# Export Posts for AI Decision Making - Complete Guide

## What This Script Does

Exports **ALL posts from all 3 WordPress sites** with complete details to CSV, so you can:
1. Upload to Claude or other AI for analysis
2. Get AI recommendations for:
   - Which site each post should be on
   - Which posts to consolidate (duplicates)
   - Which posts to delete (low-traffic)
   - Content gaps to fill

---

## Quick Start

```bash
python scripts/export_posts_for_ai_decision.py
```

**Output:** `output/reports/all_posts_for_ai_decision_TIMESTAMP.csv`

---

## What Gets Exported

### For Each Post:
- **Site**: Which website it's on (mistergeek.net, webscroll.fr, hellogeek.net)
- **Post ID**: WordPress ID
- **Status**: Published or Draft
- **Title**: Post title
- **URL**: Full post URL
- **Dates**: Published and modified dates
- **Categories & Tags**: Current categorization
- **Content Preview**: First 500 characters (for context)
- **Excerpt**: Post excerpt
- **SEO Data**:
  - Rank Math title
  - Meta description
  - Focus keyword
- **Word Count**: Content length

---

## Complete Workflow

### Step 1: Export All Posts

```bash
python scripts/export_posts_for_ai_decision.py
```

**Output:**
```
=======================================================================
EXPORTING ALL POSTS FOR AI DECISION MAKING
=======================================================================

mistergeek.net:
  Total: 328
  Published: 266
  Drafts: 62

webscroll.fr:
  Total: 17
  Published: 13
  Drafts: 4

hellogeek.net:
  Total: 23
  Published: 20
  Drafts: 3

───────────────────────────────────────────────────────────────────
Total across all sites: 368 posts
  Published: 299
  Drafts: 69
───────────────────────────────────────────────────────────────────

✓ CSV file: output/reports/all_posts_for_ai_decision_20260216_150000.csv
```

### Step 2: Open CSV and Review

```bash
open output/reports/all_posts_for_ai_decision_*.csv
```

You'll see all posts with their full details in a spreadsheet format.

### Step 3: Upload to AI for Analysis

**Option A: Claude (Recommended)**
1. Copy the CSV file path
2. Open https://claude.ai
3. Paste the CSV content or upload the file
4. Ask Claude to analyze and recommend:
   ```
   Please analyze this CSV of blog posts and:
   1. Categorize each by topic (VPN, software, gaming, torrenting, streaming, etc.)
   2. Recommend which website each should be on:
      - mistergeek.net: High-value (VPN, software, gaming, tech guides)
      - webscroll.fr: Torrenting/file-sharing content
      - hellogeek.net: Low-traffic, experimental, off-brand
   3. Identify duplicate/similar posts that should be consolidated
   4. Flag posts for deletion (very low word count or clearly spam)
   5. Provide a CSV with recommendations
   ```

**Option B: ChatGPT**
1. Upload CSV file
2. Ask same analysis questions

**Option C: Google Sheets + Claude**
1. Import CSV to Google Sheets
2. Add column: "AI Recommendation"
3. Use Claude to fill in recommendations
4. Share sheet with team for decisions

### Step 4: Create Master Decision Spreadsheet

Based on AI recommendations, create a master sheet with decisions:

| Site | Post ID | Title | Current Site | Recommended | Action | Priority | Notes |
|------|---------|-------|--------------|------------|--------|----------|-------|
| mistergeek.net | 2845 | Best VPN 2025 | mistergeek | mistergeek | KEEP | High | High traffic, optimize |
| mistergeek.net | 1234 | YggTorrent Guide | mistergeek | webscroll.fr | MOVE | Medium | Torrent content |
| mistergeek.net | 5678 | Random Post | mistergeek | hellogeek | MOVE | Low | Very low traffic |
| webscroll.fr | 100 | Tracker Guide | webscroll | webscroll | KEEP | High | Core content |

### Step 5: Execute Moves

```bash
# Week 1: Move posts to webscroll.fr
# Week 2: Move posts to hellogeek.net
# Week 3-4: Consolidate duplicates
# Week 5: Optimize remaining posts on mistergeek.net
```

---

## CSV Columns Explained

### Identification
- **site**: Current website
- **post_id**: WordPress post ID
- **status**: "publish" or "draft"

### Content
- **title**: Post title
- **slug**: URL slug
- **url**: Full post URL
- **excerpt**: Short excerpt if available
- **content_preview**: First 500 characters of post content (for topic analysis)
- **word_count**: Number of words in post

### Metadata
- **date_published**: When published
- **date_modified**: Last update
- **author_id**: Post author
- **categories**: WordPress categories
- **tags**: WordPress tags

### SEO
- **seo_title**: Rank Math SEO title
- **meta_description**: Rank Math or Yoast meta description
- **focus_keyword**: Primary keyword

---

## AI Prompt Examples

### Prompt 1: Basic Categorization

```
I have a CSV of 368 blog posts from 3 websites. Please:

1. Categorize each post by PRIMARY topic:
   - VPN
   - Software/Tools
   - Gaming
   - Streaming
   - Torrenting
   - File-Sharing
   - General Tech
   - Other

2. For each post, recommend which site it should be on:
   - mistergeek.net: VPN, Software, Gaming, General Tech (high-value)
   - webscroll.fr: Torrenting, File-Sharing (niche audience)
   - hellogeek.net: Other, low-traffic experimental content

3. Return a CSV with columns:
   post_id, current_site, title, recommended_site, topic, reason
```

### Prompt 2: Identify Duplicates

```
Please identify posts that cover the same or very similar topics:

1. Group similar posts together
2. For each group, identify which is the best (highest quality, most traffic)
3. Recommend keeping the best and consolidating others into it
4. Suggest which posts to delete vs merge

Return: List of duplicate groups with consolidation recommendations
```

### Prompt 3: Strategic Recommendations

```
Based on this data, provide strategic recommendations for:

1. Which topics are over-represented?
2. Which topics are under-represented?
3. What content gaps exist?
4. Which low-traffic posts should be deleted?
5. What new content should be created?
6. How to optimize each site's focus?

Consider SEO benefits of topic consolidation and site specialization.
```

---

## Using AI Recommendations

Once you get AI recommendations:

1. **Create master spreadsheet** in Google Sheets with all decisions
2. **Share with team** for final approval
3. **Document assumptions** (e.g., "Traffic = quality indicator")
4. **Plan execution** by priority and complexity
5. **Execute moves** following the [MULTI_SITE_STRATEGY.md](MULTI_SITE_STRATEGY.md) guide

---

## Expected CSV Size

- **368 posts** = ~150-200 KB CSV file
- Can be opened in:
  - Excel
  - Google Sheets
  - Apple Numbers
  - Any text editor

---

## Command Options

```bash
# Basic usage (default)
python scripts/export_posts_for_ai_decision.py

# Custom output location
python scripts/export_posts_for_ai_decision.py --output /path/to/my_export.csv
```

---

## Example AI Response Format

When you ask Claude to analyze, it might return:

```csv
post_id,current_site,title,recommended_site,topic,action,reason
2845,mistergeek.net,Best VPN 2025,mistergeek.net,VPN,KEEP,High traffic + relevance
1234,mistergeek.net,YggTorrent Guide,webscroll.fr,Torrenting,MOVE,Belongs in torrent-focused site
5678,mistergeek.net,Random Niche,hellogeek.net,Other,MOVE,Very low traffic + off-brand
...
```

---

## Best Practices

1. **Include context in prompts**: Tell AI your goal (improve SEO, consolidate authority)
2. **Ask for reasoning**: "Why should this post move?"
3. **Use multiple analyses**: Get 2-3 different recommendations and compare
4. **Manual review**: Don't blindly follow AI, use it to inform your decisions
5. **Test incrementally**: Move a few high-confidence posts first, measure impact, then scale

---

## Next Steps

1. **Run export:**
   ```bash
   python scripts/export_posts_for_ai_decision.py
   ```

2. **Upload CSV to Claude:**
   - Open https://claude.ai
   - Upload CSV file or paste content
   - Ask for categorization and site recommendations

3. **Review AI output** and create master decision spreadsheet

4. **Execute moves** following MULTI_SITE_STRATEGY.md guide

5. **Monitor results** for 30 days in Google Analytics

---

## Troubleshooting

### "No posts found"
- Check credentials in .env
- Verify WordPress sites are online
- Try running diagnostic: `python scripts/multi_site_seo_analyzer.py --diagnose https://www.mistergeek.net`

### "Authentication failed"
- Verify username and app password in .env
- Check user has read permission
- Re-generate app password in WordPress

### CSV is empty or missing columns
- Check that WordPress REST API is returning data
- Verify Rank Math plugin is active (for SEO fields)
- Check for errors in terminal output

---

Ready to export and analyze? Run:
```bash
python scripts/export_posts_for_ai_decision.py
```