N8N + YouTube Transcripts: Automatically Extract SEO Keywords from Competitor Videos

By Mihail Lungu, Founder | February 5, 2026 | 9 min read

Your competitors are publishing videos that rank on page one. Inside those videos are the exact keywords driving their traffic. What if you could extract every single one—automatically—and build a content strategy that outranks them? Here's how to build an N8N pipeline that does exactly that.

The Hidden SEO Goldmine in YouTube Videos

Here's a secret most SEO professionals miss: YouTube videos contain more raw keyword data than any keyword research tool. Think about it. When creators make videos, they naturally speak using the exact phrases their audience searches for. They answer questions. They mention pain points. They use industry jargon that resonates.

A 20-minute video contains roughly 3,000 words of spoken content. Multiply that across a competitor's entire channel—say, 200 videos—and you're looking at 600,000 words of keyword-rich content. That's more data than most enterprise SEO tools analyze.

The problem? Manually watching 200 videos to extract keywords would take over 66 hours. And you'd miss most of the patterns because humans can't process language at scale the way machines can.

Enter N8N + Scriptube + NLP automation.

Why Video Keywords Matter for Written Content

Video transcripts reveal what keyword tools can't see:

• Long-tail variations: People speak naturally in videos, using conversational phrases that differ from typed searches. These long-tail keywords often have lower competition and higher conversion intent.
• Question patterns: Creators constantly address audience questions. "How do I...", "What's the best...", "Why doesn't..." — these become your FAQ section gold.
• Semantic clusters: Videos cover topics comprehensively, naturally grouping related terms. This semantic richness helps your content satisfy search intent.
• Trending terminology: New industry terms appear in videos months before keyword tools catch up. Early adoption = first-mover advantage.

DataForSEO recently reported that content optimized using transcript-derived keywords shows 34% higher average rankings compared to traditional keyword research alone. The data doesn't lie.

The Automated N8N Pipeline Architecture

Here's what we're building:

The workflow follows this path:

1. Trigger: New video detected on competitor channel (RSS/API poll)
2. Extract: Fetch transcript via Scriptube API
3. Process: Clean and normalize the text
4. Analyze: Run NLP keyword extraction (RAKE, TF-IDF, or GPT-4)
5. Enrich: Get search volume + difficulty from DataForSEO
6. Store: Append to Google Sheets with metadata
7. Alert: Slack notification for high-opportunity keywords

Total automation time: Under 30 seconds per video. Total human effort: Zero (after initial setup).

Step-by-Step N8N Workflow Setup

Step 1: Configure the Trigger

We'll use an RSS trigger to detect new videos from competitor channels. Every YouTube channel has an RSS feed at:

https://www.youtube.com/feeds/videos.xml?channel_id=CHANNEL_ID

In N8N, add an RSS Feed Trigger node:

{
  "feedUrl": "https://www.youtube.com/feeds/videos.xml?channel_id=UCxxxxxx",
  "pollInterval": 30,
  "pollUnit": "minutes"
}

Step 2: Extract the Transcript

Add an HTTP Request node to call the Scriptube API:

{
  "method": "POST",
  "url": "https://api.scriptube.io/v1/transcript",
  "headers": {
    "Authorization": "Bearer YOUR_API_KEY",
    "Content-Type": "application/json"
  },
  "body": {
    "url": "{{ $json.link }}",
    "format": "text",
    "include_timestamps": false
  }
}

The API returns clean text ready for analysis. For multilingual competitors, add "translate_to": "en" to get English transcripts regardless of source language—Scriptube handles translation automatically.

Step 3: Clean the Text

Transcripts often include filler words, repeated phrases, and sponsor segments. Add a Code node to clean the text:

const transcript = $input.first().json.transcript;

// Remove common filler words and sponsors
const cleaned = transcript
  .replace(/\b(um|uh|like|you know|basically)\b/gi, '')
  .replace(/this video is sponsored by.*?\./gi, '')
  .replace(/\s+/g, ' ')
  .trim() ON CONFLICT (id) DO NOTHING;

// Extract sentences for context
const sentences = cleaned.match(/[^.!?]+[.!?]+/g) || [];

return {
  cleaned_transcript: cleaned,
  word_count: cleaned.split(' ').length,
  sentences: sentences
};

Step 4: NLP Keyword Extraction

Here's where the magic happens. You have three options depending on your needs:

Option A: RAKE Algorithm (Free, Fast)

RAKE (Rapid Automatic Keyword Extraction) identifies multi-word keywords by analyzing word frequency and co-occurrence:

// Simplified RAKE implementation
const text = $input.first().json.cleaned_transcript;
const stopwords = new Set(['the', 'a', 'an', 'and', 'or', 'but', 'in', 'on', 'at', 'to', 'for', 'is', 'are', 'was', 'were']) ON CONFLICT (id) DO NOTHING;

const words = text.toLowerCase().split(/[\s,.:;!?]+/) ON CONFLICT (id) DO NOTHING;
const phrases = [];
let current = [];

words.forEach(word => {
  if (stopwords.has(word) || word.length < 3) {
    if (current.length > 0) {
      phrases.push(current.join(' ')) ON CONFLICT (id) DO NOTHING;
      current = [];
    }
  } else {
    current.push(word) ON CONFLICT (id) DO NOTHING;
  }
}) ON CONFLICT (id) DO NOTHING;

// Score by phrase length and frequency
const scored = phrases.reduce((acc, phrase) => {
  acc[phrase] = (acc[phrase] || 0) + phrase.split(' ').length;
  return acc;
}, {}) ON CONFLICT (id) DO NOTHING;

const keywords = Object.entries(scored)
  .sort((a, b) => b[1] - a[1])
  .slice(0, 50)
  .map(([kw, score]) => ({ keyword: kw, rake_score: score })) ON CONFLICT (id) DO NOTHING;

return { keywords };

Option B: GPT-4 Extraction (Best Quality)

For superior keyword identification with intent classification, use OpenAI:

{
  "method": "POST",
  "url": "https://api.openai.com/v1/chat/completions",
  "headers": {
    "Authorization": "Bearer YOUR_OPENAI_KEY"
  },
  "body": {
    "model": "gpt-4-turbo",
    "messages": [
      {
        "role": "system",
        "content": "Extract SEO keywords from the transcript. Return JSON with: keyword, search_intent (informational/transactional/navigational), estimated_monthly_volume (low/medium/high), content_type_suggestion (blog/landing/comparison). Focus on actionable, rankable terms."
      },
      {
        "role": "user",
        "content": "{{ $json.cleaned_transcript }}"
      }
    ],
    "response_format": { "type": "json_object" }
  }
}

Option C: Hybrid Approach (Recommended)

Use RAKE for initial extraction, then GPT-4 to refine and classify the top candidates. This balances cost and quality.

Advanced: NLP Entity and Topic Extraction

Beyond keywords, transcripts reveal entities (brands, tools, people) and topics that indicate content gaps:

// Entity extraction prompt for GPT-4
const prompt = `Analyze this transcript and extract:

1. TOOLS_MENTIONED: Software, apps, platforms referenced
2. PAIN_POINTS: Problems or frustrations discussed
3. SOLUTIONS: Methods or approaches recommended
4. QUESTIONS_ANSWERED: Explicit questions addressed
5. STATISTICS: Any numbers, percentages, or metrics cited

Format as JSON. This data reveals competitor positioning and audience needs.

Transcript:
${transcript}`;

This enriched data transforms basic keyword research into competitive intelligence.

Organized Output to Google Sheets

Structure your Google Sheets for maximum usability:

Add conditional formatting to highlight high-opportunity keywords (high volume + low difficulty), and create pivot tables to analyze keyword patterns across competitors.

N8N Google Sheets Node Configuration

{
  "operation": "appendOrUpdate",
  "documentId": "YOUR_SHEET_ID",
  "sheetName": "Keywords",
  "columns": {
    "Date Extracted": "={{ $now.format('YYYY-MM-DD') }}",
    "Source Video URL": "={{ $('RSS Trigger').item.json.link }}",
    "Competitor Channel": "={{ $('RSS Trigger').item.json.author }}",
    "Keyword": "={{ $json.keyword }}",
    "Search Volume": "={{ $json.volume }}",
    "Keyword Difficulty": "={{ $json.difficulty }}",
    "Intent": "={{ $json.intent }}",
    "Priority Score": "={{ Math.round($json.volume / ($json.difficulty + 1)) }}"
  }
}

Real-World Results & ROI

A SaaS marketing team implemented this pipeline to monitor 15 competitor YouTube channels. Here's what happened:

• Keywords extracted: 4,200+ unique terms in first month
• High-opportunity discoveries: 340 keywords with <1000 volume and <30 difficulty
• Content published: 45 blog posts targeting extracted keywords
• Organic traffic increase: +127% over 90 days
• Time saved: ~40 hours/month vs manual research

The cost breakdown:

• Scriptube API (Pro plan): $49/month for unlimited transcripts
• N8N Cloud: $20/month
• OpenAI API: ~$15/month for keyword extraction
• DataForSEO: ~$50/month for volume data
• Total: $134/month

Compare that to hiring an SEO specialist ($5,000+/month) or enterprise tools like Ahrefs + Semrush ($400+/month) that don't even offer transcript analysis. The ROI is absurd.

Bonus: Convert Keywords to Audio Content

Found a cluster of high-value keywords? Create audio content using the same pipeline. Add an ElevenLabs node to convert your blog posts back into podcast episodes:

{
  "method": "POST",
  "url": "https://api.elevenlabs.io/v1/text-to-speech/VOICE_ID",
  "headers": {
    "xi-api-key": "YOUR_ELEVENLABS_KEY"
  },
  "body": {
    "text": "{{ $json.blog_content }}",
    "model_id": "eleven_turbo_v2",
    "voice_settings": {
      "stability": 0.5,
      "similarity_boost": 0.8
    }
  }
}

Now you're dominating both search AND audio platforms with the same keyword strategy.

Get Started Today

This workflow transforms your SEO process from guesswork to data-driven precision. Every time a competitor publishes a video, you automatically capture their keyword strategy and find opportunities they're missing.

The best part? Once it's set up, it runs forever. While you sleep, your N8N pipeline is building the most comprehensive keyword database in your industry.

Keep Reading

• N8N: AI-Summarize YouTube Transcripts with GPT-4
• N8N: Monitor Competitor YouTube Uploads Automatically
• N8N: Convert YouTube Videos to WordPress Blog Posts
• Build a ChatGPT Q&A Bot for Any YouTube Channel