seo-content-writer

Based on O-CMO Blog Writing & AI Copywriting Framework

OVERVIEW

This skill guides Claude through a complete AI-assisted workflow for creating long-form SEO content (2,000+ words). It integrates prompt engineering frameworks, E-E-A-T quality standards, JTBD-based content strategy, and the O-CMO brand writing framework.

Core philosophy:

• Every article starts with a Job-to-Be-Done (JTBD) — not a keyword
• People-first content that demonstrates first-hand experience and expertise
• Treat AI like a smart intern: smaller content units → more control → better quality
• Build iteratively: concept → outline → section → full article

PHASE 0: STRATEGY FOUNDATION

0.1 — Define the JTBD Before Anything Else

Every article — new or optimized — starts with a Job-to-Be-Done statement:

"When [situation], I want to [motivation], so I can [desired outcome]."

Legend:

• When [situation] = the context the reader is in → "When I'm comparing vendors…"
• I want to [motivation] = what they're looking for → "…I want to understand how each one delivers…"
• So I can [desired outcome] = the real reason behind the search → "…so I can avoid delays and explain my choice to the team."

Examples:

• When I realize our current marketing automation platform isn't working anymore, I want to understand what's involved in switching tools, so I can avoid data loss, downtime, and adoption issues.
• When we're planning the next lifecycle marketing phase, I want to evaluate whether gamification would improve retention, so I can present a clear case to the growth team.

Rule: One JTBD per article. Don't try to serve every ICP in one piece.

0.2 — E-E-A-T Alignment Check

Before writing, confirm the content plan passes Google's E-E-A-T framework (Experience, Expertise, Authoritativeness, Trustworthiness):

The three Google questions to ask about every article:

• Who created it? → Is authorship clear? Does it have a byline?
• How was it created? → Is the process transparent (including AI use)?
• Why was it created? → Is it primarily for readers, not for search rankings?

Content quality self-check:

• Does it provide original information, reporting, or analysis?
• Does it go beyond the obvious?
• Would you bookmark, share, or recommend this?
• Does it leave the reader feeling they've learned enough to act?
• Is every major claim supported by a primary source?

0.3 — Prompt Engineering Setup

Choose the right framework for each task:

C.O.A.S.T. — For all prompts

• C – Context: what you're doing and for whom
• O – Objective: exactly what you want
• A – Audience: who will read it
• S – Style / Voice: tone, paste voice guardrails
• T – Task details: word count, structure, required data

C.O.R.E. — AI task setup

• C – Context; O – Objective; R – Requirements; E – End use

C.R.E.A.T.E. — Content generation

• C – Command; R – Role (e.g., "expert SaaS copywriter"); E – Examples; A – Audience; T – Tone; E – Extras (word count, CTA style)

R.I.S.E.N. — Research

• R – Relevant data; I – Interesting/contrarian viewpoints; S – Sources; E – Engagement triggers; N – Narratives/emerging trends

Prompt structure rules:

• Use XML tags or ## headings to separate sections of long prompts
• Use [PLACEHOLDER] in CAPS for reusable variables
• One minimum content unit per prompt — never ask for the full article at once
• Ask for 3 options before committing: concept → approve → expand
• Use "step by step" when AI reasoning is unclear
• One example → AI copies structure. Three+ examples → AI extracts patterns

PHASE 1: RESEARCH & STRATEGY (30 min)

Step 1.1 — Voice DNA Extraction (one-time setup)

Upload 3–5 of your best articles and prompt:

Analyze these writing samples and create my "Voice DNA":
1. Tone descriptors (8–10 specific adjectives)
2. Sentence structure patterns (average length, variety)
3. Vocabulary level and word choices
4. Paragraph rhythm and flow
5. Common transitional phrases
6. How I handle examples and analogies
7. My approach to hooks and conclusions
Output a detailed style guide I can paste into every future prompt.

Step 1.2 — Voice Guardrails Template (save and reuse)

VOICE GUARDRAILS

DO:
- [Your tone requirements]
- [Sentence structure preferences]
- [Vocabulary style]

DON'T:
- [Words/phrases to avoid: "just", "delve", "enhance", "game-changing"]
- [Structural patterns to avoid: em-dash overuse, passive voice, generic corporate openers]
- [Formatting to avoid: lists of "Bold term: description" pairs — very AI-looking]

DO NOT START WITH:
- Proverbs, sayings, or "everyone knows that…"
- Generalizations: "every marketer has…", "all founders…"
- Direct questions to the reader: "Have you ever noticed…"
- Quotes from classics or famous people

VOICE EXAMPLES:
Instead of: "[Bad example]"
Write: "[Good example]"

Session warm-up prompt:

Before writing, review my voice guardrails: [paste guardrails]
After writing, self-check:
1. Does this sound like me?
2. Are sentence lengths varied and consistent with my style?
3. Have I avoided my "don't" list?
If not, revise before showing me output.

Step 1.3 — Topic Research (5 min) — Perplexity

Research [TOPIC] and provide:
1. 5 recent statistics or data points (2024–2025)
2. 3 trending subtopics people are currently discussing
3. 2 contrarian or surprising viewpoints
4. Top pain points your audience faces with this topic
5. 3 credible sources for each point
Format as: Fact | Source | Why it matters

Warning: Always verify sources manually. AI often references outdated or renamed products.

Step 1.4 — Competitive Analysis (5 min) — ChatGPT

Find 5 popular articles about [TOPIC] published in the last 6 months.
For each article, analyze:
- Main angle/approach taken
- Key points covered
- What's missing or could be improved
- Word count and structure
- Engagement elements (hooks, examples, CTAs)
I want to create something better and more comprehensive.

Step 1.5 — GSC Integration (for optimization tasks)

When optimizing an existing article:

1. Open Google Search Console → Performance → Search results
2. Filter by the live URL
3. Check Search Queries for: what's already ranking, what gets impressions without clicks, what's missing

Use findings to:

• Prioritize queries already getting impressions
• Integrate them naturally into subheadings and intro text
• Fill gaps by adding sections that address uncovered queries
• Improve relevance by matching the intent (informational, comparative, how-to)

Step 1.6 — Audience Research (5 min)

I'm writing about [TOPIC] for [TARGET AUDIENCE].
Create a detailed reader profile:
- Current knowledge level about this topic
- Biggest frustrations/pain points
- Questions they ask but can't find answers to
- Preferred language/tone style
- What would make them share this content

Step 1.7 — Angle Selection (5 min)

Based on this research: [PASTE ALL RESEARCH]
Generate 10 unique angles for a [WORD COUNT]-word article about [TOPIC]:
1. The contrarian take
2. Th