42-paa-scraper

name: 42-paa-scraper description: "Recursively scrape People Also Ask questions up to 5 levels deep plus Related Searches tree. Builds a question tree for FAQ content and featured snippet strategy. Use when user says paa scraper, people also ask, paa tree, question research, faq mining, paa extraction, related searches, question tree, paa depth, 42-paa." user-invokable: true argument-hint: " [--depth 3] [--related] [--provider dataforseo]" version: "1.0.0" tags: - seo - paa - serp - content-strategy - faq - featured-snippets allowed-tools: - Read - Write - Grep - Glob - Bash - WebFetch metadata: filePattern: "/paa,/question,**/faq" bashPattern: "pythonpaa"

42 PAA Scraper

Recursively scrape People Also Ask (PAA) questions and Related Searches from Google SERPs. Builds a hierarchical question tree that reveals exactly what users are asking about a topic, enabling FAQ content strategy and featured snippet targeting.

Why PAA Matters

People Also Ask boxes appear on approximately 65% of Google SERPs. Each PAA question, when clicked, reveals more questions -- creating an infinite expansion of user intent signals. By recursively following this tree, you build a complete map of every question Google associates with a topic.

This data directly fuels:

• FAQ sections that match exact user questions
• Featured snippet targeting (PAA answers are pulled from featured snippets)
• Content gap discovery (questions your site does not answer)
• Topical authority planning (breadth of questions reveals topic scope)
• AI citation optimization (question-answer pairs are the format AI models prefer to cite)

Commands

/42:paa-scraper <keyword> [--depth 3] [--related] [--depth 3] [--related]
/42:paa-scraper <csv-file> [--depth 3] [--related] [--depth 3] [--related]

Arguments

API Configuration

DataForSEO (Primary)

Set environment variables:

export DATAFORSEO_LOGIN="your_login"
export DATAFORSEO_PASSWORD="your_password"

API endpoint: https://api.dataforseo.com/v3/serp/google/organic/live/advanced

The script extracts PAA from the people_also_ask field and Related Searches from the related_searches field in the API response.

Workflow

Step 1: Parse Input

• If input is a keyword string: use it as the single seed keyword.
• If input is a CSV file: read all keywords from the file. Support both single-column CSVs and CSVs with a keyword column header.
• Validate that at least one keyword is provided.

Step 2: Fetch SERP for Seed Keyword

For each seed keyword:

1. Call the SERP API with the keyword, location, and language.
2. Extract the PAA questions from the response (typically 4 questions per SERP).
3. If --related is set, also extract Related Searches (typically 8 per SERP).
4. Store each question/search as a node in the tree with metadata: parent keyword, depth level, position.

Step 3: Recursive PAA Expansion

For each PAA question extracted at depth N (where N < max depth):

1. Wait --delay seconds (rate limiting).
2. Fetch the SERP for that exact question.
3. Extract child PAA questions from the response.
4. Check each child against the global deduplication set.
5. If the child is new (not seen before), add it to the tree at depth N+1.
6. If the child is a duplicate, mark it as [seen] but do not recurse into it.
7. Continue until max depth is reached or no new questions are found.

Step 4: Related Searches Tree (Optional)

If --related is enabled, build a parallel tree:

1. For each Related Search at depth N (where N < max depth):
   • Fetch its SERP.
   • Extract its Related Searches.
   • Deduplicate and add to tree.
2. Related Searches tree is kept separate from PAA tree in the output.

Step 5: Deduplication

Maintain a global set of all seen questions (normalized: lowercased, stripped of punctuation). This prevents:

• The same question appearing in multiple branches.
• Infinite loops where PAA questions reference each other.
• Wasted API calls on already-explored questions.

Step 6: Generate Output

Produce three output formats:

1. PAA-TREE.md -- Hierarchical tree with indentation showing parent-child relationships.
2. Flat list -- All unique questions as a simple bulleted list (appended to the same file).
3. CSV export -- paa-export.csv with columns: question, depth, parent, seed_keyword.

Output Format: PAA-TREE.md

# PAA Tree: [seed keyword]

**Generated:** [date]
**Depth:** [max depth]
**Total unique questions:** [count]
**API calls made:** [count]

---

## Hierarchical Question Tree

- **What is [topic]?**
  - What is [topic] used for?
    - What are the benefits of [topic]?
    - How does [topic] work in practice?
  - What is the difference between [topic] and [alternative]?
    - Which is better, [topic] or [alternative]?
  - What is [topic] in simple terms?
- **How to [topic]?**
  - How to [topic] for beginners?
    - What is the easiest way to [topic]?
  - How to [topic] step by step?
- **Is [topic] worth it?**
  - How much does [topic] cost?
  - What are the disadvantages of [topic]?

## Related Searches Tree (if --related)

- [related search 1]
  - [child related search]
- [related search 2]

---

## Flat Question List (for content planning)

1. What is [topic]?
2. What is [topic] used for?
3. What are the benefits of [topic]?
...

---

## Export

CSV exported to: `paa-export.csv`

Rate Limiting and Cost Awareness

For a CSV with 10 keywords at depth 3: ~210 API calls, ~$0.42.

Default delay of 1 second between requests. Increase with --delay 2 for conservative usage.

Script Location

scripts/paa_scraper.py

Run directly:

python scripts/paa_scraper.py "best crm software" --depth 3 --related
python scripts/paa_scraper.py keywords.csv --depth 2

Cross-References

• 42:seo-agi -- Use PAA tree output to inform content brief structure and FAQ sections.
• 42:content -- PAA questions map directly to H2/H3 headings and FAQ schema.
• 42:blog-geo -- PAA questions are the exact format AI models cite; use for AI-optimized content.
• 42:topical-map -- PAA tree reveals topic boundaries and subtopic relationships.
• 42:striking-distance -- Cross-reference PAA questions with pages already ranking positions 4-20.