Content Reviewer — Claude AI Skill for SEO Content Quality Control

Without Content Review

With Content Review

Vague feedback with no score — you don’t know if the article is 60% or 90% of the way there

Four-pillar scorecard out of 100 — SEO, search intent, structure, and factual accuracy each scored /25 with a justified rating

No SERP check — Claude doesn’t know what competitors are covering before deciding what’s missing from your article

Live competitor benchmarking before writing a single word of review — top 3–5 organic results analysed for word count, headings, topics, and structured elements

No keyword density check — you have no idea if the primary keyword is under or over-used relative to the target range

Keyword density calculated precisely — flags if below 0.4% (underoptimised) or above 2% (stuffing risk), checks placement in H1, first 100 words, H2s, meta title, and meta description

No publish verdict — you’re left deciding on your own whether the article is ready, needs work, or should be sent back

Clear verdict every time — Pass (80–100), Revise (60–79), or Reject (below 60), with the top 3 priority fixes listed before the detailed review

Different review quality every time — no consistent standard between articles, writers, or reviewers

Identical workflow every review — same questions, same pillar order, same scoring philosophy, same report format regardless of article or writer

No deliverable — feedback exists only in the chat, with nothing to send back to the writer or keep on file

Downloadable .docx report generated automatically — professionally formatted with cover block, scorecard, verdict banner, pillar sections, and fix instructions, ready to send to the writer

The skill collects everything it needs before touching the article: the .docx upload, primary keyword, secondary keywords, website URL, whether it’s a new or edited piece, target audience, and funnel stage. Questions appear in the same fixed order every time. No assumptions — no skipping steps. If reviewing an edited article, it also collects the original version for diff comparison.

Intake questions (fixed sequence):
  Q1 · Upload the article (.docx)
  Q2 · Primary target keyword
  Q3 · Secondary keywords (comma-separated)
  Q4 · Website URL
  Q5 · New article or edited version?
  Q6 · Target audience + funnel stage (TOFU / MOFU / BOFU)

Searches the primary keyword and analyses the top 3–5 organic results (excluding ads, featured snippets, and YouTube). Records approximate word count, heading structure, topic coverage, and structured elements for each competitor. Also fetches the provided website URL to identify tone, internal linking opportunities, and meta tag patterns.

Competitor benchmark for: [primary keyword]
Average word count:         1,840 words
This article:               1,220 words  (-620 vs avg)
Topic gaps identified:      3 missing sections
Heading patterns top rankers use: FAQ section, comparison table, step-by-step numbered lists
Structured elements common in top results: tables (4/5 competitors), FAQ (5/5), numbered steps (3/5)

Each pillar produces a score, a one-sentence verdict, a list of what’s working, a list of issues found with exact line references where possible, and numbered fix instructions that are specific enough for a writer to act on without further clarification. Scoring is calibrated — 18–20 for solid fundamentals, 22–25 for exceptional execution, below 15 only for genuinely missing core elements.

After all four pillars, the skill compiles the total score, applies the verdict threshold (Pass 80+, Revise 60–79, Reject below 60), lists the top 3 must-fix items, then generates a downloadable Word document containing the full report — formatted for Google Docs compatibility with cover block, scorecard table, verdict banner, and pillar sections.

When reviewing a revised article, the skill compares it structurally against the original — identifying what was added, removed, or changed, confirming improvements, flagging anything still outstanding from the first review, and catching regressions the edit introduced.