lookalike-audience-builder

name	lookalike-audience-builder
description	Build a lookalike audience from a seed person, company, lead list, or website with customer logos, then search for similar profiles with match reasoning and personalization fields.
metadata	{"author":"amplemarket","version":"1.0.4","category":"Prospecting & Lead Generation"}
compatibility	Requires Amplemarket MCP server

Lookalike Audience Builder

Build a lookalike audience from a seed person, company, lead list, or website with customer logos, then search for similar profiles with match reasoning and personalization fields.

Instructions

When a user wants to find prospects similar to an existing person, company, lead list, or set of customers visible on a website, follow these steps to extract seed attributes, refine criteria, and search for lookalike matches.

Steps

1. Ask the user for a seed. The seed can be any of:

   • A LinkedIn URL or email address (person seed)
   • A person's name + company (person seed)
   • A company name or domain (company seed)
   • An Amplemarket lead list name or ID (lead list seed)
   • A website URL that displays customer logos (website seed)

   If the user's request is ambiguous, ask which seed type they intend.

2. Determine seed type and extract attributes. Process the seed based on its type:

   • Person seed: Call mcp__claude_ai_Amplemarket__enrich_person with the LinkedIn URL, email, or name + company. Then call mcp__claude_ai_Amplemarket__enrich_company with the person's company domain. Extract:
     • Title and role keywords
     • Seniority level
     • Department / job function
     • Industry and sub-industry
     • Company size range
     • Company type
     • Location (person and company)
   • Company seed: Call mcp__claude_ai_Amplemarket__enrich_company with the company domain or LinkedIn URL. Extract:
     • Industry and sub-industry
     • Company size range
     • Company type (public, private, etc.)
     • Headquarters location
     • Tech stack and key signals
   • Lead list seed: Call mcp__claude_ai_Amplemarket__get_lead_list with the list ID or name. Analyze the leads to find common patterns:
     • Most frequent titles and seniority levels
     • Most common industries
     • Dominant company size ranges
     • Top locations
     • Shared departments or job functions
   • Website seed: Call WebFetch with the URL to retrieve the page content. Identify customer logos, company names, or "Trusted by" sections. For each identified customer company, call mcp__claude_ai_Amplemarket__enrich_company with the company domain. Aggregate the enrichment results to extract common attributes across the customer base:
     • Most frequent industries
     • Dominant company size ranges
     • Common company types
     • Geographic clusters

3. Present extracted seed attributes to the user in a clear summary. For person seeds, show a profile card. For company seeds, show a company card. For list and website seeds, show a pattern analysis table with frequency counts.

4. Ask the user to prioritize attributes. Present: "Based on the seed, here are the common attributes: [list]. Which are most important for finding lookalikes? Any attributes to exclude or override?" Wait for the user's response before proceeding.

5. Resolve enum values by calling:

   • mcp__claude_ai_Amplemarket__get_industries to validate and map industry terms to exact API values.
   • mcp__claude_ai_Amplemarket__get_job_functions to validate and map job function terms to exact API values.

   Match the user's prioritized attributes to the correct enum values.

6. Run the lookalike search by calling mcp__claude_ai_Amplemarket__search_people with refined filters based on the user's priorities. Set full_output to true and page_size to 20. Apply filters:

   • person_titles: title keywords from the seed (use variations)
   • person_seniorities: matched seniority level(s)
   • person_departments: matched department(s)
   • person_locations: seed location(s) or user-specified locations
   • company_industries: validated industry values
   • company_sizes: seed company size range (expand by one tier in each direction)
   • company_types: seed company type if relevant
   • Exclude the seed person or company from results where possible using negative filters.

7. Present results with match reasoning. Format results as a table with columns: Name, Title, Company, Location, Match Reason. For each prospect, generate a concise match reason explaining which attributes align with the seed (e.g., "Same seniority + industry + company size range as seed").

8. Generate dynamic fields for each matched prospect. Populate the {{lookalike_*}} fields described in the Dynamic Fields Generated section below. These fields can be used in outreach templates to reference the similarity between the prospect and the seed.

9. Offer to create a lead list with the lookalike results. If the user agrees, call mcp__claude_ai_Amplemarket__create_lead_list with:

   • name: a descriptive name (e.g., "Lookalikes of Sarah Chen - VP Marketing - Mar 2026")
   • type: "linkedin" (if LinkedIn URLs are available) or "email"
   • leads: array of lead objects from the search results
   • options: ask about enrichment preferences (enrich, validate_email, reveal_phone_numbers)

   Include the lookalike context fields in the list description so downstream outreach can reference the match reasoning.

Important Notes

• Always validate industry and job function values via the API before searching. Never guess enum values.
• For website seeds, only extract companies that are clearly identifiable as customers (look for "Trusted by", "Our customers", "Used by" sections). Do not include partners or integration logos.
• When the seed is a lead list, analyze at least 10 leads (or all leads if fewer than 10) to establish reliable patterns.
• If the lookalike search returns zero results, relax filters one at a time in priority order: location first, then company size, then seniority, then industry.
• Exclude the original seed person or company from the results to avoid duplicate outreach.

Dynamic Fields Generated

Field	Description
{{lookalike_seed_name}}	Name of the seed person or company used as the basis for the search
{{lookalike_match_reason}}	Why this prospect matches the seed (e.g., "Same title + industry + company size as seed")
{{lookalike_shared_attributes}}	Comma-separated list of shared attributes (e.g., "VP-level, SaaS industry, 201-500 employees")
{{lookalike_company_similarity}}	How the prospect's company compares to the seed company (e.g., "Similar stage Series B SaaS, 30% smaller headcount")
{{lookalike_seniority_match}}	Whether seniority level matches the seed (e.g., "Exact match: VP-level")
{{lookalike_industry_match}}	Whether industry matches the seed (e.g., "Same industry: Computer Software")
{{lookalike_title_similarity}}	How the prospect's title compares to the seed (e.g., "Equivalent role: VP Marketing vs. Head of Marketing")
{{lookalike_location_match}}	Whether location aligns with the seed (e.g., "Same metro: San Francisco Bay Area")
{{lookalike_company_size_match}}	Whether company size is similar to the seed (e.g., "Adjacent range: 501-1000 vs. seed 201-500")
{{lookalike_suggested_opener}}	Draft opening line referencing the similarity (e.g., "I work with several VP Marketing leaders at Series B SaaS companies like [seed company]. Thought you might face similar challenges at [prospect company].")

Examples

Example 1: Person Seed (LinkedIn URL)

User prompt: "Find people like linkedin.com/in/sarah-chen-vp-marketing"

What the skill does:

1. Calls mcp__claude_ai_Amplemarket__enrich_person with linkedin_url: "https://linkedin.com/in/sarah-chen-vp-marketing", reveal_email: true.
2. Calls mcp__claude_ai_Amplemarket__enrich_company with domain from enrichment (e.g., "cloudmetrics.io").
3. Extracts seed attributes: VP of Marketing, VP seniority, Marketing department, Cloud Analytics / Computer Software industry, 180 employees (201-500 range), Austin TX.
4. Presents: "Seed profile - Sarah Chen, VP Marketing at CloudMetrics. Key attributes: VP-level, Marketing, Software/Analytics industry, 201-500 employees, Austin TX. Which matter most?"
5. User says: "Title and industry are most important, location doesn't matter."
6. Calls mcp__claude_ai_Amplemarket__get_industries and mcp__claude_ai_Amplemarket__get_job_functions to validate values.
7. Calls mcp__claude_ai_Amplemarket__search_people with:

• person_titles: ["VP of Marketing", "VP Marketing", "Vice President of Marketing", "Head of Marketing"]
• person_seniorities: ["VP", "Head"]
• person_departments: ["Marketing"]
• company_industries: [matched software/analytics values]
• company_sizes: ["51-200 employees", "201-500 employees", "501-1000 employees"]
• full_output: true, page_size: 20

Example output:

Name	Title	Company	Location	Match Reason
David Park	VP of Marketing	DataLoom	Boston, MA	Same title + industry + company size as seed
Lisa Nguyen	Head of Marketing	MetricFlow	Denver, CO	Equivalent seniority, same industry, adjacent size
Tom Rivera	VP Marketing & Growth	AnalyticsPro	Chicago, IL	Same title + industry, slightly larger company

Found 83 total lookalike matches. Showing page 1 of 5.

Dynamic fields for David Park:

• {{lookalike_seed_name}}: Sarah Chen (VP Marketing, CloudMetrics)
• {{lookalike_match_reason}}: Same title, industry, and company size range as seed
• {{lookalike_shared_attributes}}: VP-level, Marketing, Computer Software, 201-500 employees
• {{lookalike_company_similarity}}: DataLoom is a similar-stage analytics company with 220 employees vs. CloudMetrics' 180
• {{lookalike_seniority_match}}: Exact match: VP-level
• {{lookalike_industry_match}}: Same industry: Computer Software
• {{lookalike_title_similarity}}: Exact match: VP of Marketing
• {{lookalike_location_match}}: Different city (Boston vs. Austin), user excluded location
• {{lookalike_company_size_match}}: Same range: 201-500 employees
• {{lookalike_suggested_opener}}: "I work with several VP Marketing leaders at analytics companies like CloudMetrics. Thought you might face similar demand gen challenges scaling at DataLoom."

Would you like me to create a lead list from these 83 lookalike matches?

Example 2: Website with Customer Logos Seed

User prompt: "Build a lookalike audience from the customers on acme-analytics.com/customers"

What the skill does:

1. Calls WebFetch with URL "https://acme-analytics.com/customers" to retrieve the page content.
2. Identifies customer logos and company names from the "Trusted by" section: Stripe, Notion, Figma, Datadog, Plaid.
3. Calls mcp__claude_ai_Amplemarket__enrich_company for each: stripe.com, notion.so, figma.com, datadoghq.com, plaid.com.
4. Analyzes patterns across the 5 companies:

• Industries: 4/5 Computer Software, 3/5 Fintech-adjacent
• Sizes: 3/5 in 1001-5000 range, 2/5 in 5001-10000 range
• Types: 3/5 Privately Held, 2/5 Public Company
• Locations: 4/5 San Francisco Bay Area

5. Presents pattern analysis and asks: "Common profile across these customers: mid-to-late-stage software companies, 1000-10000 employees, mostly Bay Area. Which attributes matter most?"
6. User says: "Industry and size are key. Search for VP Engineering and CTO at similar companies."
7. Validates enums, then calls mcp__claude_ai_Amplemarket__search_people with:

• person_titles: ["VP of Engineering", "VP Engineering", "CTO", "Chief Technology Officer"]
• person_seniorities: ["VP", "C-Suite"]
• person_departments: ["Engineering & Technical"]
• company_industries: [matched software values]
• company_sizes: ["1001-5000 employees", "5001-10000 employees"]
• full_output: true, page_size: 20

8. Returns results with match reasoning referencing the customer pattern, not a single seed.

Example output:

Name	Title	Company	Location	Match Reason
Raj Mehta	VP of Engineering	Streamline AI	San Francisco, CA	Same industry + size range as 4/5 seed customers
Kim Okada	CTO	PaymentGrid	New York, NY	Fintech-adjacent software, 2000 employees matches seed pattern
Alex Werner	VP Engineering	CloudVault	Austin, TX	Software industry, 1500 employees, matches seed company profile

Found 156 total lookalike matches. Showing page 1 of 8.

Dynamic fields for Raj Mehta:

• {{lookalike_seed_name}}: Acme Analytics customer base (Stripe, Notion, Figma, Datadog, Plaid)
• {{lookalike_match_reason}}: Company profile matches 4/5 seed customer attributes: Software industry, 1001-5000 employees, Privately Held, Bay Area
• {{lookalike_shared_attributes}}: Computer Software, 1001-5000 employees, Privately Held, San Francisco Bay Area
• {{lookalike_company_similarity}}: Streamline AI is a mid-stage AI software company with 1,200 employees, similar profile to Datadog and Plaid at comparable stages
• {{lookalike_seniority_match}}: VP-level as requested
• {{lookalike_industry_match}}: Same industry: Computer Software
• {{lookalike_title_similarity}}: VP of Engineering - matches requested persona
• {{lookalike_location_match}}: San Francisco Bay Area - matches 4/5 seed customers
• {{lookalike_company_size_match}}: Same range: 1001-5000 employees (matches 3/5 seed customers)
• {{lookalike_suggested_opener}}: "Companies like Stripe and Datadog trust Acme Analytics for their cloud metrics. Given Streamline AI is at a similar stage, I thought it might be relevant to your engineering team too."

Would you like me to create a lead list from these 156 lookalike matches?

Example 3: Lead List Seed

User prompt: "Find more prospects similar to my 'Q4 Closed Won' lead list"

What the skill does:

1. Calls mcp__claude_ai_Amplemarket__list_lead_lists to find the list, then mcp__claude_ai_Amplemarket__get_lead_list with the matching list ID.
2. Analyzes the leads in the list (e.g., 25 leads). Finds common patterns:

• Titles: 60% Director-level, 25% VP-level, 15% Manager
• Departments: 80% Revenue/Sales, 20% Marketing
• Industries: 70% Financial Services, 20% Insurance
• Company sizes: 65% 501-1000, 25% 1001-5000
• Locations: 50% Northeast US, 30% Midwest

3. Presents: "Your closed-won leads are predominantly Director/VP-level in Revenue at financial services companies with 500-5000 employees, concentrated in the Northeast. Which patterns should I prioritize?"
4. User says: "Focus on the Director-level Revenue people at financial services. Expand to all US locations."
5. Validates enums and runs:

• person_seniorities: ["Director", "VP"]
• person_departments: ["Revenue"]
• company_industries: [matched financial services values]
• company_sizes: ["501-1000 employees", "1001-5000 employees"]
• person_locations: ["United States"]
• full_output: true, page_size: 20

6. Returns results with match reasoning referencing the closed-won pattern.

Example output:

Name	Title	Company	Location	Match Reason
Angela Torres	Director of Sales	Heritage Financial Group	Dallas, TX	Matches 4/5 closed-won attributes: Director, Revenue, Financial Services, 501-1000
Brian Caldwell	VP of Revenue	Midwest Bancorp	Chicago, IL	Matches 5/5 closed-won attributes: VP, Revenue, Financial Services, 1001-5000, Midwest
Nadia Hassan	Director of Business Development	SecureInsure	Atlanta, GA	Matches 3/5: Director, Revenue-adjacent, Insurance (related industry)

Found 210 total lookalike matches. Showing page 1 of 11.

Dynamic fields for Angela Torres:

• {{lookalike_seed_name}}: Q4 Closed Won lead list (25 leads)
• {{lookalike_match_reason}}: Matches 4/5 closed-won pattern attributes: Director-level, Revenue department, Financial Services, 501-1000 employees
• {{lookalike_shared_attributes}}: Director-level, Revenue, Financial Services, 501-1000 employees
• {{lookalike_company_similarity}}: Heritage Financial Group is a mid-size financial services firm with 650 employees, aligning with the 501-1000 range dominant in closed-won deals
• {{lookalike_seniority_match}}: Exact match: Director-level (60% of closed-won leads)
• {{lookalike_industry_match}}: Same industry: Financial Services (70% of closed-won leads)
• {{lookalike_title_similarity}}: Director of Sales - consistent with Revenue department pattern
• {{lookalike_location_match}}: Dallas, TX - outside the dominant Northeast cluster but within expanded US scope per user request
• {{lookalike_company_size_match}}: Same range: 501-1000 employees (65% of closed-won leads)
• {{lookalike_suggested_opener}}: "We have helped several Directors of Sales at financial services companies similar to Heritage Financial Group. Happy to share what has worked for them if useful."

1. Offers to create a "Lookalikes of Q4 Closed Won - Mar 2026" lead list.

Troubleshooting

Problem	Solution
Zero results from lookalike search	Relax filters one at a time in this order: 1) Remove location filter. 2) Expand company size by one tier in each direction. 3) Broaden seniority to include one adjacent level. 4) Add related industry values from mcp__claude_ai_Amplemarket__get_industries. Report each change to the user.
Website seed yields no customer logos	Ask the user to provide specific company names instead. Alternatively, try fetching alternate pages like /customers, /case-studies, or /about.
Lead list seed has too few leads for pattern analysis	If fewer than 5 leads, warn the user that patterns may not be reliable. Suggest supplementing with additional seed data or treating the top 1-2 leads as individual person seeds.
Enrichment fails for seed person or company	Fallback chain: 1) Try alternate identifiers (domain instead of name, LinkedIn URL instead of email). 2) If person enrichment fails, fall back to company-only seed. 3) Ask the user for additional identifiers.
Too many results, audience is too broad	Add more filters from the seed attributes the user deprioritized. Tighten company size range. Add person_departments or company_types constraints. Show a refined preview after each adjustment.
Match reasoning feels generic	Ensure you are comparing specific attribute values, not just categories. Reference exact titles, company sizes, and industries rather than saying "similar profile."
Seed company exists but enrich_company returns sparse data	Fallback chain: 1) Try the company LinkedIn URL. 2) Try mcp__claude_ai_Amplemarket__search_companies with the domain. 3) Use whatever partial data is available and note gaps to the user.
WebFetch returns blocked or empty page	Some sites block automated fetches. Ask the user to paste the customer list manually, or try an alternate URL on the same domain (e.g., /about, /case-studies).