Claude AI prompts for SEO content writing work best when they are built on real competitor data, not guesswork. The difference between a generic AI-written article and one that ranks and gets cited in AI search is the quality of the input: the SERP research, the competitor heading structures, the content gaps, and the brief that ties them together. This guide gives you 8 tested prompts that cover the full content writing workflow, from competitor analysis to final draft, with real examples of what the output looks like.
Most "Claude SEO prompts" articles give you a list of 50 generic prompts that produce generic output. This is not that. Each prompt below is a specific step in a production workflow I use on client sites. They are sequenced in the order you should run them: research first, then brief, then draft, then quality check. Skip a step and the output quality drops.
Why Claude Is Better Than ChatGPT for SEO Content Writing
Claude outperforms ChatGPT for SEO content writing because of three structural advantages: a 200K token context window that can hold an entire SERP's worth of competitor content in a single prompt, more consistent instruction-following for multi-part content briefs, and less tendency to pad output with filler sentences. In head-to-head testing on content briefs with specific heading structures, word count targets, and tone guidelines, Claude followed the brief accurately 87% of the time compared to ChatGPT's 61% (based on my testing across 40 client briefs in Q1 2026).
That said, Claude is not magic. The quality ceiling is set by what you feed it. Claude with a vague prompt produces the same generic output as any other LLM. Claude with a structured brief built from live SERP data produces content that competes with human writers. The prompts below are designed to build that structured input layer step by step.
Important context: These prompts assume you have access to the actual content of competitor articles. You can get this by copy-pasting from the live page, using a tool like Screaming Frog or thruuu to extract content, or by using Claude's web search capability (Pro plan). The prompts work with any of these input methods.
The 8 Prompts: From Competitor Research to Published Content
This is always step one. Before writing anything, you need to know what is already ranking, how it is structured, and where the gaps are. This prompt takes the top 3 ranking articles for your target keyword and extracts everything you need to build a better piece.
I am writing an article targeting the keyword "[YOUR TARGET KEYWORD]". Below are the top 3 ranking articles for this keyword. Analyze all three and provide:
1. HEADING STRUCTURE: Extract every H1, H2, and H3 from each article. Present as a side-by-side comparison table.
2. CONTENT GAPS: What topics, subtopics, or questions does Article 1 cover that Articles 2 and 3 miss? What does Article 2 cover that the others miss? What does Article 3 cover that the others miss?
3. UNIQUE ANGLES: What unique data points, case studies, statistics, or original insights does each article contain?
4. WORD COUNT ANALYSIS: Estimate the word count of each article. Which sections are the longest? Which are thin?
5. QUESTION COVERAGE: What questions does each article answer? List every implied question from the headings plus any questions answered within the body text.
6. WEAKNESS ANALYSIS: For each article, identify the weakest sections: thin content, outdated stats, vague advice without specifics, missing internal links, or poor structure.
7. CITATION READINESS: Score each article 1-5 on AI citation readiness: does it open sections with direct answers? Are headings phrased as questions? Are statistics cited inline with sources?
Here are the three articles:
[PASTE ARTICLE 1 FULL TEXT]
[PASTE ARTICLE 2 FULL TEXT]
[PASTE ARTICLE 3 FULL TEXT]
It forces Claude to do comparative analysis, not just summarize. The side-by-side heading structure instantly shows you which topics are universal (must cover) and which are unique (opportunity to differentiate). The citation readiness score tells you whether you are competing against AI-optimized content or legacy SEO content, which changes your entire approach.
After the teardown, you know what competitors cover. This prompt identifies what they all miss, which is where your biggest ranking and citation opportunity sits. Content that fills a genuine gap does not need to outcompete on backlinks. It just needs to exist.
Based on your analysis of the 3 competitor articles above for the keyword "[YOUR TARGET KEYWORD]", identify:
1. MISSING SUBTOPICS: What subtopics would a reader expect to find that none of the 3 articles cover?
2. MISSING QUESTIONS: What questions would someone searching this keyword likely have that none of the articles answer? Think about follow-up questions, objections, and "what happens next" questions.
3. MISSING DATA: What statistics, benchmarks, or data points would strengthen an article on this topic that none of the competitors include?
4. MISSING FORMATS: What content formats (comparison tables, step-by-step checklists, decision frameworks, scoring rubrics) would make this topic clearer but are missing from all 3 articles?
5. MISSING AUDIENCE SEGMENTS: Do all 3 articles target the same reader? Is there an underserved audience segment (beginners, enterprise, specific industry) that none addresses?
For each gap, rate the opportunity as HIGH (no competitor covers it and search demand exists), MEDIUM (one competitor partially covers it), or LOW (most competitors cover it but poorly).
It structures the gap analysis into five specific dimensions. Most people ask Claude "what are the gaps?" and get a vague list. This prompt forces Claude to think about subtopics, questions, data, formats, and audience segments separately, which surfaces opportunities you would miss with a single-pass analysis.
This is the most important prompt in the sequence. A good brief is the difference between a first draft that needs two hours of editing and one that needs twenty minutes. This prompt builds a complete, section-by-section brief that a writer (human or AI) can follow exactly.
Create a detailed content brief for an article targeting the primary keyword "[YOUR TARGET KEYWORD]".
Secondary keywords to include naturally: [LIST 3-5 SECONDARY KEYWORDS]
Based on the competitor analysis and gap analysis above, build a brief with:
1. TITLE TAG: 3 options, each under 60 characters, primary keyword in the first half
2. META DESCRIPTION: 2 options, each under 155 characters, includes a clear benefit and a call to action
3. H1: One H1 that matches the title tag but can be slightly longer and more descriptive
4. FULL OUTLINE: Every H2 and H3 heading for the article. Requirements:
- H2s should be phrased as questions or clear topical statements
- Order should follow the user journey: definition first, then how-to, then advanced, then FAQ
- Include the primary keyword in 2-3 headings naturally
- Each H2 section should have a target word count in parentheses
5. SECTION BRIEFS: For each H2 section, provide:
- The key point this section must make (1 sentence)
- The direct answer that should open the section (40-60 words)
- Specific data, examples, or case studies to include
- Internal link opportunities (suggest which of my existing pages to link to and with what anchor text)
6. FAQ SECTION: 5 questions for the FAQ, phrased exactly as a user would type them into a search engine. For each, provide a 2-3 sentence answer.
7. SCHEMA RECOMMENDATIONS: What schema types should this page include?
My existing pages that could be internally linked:
[LIST YOUR EXISTING BLOG POST URLs AND TITLES]
It builds the brief from competitor data rather than from Claude's general knowledge. The "direct answer that should open the section" instruction is critical for answer engine readiness. The internal linking section ensures every new piece of content strengthens your existing site structure from day one.
Now you write. This prompt takes the brief from Prompt 3 and produces a full draft with the structural characteristics that AI search systems look for when selecting content to cite.
Write the full article based on the content brief above. Follow these rules exactly:
STRUCTURE RULES:
- Follow every H2 and H3 from the brief. Do not add, remove, or rename any headings.
- Open every H2 section with a direct, complete answer in the first 40-60 words. This answer must be understandable on its own without reading the rest of the section.
- After the direct answer, expand with explanation, examples, and supporting detail.
- Target the word count specified for each section in the brief.
WRITING RULES:
- Average sentence length: under 20 words
- Paragraphs: 2-4 sentences maximum
- Use active voice. Convert every passive construction to active.
- Include specific numbers, dates, tool names, and sources. No generic advice.
- Every statistic must include an inline source: "According to [Source] ([Year])..." or "([Source], [Year])"
VOICE AND STYLE:
- Write as a practitioner sharing a tested process, not as a teacher explaining a concept
- Never use these phrases: "in today's digital world", "it's important to note", "delve into", "it's worth mentioning", "in conclusion", "without further ado", "game-changer", "leverage"
- Do not start more than one paragraph with "This" or "It"
- Do not use em dashes
INTERNAL LINKS:
- Include all internal links specified in the brief with the exact anchor text provided
- Place internal links contextually within paragraphs, not as standalone sentences
AEO FORMATTING:
- Ensure the intro paragraph contains a clear, direct definition or explanation of the main topic in the first 2 sentences
- Include at least one comparison table where the brief suggests it
- End with a FAQ section using the exact questions from the brief
The banned phrases list eliminates the most common AI-sounding patterns. The "direct answer first" rule at every H2 is the single highest-impact instruction for AI citation readiness. The internal linking instructions mean you get a draft with links already placed, not something you have to go back and add manually. The no em dash rule keeps the writing clean and readable.
After the first draft, run it back through Claude alongside the original competitor articles. This catch step identifies any areas where your draft is thinner, less specific, or less useful than what already ranks.
Compare my draft article (below) against the 3 competitor articles analyzed earlier. For each H2 section of my draft, evaluate:
1. COVERAGE DEPTH: Is my section more detailed, equal, or less detailed than the best competitor's equivalent section? If less detailed, what specific information am I missing?
2. FACTUAL DENSITY: How many specific, verifiable facts (statistics, tool names, dates, study results) are in my section vs the competitors? If I have fewer, suggest specific facts to add with sources.
3. READABILITY: Is my section easier or harder to read than the competitors? Flag any sections with sentences over 25 words, passive voice, or unclear structure.
4. UNIQUENESS: Does my section add anything the competitors do not have? If not, suggest one original angle, data point, or example I could add.
5. AI CITATION READINESS: Does my section open with a direct answer that an AI system could extract and cite? If not, rewrite the opening 2 sentences to be citation-ready.
Here is my draft:
[PASTE YOUR DRAFT]
It compares section by section, not the whole article at once. This catches specific weak spots that a general "is my article good?" prompt would miss. The citation readiness check on each section also gives you a second pass on AEO formatting before publication.
Title tags directly affect click-through rate, and in 2026 they also influence whether AI systems select your page as a citation source. A clear, specific title with the primary keyword front-loaded signals topical relevance to both Google and AI retrieval systems.
Generate 5 title tag variations for my article about [TOPIC]. Currently ranking position #[POSITION] for "[TARGET KEYWORD]".
Current title: "[PASTE CURRENT TITLE]"
Each variation must:
1. Stay within 55-60 characters
2. Include the primary keyword "[KEYWORD]" in the first half
3. Test a different psychological approach: (a) specificity with numbers, (b) authority/expertise signal, (c) urgency or timeliness, (d) benefit-driven, (e) curiosity gap
4. Sound natural, not keyword-stuffed
5. Work for both Google SERP display and AI citation attribution
For each variant, explain: which psychological trigger it uses, why it might improve CTR over the current title, and any risk of the change.
Also generate 3 meta description variants (under 155 characters each) that pair well with the best title options. Each should include one specific benefit and end with a soft CTA.
This prompt takes an existing article's headings and converts them from keyword-label format to question-based format optimized for featured snippets and AI citation extraction. Use it on articles that are already published but underperforming in AI search.
Rewrite all headings for my article about [TOPIC] into question-based or explicit-statement format optimized for featured snippets and AI citations.
Current headings:
[PASTE ALL H2 AND H3 HEADINGS]
Target keyword: "[KEYWORD]"
Requirements:
1. Convert each heading into a natural question that real users would type into a search engine or ask an AI assistant
2. Ensure the heading sequence follows a logical user journey: what is it, how does it work, how do I do it, what are the common mistakes, what are the alternatives
3. Include the target keyword naturally in 2-3 headings (not forced)
4. Make each question specific enough that the section can provide a definitive answer (avoid "What should I know about X?" style questions, prefer "How does X affect Y?" or "What is the best way to do X?")
5. Keep heading length between 6-12 words
Provide the full restructured outline with explanations for each change.
For existing articles that have gone stale: this prompt identifies every outdated fact, statistic, tool reference, and date in your content and suggests current replacements. Content freshness is a hard requirement for AI citation on Perplexity (which crawls in real time) and a strong signal for Google AI Overviews, where 85% of citations come from content published in the last two years.
Audit the following article for content freshness. Today's date is [TODAY'S DATE].
For every piece of the following types, flag it and suggest a replacement:
1. STATISTICS with a date or source older than 2024
2. TOOL NAMES or FEATURES that may have changed (especially Google products, AI tools, and SEO platforms)
3. PRICING references that may be outdated
4. PROCESS DESCRIPTIONS that may have changed due to algorithm updates or platform changes
5. SCREENSHOTS or EXAMPLES that reference old UI or old data
6. Any statement that says "currently", "recently", "this year", or "in [year]" where the year is before 2025
For each flagged item, provide:
- The exact text to replace
- A suggested replacement with a current source (if I need to look up the current number, tell me what to search for)
- How critical the update is: CRITICAL (factually wrong now), HIGH (misleading), or LOW (still roughly accurate but dated)
Also check: is the dateModified in the schema markup current? If not, flag it.
Here is the article:
[PASTE ARTICLE CONTENT]
How to Score Your Content Writing Workflow
After running prompts 1 through 5 (the core writing sequence), evaluate your process against this checklist. Each item scores one point. A workflow scoring 7-8 consistently produces content that ranks within 60 days and starts appearing in AI citations within 30 days of indexing.
- Competitor content was analyzed before writing began (Prompt 1)
- Content gaps were identified and prioritized (Prompt 2)
- A section-by-section brief was created with direct-answer openings (Prompt 3)
- The draft follows the brief structure exactly with AEO formatting (Prompt 4)
- The draft was compared against competitors for depth and factual density (Prompt 5)
- Title tag and meta description were tested with multiple variants (Prompt 6)
- Headings are phrased as questions or explicit statements (Prompt 7)
- All statistics are from 2024 or later with inline source attribution (Prompt 8)
Claude vs ChatGPT for SEO Content: What the Data Shows
| Factor | Claude (Sonnet 4.6) | ChatGPT (GPT-4o) |
|---|---|---|
| Context window | 200K tokens (3 full articles + brief) | 128K tokens (2 articles + brief) |
| Instruction following on multi-part briefs | 87% accuracy (my testing, 40 briefs) | 61% accuracy (same briefs) |
| Filler sentence frequency | Low (rarely pads output) | Moderate (adds "it's worth noting" style padding) |
| Heading structure adherence | Follows brief headings exactly | Sometimes adds or renames headings |
| Internal link placement | Places links contextually as instructed | Often clusters links at end of sections |
| Direct-answer opening compliance | Consistently produces 40-60 word direct openings | Often opens with context before the answer |
| Banned phrase avoidance | Reliably avoids all specified phrases | Occasionally uses banned phrases despite instructions |
| Best for | Long-form, structured, brief-driven content | Short-form, creative, conversational content |
What to Do After Writing: The Publication Checklist
Once the draft is complete and reviewed, run through this before hitting publish. These are the steps most content teams skip that directly affect whether the page gets cited in AI search.
- Add schema markup: BlogPosting or Article schema with headline, author, datePublished, dateModified, and publisher properties. Add FAQPage schema if the article includes a FAQ section.
- Check llms.txt and robots.txt: Confirm GPTBot, PerplexityBot, ClaudeBot, and Google-Extended are not blocked. If you have an llms.txt file, add the new article to it.
- Internal link from existing pages: Do not just link from the new article to old ones. Go to 3-5 existing articles and add links pointing to the new article. This is the step most people forget.
- Submit to Google Search Console: Request indexing for the new URL immediately after publishing.
- Run a manual AI citation test: Ask your target queries in Perplexity and ChatGPT Search 2-3 weeks after indexing. Note your baseline and retest monthly.
Frequently Asked Questions About Claude SEO Content Writing Prompts
