Linkedin Researcher

Phase 1: Influencer Discovery

Identify top voices in the niche using WebSearch:

Discovery queries:

• site:linkedin.com/in "{service_keyword}" "followers" "{industry}"
• site:linkedin.com/posts "{service_keyword}" "likes" OR "comments"
• "top linkedin influencer" "{industry}" OR "{service_keyword}"
• "linkedin creator" "{industry}" "{icp_title}"
• "best linkedin posts" "{service_keyword}" {time_window}

For each discovered influencer, capture:

{
  "name": "Full name",
  "profile_url": "LinkedIn URL",
  "headline": "Their LinkedIn headline",
  "follower_estimate": "Approximate follower count if visible",
  "niche": "Their primary topic area",
  "relevance_to_icp": "HIGH | MEDIUM | LOW"
}

Filtering rules:

• Prioritize creators whose audience overlaps with ICP titles (founders, CMOs, heads of ecommerce).
• Skip profiles with fewer than 1,000 estimated followers (micro-influencers handled by influencer-finder).
• Target 10-20 influencers for post analysis.

Phase 2: Post Collection

For each identified influencer and for niche keywords generally, search for high-performing posts:

Post search queries:

• site:linkedin.com/posts "{influencer_name}" "{service_keyword}"
• site:linkedin.com/posts "{service_keyword}" "agree" OR "this" OR "100%" (engagement markers)
• site:linkedin.com/pulse "{service_keyword}" "{industry}"
• "{influencer_name}" linkedin post "{topic_keyword}"

For each post found, extract:

{
  "post_url": "URL if available",
  "author": "Name",
  "hook": "First 2 lines of the post (the scroll-stopper)",
  "full_text": "Complete post text (first 500 chars if truncated)",
  "format": "text_only | listicle | story | contrarian | how_to | carousel | poll | video | article",
  "topic": "Primary topic of the post",
  "engagement_signals": "Likes/comments/reposts if visible in search snippet",
  "posted_date": "Date if available",
  "cta_type": "question | link | dm_me | comment_below | none",
  "length": "short (<500 chars) | medium (500-1500) | long (1500+)"
}

Phase 3: Pattern Analysis

Analyze collected posts to identify winning patterns:

Hook Analysis

Categorize all hooks into types:

• Contrarian: "Stop doing X" / "X is dead" / "Unpopular opinion:"
• Story opener: "Last week I..." / "3 years ago..." / "True story:"
• Data lead: "We analyzed X..." / "97% of..." / "After X conversions..."
• List promise: "X things I learned..." / "X mistakes that..." / "The X framework for..."
• Question: "Why do most..." / "What if..." / "Ever wondered..."
• Bold claim: "This one change..." / "The secret to..." / "Nobody talks about..."

Count frequency and estimate engagement per hook type.

Format Analysis

For each content format (text, carousel, poll, article, video):

• Count of posts found
• Average engagement signals
• Best-performing examples
• Common structural patterns

Topic Cluster Analysis

Group posts into topic clusters:

• Map each post to 1-2 topic clusters from service keywords
• Identify which clusters have highest engagement
• Find underserved topics (low competition, relevant to ICP)

Posting Pattern Analysis

If dates are available:

• Day of week distribution for high-performing posts
• Posting frequency of top creators
• Consistency patterns

Phase 4: Template Generation

Based on patterns identified, generate reusable templates:

For each of the top 5 hook types, create:

{
  "hook_type": "contrarian",
  "template": "Stop [common practice]. Here's what [top performers] do instead:",
  "example_filled": "Stop A/B testing your homepage hero. Here's what brands doing 8-figure revenue do instead:",
  "when_to_use": "When challenging conventional wisdom in your space",
  "engagement_prediction": "HIGH -- contrarian hooks get 2-3x more comments"
}

For each winning format, create a structural template:

• Text post template with hook, body, CTA
• Carousel outline template with slide-by-slide guidance
• Poll template with option framing guidance
• Article template with section structure

Content calendar seeds:

Generate 20 specific post ideas mapped to:

• Service keyword they promote
• ICP segment they target
• Hook type to use
• Format to use
• Estimated effort (low/medium/high)

Phase 5: Output

Return structured playbook:

{
  "research_summary": {
    "posts_analyzed": 50,
    "influencers_studied": 15,
    "time_period": "past 30 days",
    "platforms": ["LinkedIn"]
  },
  "top_influencers": [
    {
      "name": "...",
      "profile_url": "...",
      "headline": "...",
      "follower_estimate": "...",
      "niche": "...",
      "top_post_hook": "...",
      "content_style": "..."
    }
  ],
  "hook_analysis": {
    "contrarian": { "frequency": 12, "avg_engagement": "high", "examples": [] },
    "story_opener": { "frequency": 8, "avg_engagement": "medium", "examples": [] }
  },
  "format_analysis": {
    "text_only": { "count": 25, "avg_engagement": "medium", "best_example": "..." },
    "carousel": { "count": 10, "avg_engagement": "high", "best_example": "..." }
  },
  "topic_clusters": [
    { "topic": "Shopify CRO", "post_count": 8, "engagement": "high", "saturation": "medium" }
  ],
  "templates": [],
  "content_calendar": [],
  "recommendations": {
    "posting_frequency": "3-4x per week",
    "best_days": ["Tuesday", "Wednesday", "Thursday"],
    "top_formats": ["text_only", "carousel"],
    "top_hooks": ["contrarian", "data_lead"],
    "topics_to_own": ["...", "..."],
    "voice_notes": "Align with agency tone: direct, helpful, zero fluff"
  }
}

Present a formatted summary alongside the JSON:

LINKEDIN CONTENT PLAYBOOK
Analyzed: {N} posts from {M} influencers

TOP HOOKS THAT WORK:
1. {hook_type} -- used {N} times, {engagement} engagement
   Template: "{template}"

WINNING FORMATS:
1. {format} -- {count} posts, {engagement} avg

TOPIC OPPORTUNITIES:
1. {topic} -- {saturation} saturation, {engagement} potential

CONTENT CALENDAR (Next 20 posts):
1. [{format}] {topic} -- Hook: {hook_type} -- Effort: {level}
...

POSTING CADENCE: {frequency} on {best_days}

Example Usage

Trigger phrases:

• "Research LinkedIn content in our niche"
• "What's working on LinkedIn for Shopify agencies?"
• "Build a LinkedIn content playbook"
• "Analyze top LinkedIn posts about ecommerce"
• "Find viral LinkedIn content about D2C"
• "What hooks work best on LinkedIn?"

User: Research what's working on LinkedIn for Shopify and ecommerce content
Assistant: [reads config, discovers top influencers, collects high-performing posts, analyzes hooks/formats/topics, generates templates and content calendar, presents playbook]

User: Build a LinkedIn playbook focused on CRO content, study these 5 creators: [names]
Assistant: [same flow but focused on CRO keywords, studies specified creators plus discovers additional ones]