, refrigerants) that signal topical authority.
2. Protected Elements (Do NOT Penalize)
   - Placeholder tags (e.g., {#SEOCITY1RAW}, {#COMPANYNAME}, {#COMPANYABBV})
   - CTA/button text (e.g., “Read Our Story”, “See All Services”)
   - Close keyword variants
   - Business claims the user has told you to treat as verified.
   - Normal contractor-site overlap by itself
   - Reasonable persuasive language, as long as the page still stays mostly concrete.
3. Deduction Discipline (Critical)
   - If a score is below 10, every lost point must be clearly accounted for in the deductions. No vague scoring. No “No meaningful deductions” if points were actually lost. If an item is fixed, the score should go up the same amount you deducted.
   - Each issue is deducted once, in one category only.
   - Do not hide deductions.
   - Every meaningful issue MUST include: The exact sentence/phrase, a clear explanation of what’s wrong, and a clear How to fix this.
4. Sales / Reassurance Language Guardrail
   - Do not flag "Expert Reassurance" (e.g., "NATE-certified," "70 years experience").
   - Only flag "Empty Superlatives" (e.g., "We are the best, most amazing ever") if they are repeated excessively or replace specific proof.
   - Generic language is okay as long as it is not overwhelming or more than 20% of the copy (i.e., you can say we keep you comfortable or are trustworthy).
5. Local Relevance Guidance
   - For non-homepages, assess whether the page feels locally grounded, not just city-swapped.
   - Look for: Real local references, natural integration of city/service area language, climate/weather relevance, landmarks, local utilities, building patterns, or homeowner realities.
   - Do not force city-page style localization onto homepages and service pages. For homepages, judge local relevance based on service-area clarity, geographic targeting, and business grounding.
6. Template / AI Pattern Risk Guidance
   - Do not call something AI just because it covers standard industry topics.
   - Focus this section on real pattern risk, such as: Repetitive sentence construction, stock phrasing, empty reassurance, mechanical transitions, overly symmetrical copy, vague generalities, recycled heading formulas, and templated rhetorical patterns.
   - Do not penalize placeholder tags or normal service-page similarity by default.

Step 3: Parse and Record Audit Outputs

Confirm the generated audit report contains all required output fields:

Overall Score: X/10 (grading page quality, competitiveness, and ranking/conversion outlook).
Score Table:
- SEO Effectiveness (X/10)
- EEAT / Trust Signals (X/10)
- Content Quality & Clarity (X/10)
- Local Relevance (X/10)
- Conversion Strength (X/10)
- Likelihood to Rank (X/10)
- Template / AI Pattern Risk (X/10)
- Comprehensiveness (X/10)
- AI Citation / Retrieval Readiness (X/10)
Point Deductions: Listed with exact sentence, points deducted (-0.X), reason, and fix.
Must-Fix Issues: Numbered, citing exact sentences and providing rewrite examples.
What’s Working Well: Highlighting strong sections.
Missing Keywords: Classification table (Present | Partially Present | Missing).
Order / Flow Review: Section-by-section transition review.
SEO & Structural Opportunities: Headings, entities, schema suggestions.
Local Signals Review: Alignment with local climate or utility references.
Conversion & Trust Review: Direct feedback on CTAs.
AI Readiness Review: Identification of citation-ready sentences.
Flagged Syntax Issues: Line items showing exact phrasing issues and corrections.
Required Verification Line: Document must end with: "Chain of Verification was used before final scoring and recommendations."

QA / Troubleshooting

Superlative Flags: If a writer uses terms like "premier provider" or "top-notch service," verify if they have years of business history or ratings justifying the claim before deducting points under the Sales Language rule.
AI Rejection: If the AI Pattern Risk score drops below 7/10, mark the draft Rejected and return to Gate 3 in Content Process.