Files
seo/ENHANCED_ANALYSIS_GUIDE.md
Kevin Bataille 1744d8e7db Add enhanced analysis with selective field analysis and category proposer
New Features:
- Selective field analysis: Choose which fields to analyze (title, meta_description, categories, site)
- In-place CSV updates: Update input CSV with new columns (automatic backup created)
- Category proposer: Dedicated command for AI-powered category suggestions

New Commands:
- seo analyze -f title categories: Analyze specific fields only
- seo analyze -u: Update input CSV with recommendations
- seo category_propose: Propose categories based on content

New Scripts:
- enhanced_analyzer.py: Enhanced AI analyzer with selective analysis
- category_proposer.py: Dedicated category proposal tool

CLI Options:
- --fields, -f: Specify fields to analyze
- --update, -u: Update input CSV (creates backup)
- --output, -o: Custom output file path

Output Columns:
- proposed_title, title_reason (for title analysis)
- proposed_meta_description, meta_reason (for meta analysis)
- proposed_category, category_reason (for category analysis)
- proposed_site, site_reason (for site analysis)
- ai_confidence, ai_priority (common to all)

Documentation:
- ENHANCED_ANALYSIS_GUIDE.md: Complete guide with examples

Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
2026-02-16 14:57:42 +01:00

302 lines
7.2 KiB
Markdown

# Enhanced Analysis Guide
## Overview
The SEO automation tool now provides enhanced analysis capabilities with:
1. **Selective Field Analysis** - Choose which fields to analyze (title, meta description, categories, site)
2. **In-place CSV Updates** - Update input CSV with new columns (with automatic backup)
3. **Category Proposer** - Dedicated command for AI-powered category suggestions
## Commands
### 1. Enhanced Analysis (`seo analyze`)
Analyze posts with AI and add recommendation columns to your CSV.
#### Basic Usage
```bash
# Analyze all fields (default behavior)
./seo analyze
# Analyze specific CSV file
./seo analyze output/all_posts_2026-02-16.csv
```
#### Selective Field Analysis
```bash
# Analyze only titles
./seo analyze -f title
# Analyze titles and categories
./seo analyze -f title categories
# Analyze meta descriptions only
./seo analyze -f meta_description
# Analyze all fields
./seo analyze -f title meta_description categories site
```
#### Update Input CSV (In-place)
```bash
# Update input CSV with new columns (creates backup automatically)
./seo analyze -u
# Update with specific fields only
./seo analyze -u -f title meta_description
# Specify custom output file
./seo analyze -o output/my_analysis.csv
```
#### Output Columns
Depending on selected fields, the following columns are added:
**Title Analysis:**
- `proposed_title` - AI-suggested improved title
- `title_reason` - Explanation for title change
**Meta Description Analysis:**
- `proposed_meta_description` - AI-suggested meta description (120-160 chars)
- `meta_reason` - Explanation for meta description change
**Category Analysis:**
- `proposed_category` - AI-suggested best category
- `category_reason` - Explanation for category choice
**Site Analysis:**
- `proposed_site` - AI-suggested best site
- `site_reason` - Explanation for site recommendation
**Common Fields:**
- `ai_confidence` - AI confidence level (High/Medium/Low)
- `ai_priority` - Priority level (High/Medium/Low)
### 2. Category Proposer (`seo category_propose`)
Dedicated command for AI-powered category suggestions based on post content.
#### Usage
```bash
# Propose categories for latest export
./seo category_propose
# Propose categories for specific CSV
./seo category_propose output/all_posts_2026-02-16.csv
# Save to custom output file
./seo category_propose -o output/my_category_proposals.csv
```
#### Output Columns
- `post_id` - Post identifier
- `title` - Post title
- `current_categories` - Current categories
- `proposed_category` - AI-suggested category
- `alternative_categories` - Alternative category suggestions
- `category_reason` - Explanation for suggestion
- `category_confidence` - Confidence level
## Examples
### Example 1: Analyze Titles Only
```bash
# Analyze only titles for SEO optimization
./seo analyze -f title
# Output: analyzed_posts_YYYYMMDD_HHMMSS.csv
# Contains: original columns + proposed_title + title_reason + ai_confidence
```
### Example 2: Update CSV with Meta Descriptions
```bash
# Update input CSV with proposed meta descriptions
./seo analyze -u -f meta_description
# Creates:
# - all_posts_2026-02-16_backup_YYYYMMDD_HHMMSS.csv (backup)
# - all_posts_2026-02-16.csv (updated with new columns)
```
### Example 3: Full Category Analysis
```bash
# Propose categories for all posts
./seo category_propose
# Review proposals
open output/category_proposals_*.csv
# Apply approved categories manually in WordPress
```
### Example 4: Multi-Field Analysis
```bash
# Analyze titles and categories together
./seo analyze -f title categories
# Output includes:
# - proposed_title, title_reason
# - proposed_category, category_reason
# - ai_confidence, ai_priority
```
### Example 5: Targeted Analysis with Output
```bash
# Analyze meta descriptions, save to specific file
./seo analyze -f meta_description -o output/meta_analysis.csv
```
## Workflow Examples
### Workflow 1: SEO Title Optimization
```bash
# 1. Export posts
./seo export
# 2. Analyze titles only
./seo analyze -f title
# 3. Review proposed titles
open output/analyzed_posts_*.csv
# 4. Manually update best titles in WordPress
```
### Workflow 2: Category Reorganization
```bash
# 1. Export posts
./seo export
# 2. Get category proposals
./seo category_propose
# 3. Review proposals
open output/category_proposals_*.csv
# 4. Apply approved category changes
```
### Workflow 3: Complete SEO Audit
```bash
# 1. Export posts
./seo export
# 2. Analyze all fields
./seo analyze -f title meta_description categories site
# 3. Review comprehensive analysis
open output/analyzed_posts_*.csv
# 4. Implement changes based on AI recommendations
```
### Workflow 4: Incremental Analysis
```bash
# 1. Export posts
./seo export
# 2. Analyze titles (fast, low cost)
./seo analyze -f title
# 3. Later, analyze meta descriptions
./seo analyze -u -f meta_description
# 4. Later, analyze categories
./seo analyze -u -f categories
# Result: CSV progressively enriched with AI recommendations
```
## Cost Optimization
### Reduce API Costs
```bash
# Analyze only needed fields (saves tokens)
./seo analyze -f title # Cheaper than analyzing all fields
# Use smaller batch sizes for better control
# (edit script or use environment variable)
# Analyze in stages
./seo analyze -f title
./seo analyze -u -f meta_description
# Total cost similar, but better control over each step
```
### Token Usage by Field
Approximate token usage per 100 posts:
- **title**: ~500 tokens (lowest cost)
- **meta_description**: ~800 tokens
- **categories**: ~600 tokens
- **site**: ~400 tokens (lowest cost)
- **All fields**: ~2000 tokens (best value)
## Best Practices
1. **Start Small**: Test with `-f title` first to see AI quality
2. **Review Before Applying**: Always review AI suggestions before implementing
3. **Use Backups**: The `-u` flag creates automatic backups
4. **Batch Analysis**: Analyze related fields together for better context
5. **Confidence Matters**: Pay attention to `ai_confidence` column
6. **Iterative Process**: Enrich CSVs incrementally for better control
## Troubleshooting
### No CSV File Found
```bash
# Error: No CSV file found
# Solution: Run export first or provide file path
./seo export
./seo analyze
# Or specify file directly
./seo analyze path/to/your/file.csv
```
### API Key Not Set
```bash
# Error: OPENROUTER_API_KEY not set
# Solution: Add to .env file
echo "OPENROUTER_API_KEY=your_key_here" >> .env
```
### High API Costs
```bash
# Reduce costs by analyzing fewer fields
./seo analyze -f title # Instead of all fields
# Or analyze in batches
./seo analyze -f title
./seo analyze -u -f meta_description
```
## File Formats
### Input CSV Requirements
Must contain at minimum:
- `post_id` - Unique identifier
- `title` - Post title (for title analysis)
- `meta_description` - Current meta (for meta analysis)
- `categories` - Current categories (for category analysis)
- `site` - Current site (for site analysis)
- `content_preview` or `content` - Post content (recommended for all analyses)
### Output CSV Format
Standard output includes:
- All original columns
- New `proposed_*` columns for analyzed fields
- `*_reason` columns with explanations
- `ai_confidence` and `ai_priority` columns
---
**Version**: 1.0.0
**Last Updated**: 2026-02-16
**Related**: See ARCHITECTURE.md for system overview