---
name: geo-expert
description: Expert system for writing high-quality, GEO-optimized B2B blog articles following Google's E-E-A-T principles. GEO (Generative Engine Optimization) focuses on getting content cited by AI search engines (ChatGPT, Google AI Overview, Perplexity, Gemini). Use when user requests to (1) write a blog article or post, (2) create GEO/SEO-optimized content, (3) provides a topic/product description/materials/URL for article creation, (4) asks for enterprise/B2B technical content, or (5) mentions keywords like "GEO", "SEO article", "blog post", "content marketing". Generates complete Word documents with title, search title, keywords, meta description, and article body (within 2500 words).
---

# GEO Expert - Enterprise Content Creation

As a GEO (Generative Engine Optimization) expert, create high-quality articles optimized for both traditional search engines AND AI-powered search platforms (ChatGPT, Google AI Overview, Perplexity, Gemini). Articles should follow E-E-A-T principles and be structured for AI citation and extraction.

## What is GEO?

GEO = Generative Engine Optimization. Unlike traditional SEO (optimizing for rankings and clicks), GEO optimizes content to be **cited and referenced by AI search engines** when they generate answers. The goal shifts from "rank on page 1" to "become the source AI quotes."

Key differences from SEO:
- **Optimization granularity**: Page-level → Paragraph-level (AI extracts individual passages)
- **Content style**: Can be marketing-flavored → Must be encyclopedic, neutral, fact-dense
- **Success metric**: Rankings & clicks → AI citation rate & brand mention frequency
- **Keyword approach**: Keyword density → Semantic concept coverage

## Workflow Overview

Follow these steps in order:

1. **Research & Analysis** - Understand the topic, conduct keyword research, and analyze AI citations
2. **Content Planning** - Structure the article for both E-E-A-T and AI extractability
3. **Writing** - Create the article with GEO-specific writing techniques
4. **Optimization** - Ensure all SEO + GEO elements are properly implemented
5. **Quality Check** - Verify content meets all requirements

## Step 1: Research & Analysis

### Understanding the Input

Users may provide:
- A specific topic or title
- A product/service description
- Reference materials or documents
- URLs to analyze or reference
- Target audience information

**If URL provided**: Use web_fetch to retrieve and analyze the content before proceeding.

**If materials uploaded**: Read and extract key information to inform the article.

### Keyword Research

Use web_search to identify optimal keywords:

1. **Primary keyword research**:
   - Search for "[topic] trends 2025"
   - Search for "[topic] best practices"
   - Search for "[topic] enterprise solutions"
   - Identify the most relevant keyword with good search volume

2. **Competitor & AI citation analysis**:
   - Search for primary keyword to see ranking articles
   - Analyze top 3-5 results using web_fetch
   - Note their structure, length, and keyword usage
   - Identify content gaps to address
   - **GEO-specific**: Test the primary keyword in AI search context — what would ChatGPT/Perplexity cite? Note what types of content (FAQ, comparison, definition) get referenced

3. **Secondary keywords**:
   - Find 3-5 related keywords from search results
   - Include long-tail variations (3-5 words)
   - **GEO-specific**: Include question-format keywords (how/what/why) that match AI search patterns

4. **Fan-out query mapping** (GEO-specific):
   AI search engines decompose complex questions into sub-queries. Map these to ensure comprehensive coverage:
   - Search primary keyword in Google → expand all "People Also Ask" → record sub-questions
   - If possible, test in Perplexity/ChatGPT → note the sub-topics covered in AI responses
   - Identify 4-6 sub-queries that your article's H2/H3 sections should directly answer
   - These sub-queries become your article's section headings (can be rewritten as statements)

**Document your findings**:
- Primary keyword: [keyword]
- Search volume estimate: [estimate]
- Secondary keywords: [list]
- Question-format keywords: [list] ← GEO addition
- Fan-out sub-queries: [list of 4-6 sub-questions] ← GEO addition
- Content gaps identified: [gaps]

## Step 2: Content Planning

### Read Reference Materials

Before writing, always read these reference files:

```bash
view references/geo-optimization-guide.md  # GEO core principles — READ FIRST
view references/eeat-guidelines.md
view references/seo-best-practices.md
view references/writing-style-guide.md
```

**`geo-optimization-guide.md` is the most important reference file** — it defines the eight core GEO optimization strategies that must be applied throughout the article.

### Article Structure (GEO-Optimized)

Plan your article following this GEO-optimized structure:

**TL;DR / Summary (50-80 words)**:
- 3-4 sentences summarizing the entire article
- Include the primary keyword and 1-2 key statistics
- This block is designed to be directly extractable by AI

**Introduction (150-200 words)**:
- **Answer-first**: Open with a direct, definitive statement answering the title's implied question
- Include primary keyword in first 50 words
- Include one compelling statistic with source attribution
- Promise clear value/outcomes
- Mention brand/product entity name within first 100 words

**Body (1800-2000 words)**:
- 3-5 main sections (H2 headings — use question-format when possible)
- Each section: 400-500 words
- **Answer-first**: Each H2 section starts with a 40-60 word direct answer
- **Self-contained paragraphs**: Each paragraph works independently
- **Statistic density**: At least 1 data point per 150-200 words
- **Inline citations**: Attributed sources ("According to [Source]...")
- Include subsections (H3) for complex topics

**Conclusion (100-150 words)**:
- Summarize key takeaways (extractable format)
- Provide actionable next steps
- End with appropriate CTA

**FAQ Section (3-5 questions)**:
- Cover what/why/how/comparison question types
- Each answer: 2-4 sentences, answer-first, with data
- Use natural language that matches real search queries

### E-E-A-T + GEO Implementation Checklist

- [ ] **Experience**: Include practical examples from real scenarios
- [ ] **Expertise**: Demonstrate deep technical knowledge with specific data
- [ ] **Authoritativeness**: Reference reputable sources with inline attribution (3-5 minimum)
- [ ] **Trustworthiness**: Acknowledge limitations, provide balanced perspective
- [ ] **Extractability**: Every paragraph stands alone as a citable unit
- [ ] **Fact density**: Statistics every 150-200 words
- [ ] **Entity clarity**: Brand/product names mentioned consistently

## Step 3: Writing the Article

### Initialize Document

Before writing, check the docx skill for best practices:

```bash
view /mnt/skills/public/docx/SKILL.md
```

### GEO Writing Rules (MUST FOLLOW)

These six rules distinguish GEO content from traditional SEO content:

#### Rule 1: Answer-First Writing
Every section (H2) must open with a direct answer in the first 40-60 words. Do NOT build up to the answer — state it immediately, then elaborate.

#### Rule 2: Self-Contained Paragraphs
Each paragraph must be independently understandable. Avoid:
- "As mentioned above..." / "如上所述..."
- "The aforementioned..." / "前面提到的..."
- Any expression that requires reading previous paragraphs to understand

Each paragraph = one complete idea + supporting evidence.

#### Rule 3: Statistic Density
Embed a concrete data point (percentage, number, metric) at least once every 150-200 words. Prioritize:
1. First-party data (product metrics, customer results)
2. Authoritative third-party (Gartner, IDC, Forrester)
3. Academic research
4. Industry benchmarks

#### Rule 4: Inline Citation Attribution
Use attributed citations throughout, not vague references:
- ✅ "According to Gartner's 2024 CPaaS report, the market will reach $34B by 2028."
- ✅ "IDC research shows enterprises using cloud communications reduce costs by 35%."
- ❌ "Research shows..." / "Studies indicate..." / "Experts say..."

Minimum 3-5 attributed citations per article.

#### Rule 5: FAQ Section
Include 3-5 FAQ items at the end. Format:
- Question in natural search language
- Answer starts with direct conclusion
- Answer includes at least one data point
- 2-4 sentences maximum per answer

#### Rule 6: Entity Clarity
- Use full brand/product name on first mention, then consistent short form
- Competitor names use official names
- Key entities appear 2-3+ times throughout
- Primary brand entity appears within first 100 words

### Create the Article

Use python-docx to create a professional Word document with:

1. **Metadata section** (at the top):
   - Search Title (搜索标题): 50-60 characters
   - Meta Description (摘要): 150-160 characters
   - Primary Keyword (主关键词): [keyword]
   - Secondary Keywords (次关键词): [keyword list]
   - Question Keywords (问题关键词): [question-format keywords] ← GEO addition
   - URL Slug: [keyword-phrase]
   - Target Audience: [enterprise decision-makers/technical teams]
   - Schema Markup Recommendation: [FAQPage / HowTo / Article] ← GEO addition

2. **Article content**:
   - TL;DR summary block ← GEO addition
   - H1: Article title (include primary keyword)
   - Introduction (answer-first, with statistic and source)
   - H2: Main sections (3-5 sections, question-format headings preferred)
   - H3: Subsections as needed
   - Conclusion
   - FAQ section (3-5 Q&As) ← GEO addition
   - Optional: References section

### Writing Guidelines

**Follow these rules strictly**:

- **Length**: Within 2,500 words total (approximately 2,500 Chinese characters or 2,500 English words)
- **Paragraphs**: 2-4 sentences maximum, each self-contained
- **Sentences**: 15-20 words average
- **Tone**: Professional, encyclopedic, neutral (avoid first-person marketing language for GEO)
- **Language**: Clear, concrete, no jargon without explanation
- **Evidence**: Every claim backed by specific data + attributed source
- **Examples**: Include at least 2-3 concrete cases or scenarios
- **Statistics**: At least 1 per 150-200 words

**Keyword optimization**:
- Primary keyword: Natural integration (not density-focused — GEO uses semantic matching)
- Include in: title, first 50 words, 1-2 H2 headings, FAQ, conclusion
- Use variations and synonyms naturally
- **GEO shift**: Focus on concept coverage over keyword repetition

**GEO tone guidance**:
- Prefer neutral third-person over first-person ("Research indicates..." over "We recommend...")
- This aligns with encyclopedic sources that AI platforms (especially ChatGPT) prefer to cite
- Exception: Customer case studies can use first-person for authenticity

**B2B enterprise focus**:
- Address business impact and ROI with specific numbers
- Consider security, compliance, scalability
- Discuss implementation complexity
- Acknowledge costs and trade-offs
- Use enterprise-relevant examples

### Code Implementation Pattern

```python
from docx import Document
from docx.shared import Pt, Inches
from docx.enum.text import WD_ALIGN_PARAGRAPH

# Create document
doc = Document()

# Configure default style
style = doc.styles['Normal']
style.font.name = 'Arial'
style.font.size = Pt(11)

# Add metadata section
doc.add_heading('SEO / GEO Metadata', level=2)
metadata = [
    ('Search Title', '[your optimized title]'),
    ('Meta Description', '[your description]'),
    ('Primary Keyword', '[keyword]'),
    ('Secondary Keywords', '[keyword list]'),
    ('Question Keywords', '[question-format keywords]'),
    ('URL Slug', '[slug]'),
    ('Target Audience', '[audience]'),
    ('Schema Markup', 'FAQPage, Article'),
]
for label, value in metadata:
    p = doc.add_paragraph()
    p.add_run(f'{label}: ').bold = True
    p.add_run(value)

# Add separator
doc.add_paragraph('_' * 80)

# Add TL;DR
doc.add_heading('TL;DR', level=2)
doc.add_paragraph('[3-4 sentence summary with key statistic]')

# Add article content
doc.add_heading('[Article Title]', level=1)
doc.add_heading('Introduction', level=2)
doc.add_paragraph('[Answer-first opening with statistic and source attribution]')

# Main sections with question-format headings
doc.add_heading('[What/How/Why Question as Section Title]', level=2)
doc.add_paragraph('[Direct answer in first 40-60 words, then elaborate with data]')

# FAQ section
doc.add_heading('FAQ', level=2)
doc.add_paragraph('Q: [Natural language question]', style='List Number')
doc.add_paragraph('A: [Direct answer with data point, 2-4 sentences]')

# Save
doc.save('/mnt/user-data/outputs/article_title.docx')
```

## Step 4: SEO + GEO Optimization

### Verify All Optimization Elements

Before finalizing, check:

**Traditional SEO Metadata**:
- [ ] Search title: 50-60 chars, includes primary keyword
- [ ] Meta description: 150-160 chars, includes primary + secondary keyword
- [ ] URL slug: Short, clean, includes primary keyword

**GEO Content Structure**:
- [ ] TL;DR summary block at top (3-4 sentences, extractable)
- [ ] Answer-first opening in every H2 section (first 40-60 words = direct answer)
- [ ] Self-contained paragraphs (no backward references)
- [ ] FAQ section with 3-5 Q&As
- [ ] Question-format H2 headings (at least 2-3)

**Fact Density & Citations**:
- [ ] Statistics every 150-200 words (minimum 8-10 per 2000-word article)
- [ ] Inline attributed citations: 3-5 minimum ("According to [Source]...")
- [ ] First-party data included where available
- [ ] All statistics include source and timeframe

**Entity & Brand Signals**:
- [ ] Primary brand/product name in first 100 words
- [ ] Consistent naming throughout (no mixed abbreviations)
- [ ] Competitor names use official full names
- [ ] Key entities mentioned 2-3+ times

**Schema Markup Recommendation**:
- [ ] Recommended schema types noted in metadata (FAQPage, Article/BlogPosting, HowTo)
- [ ] FAQ content formatted to match FAQPage schema structure

**Traditional SEO Elements**:
- [ ] Primary keyword in first 50 words
- [ ] Primary keyword in H1 title
- [ ] Keyword variations in 1-2 H2 headings
- [ ] 1-2 external links to authoritative sources
- [ ] 1-2 internal link suggestions (placeholder text)

**Content Structure**:
- [ ] Clear H1, H2, H3 hierarchy
- [ ] Short paragraphs (2-4 sentences, each self-contained)
- [ ] Bullet points for lists
- [ ] Readable sentence length (15-20 words avg)

**E-E-A-T Signals**:
- [ ] Evidence-based claims with attributed sources
- [ ] Practical examples included
- [ ] Sources cited with specific names and dates
- [ ] Honest about limitations
- [ ] Actionable recommendations

## Step 5: Quality Check

### Final Review

Read through the completed article and verify:

1. **GEO readiness** (NEW — check first):
   - Extract any 3 random paragraphs — do they each make sense standalone?
   - Does every H2 section answer its implied question in the first 2 sentences?
   - Count statistics — is there at least 1 per 150-200 words?
   - Count attributed citations — are there at least 3-5?
   - Is the FAQ section complete with data in each answer?

2. **Content quality**:
   - Provides genuine value to enterprise readers
   - Addresses search intent completely
   - Includes specific, actionable information
   - Avoids fluff and generic statements
   - Tone is encyclopedic and neutral (not marketing-speak)

3. **Technical accuracy**:
   - All technical details are correct
   - Version numbers specified where relevant
   - Examples are valid and tested
   - External links are working and appropriate

4. **Readability**:
   - Professional but accessible tone
   - Clear logical flow between sections
   - Complex concepts explained simply
   - No unnecessary jargon

5. **Word count**: Within 2,500 words (not including metadata and FAQ)

### Output Delivery

Save the final document to `/mnt/user-data/outputs/` with a descriptive filename:

```python
output_filename = f"{primary_keyword.replace(' ', '_')}_article.docx"
doc.save(f'/mnt/user-data/outputs/{output_filename}')
```

Present the file to the user with a brief summary of:
- Article topic and primary keyword
- Word count
- Key sections covered
- Number of statistics and cited sources
- Recommended schema markup types
- Any notable GEO features or recommendations

### GEO Monitoring Appendix

Every article delivered should include a **GEO Monitoring** section at the end of the Word document (after References, before the document ends). This section helps the content team track AI visibility after publishing.

Include:

```
## GEO 监测建议

### 目标 AI 搜索 Prompt
基于本文关键词，建议用以下 prompt 在 AI 平台中测试本文是否被引用：
1. [prompt 1 — 基于主关键词的自然语言问题]
2. [prompt 2 — 基于 fan-out 子查询]
3. [prompt 3 — 品牌+关键词组合]

### 监测平台
- ChatGPT (chatgpt.com)
- Google AI Overview (google.com)
- Perplexity (perplexity.ai)

### 监测频率
- 发布后第 1 周：首次检查是否被收录
- 此后每月 1 次：检查引用状态变化

### 关注指标
- Citation：本文是否出现在 AI 回答的引用来源中
- Brand Mention：品牌名称是否在 AI 回答正文中被提及
- Accuracy：AI 对产品/品牌的描述是否准确
- Share of Voice：在同类查询中，本品牌 vs 竞品的出现频率
```

## Additional Notes

### When User Provides URL

If user provides a URL for reference:

1. Use web_fetch to retrieve the content
2. Analyze for key points and insights
3. Identify content gaps to address
4. Reference appropriately (with citation) in your article
5. Do NOT copy content - create original analysis and perspective

### Handling Multiple Topics

If user provides multiple topics or a broad scope:

1. Ask user to prioritize or select primary focus
2. Suggest breaking into multiple articles if needed
3. For a single article, choose the most cohesive angle

### Asset Template Usage

A Word template is available at `assets/geo_article_template.docx` showing the expected structure. You can reference this for formatting, but always create documents programmatically using python-docx for consistency.

### Iterative Refinement

If user requests revisions:
- Use str_replace to edit the existing document file
- Or regenerate sections as needed
- Maintain consistency with original keyword strategy

## Common Patterns

### Technical Tutorial Article
Focus: How to implement a specific technology/solution
Structure: Problem → Solution → Implementation → Best Practices → FAQ → Conclusion
GEO tip: Use "How to..." H2 headings, include step-by-step with specific metrics

### Comparison Article
Focus: Tool A vs Tool B for enterprise use
Structure: Overview → Criteria → Detailed Comparison → Recommendations → FAQ → Conclusion
GEO tip: Comparison articles account for 32.5% of AI citations — include structured comparison tables

### Best Practices Guide
Focus: Recommended approaches for a challenge
Structure: Context → Core Principles → Implementation → Common Pitfalls → FAQ → Conclusion
GEO tip: Number your best practices, include measurable outcomes for each

### Trend Analysis
Focus: Industry developments and implications
Structure: Current State → Emerging Trends → Impact Analysis → Preparation Strategies → FAQ → Conclusion
GEO tip: Heavy on recent statistics, update frequently (AI platforms favor content updated within 30 days)

## Troubleshooting

**If keyword research yields no clear primary keyword**:
- Broaden the search terms
- Look for related concepts
- Check for enterprise-specific variations
- Consult with user on target search terms

**If competitor content is too technical/not technical enough**:
- Aim for balanced depth appropriate for enterprise decision-makers
- Include both strategic and technical perspectives
- Use executive summaries or TL;DR sections

**If word count is too short**:
- Add more examples and case studies
- Expand on implementation details
- Include more data and research
- Add a FAQ or common challenges section

**If word count is too long**:
- Remove redundant information
- Tighten paragraphs and sentences
- Consider splitting into multiple articles
- Move excessive detail to appendix

**If statistics are hard to find**:
- Use web_search to find industry reports and benchmarks
- Check vendor blogs for performance data
- Use general industry benchmarks as reference points
- Clearly state when data is estimated vs. verified