How Lookalike Expansion Works
Lookalike audiences are created by uploading a "seed audience", a list of your best customers, highest-value purchasers, or qualified leads, to a platform like Meta, TikTok, LinkedIn, or Snapchat. The platform's machine learning models analyze hundreds of behavioral, demographic, and interest signals from seed audience members and identify other users with similar profiles who have no prior relationship with your brand. This produces a prospecting audience with substantially higher baseline conversion affinity than generic interest targeting.
Why Lookalike Expansion Matters for B2B Marketing
On Meta, lookalike audiences are created in percentages from 1% to 10% of a country's total population. A 1% lookalike is the most similar to the seed audience (smallest, highest quality); a 10% lookalike is the broadest (largest, lower similarity). Best practice is to launch with 1-3% lookalikes for quality, then expand to 3-5% as the campaign scales. The quality of the seed audience is the single most important factor, a seed built from 200 top-paying customers will outperform a seed built from 2,000 general website visitors.
Lookalike Expansion: Best Practices & Strategic Application
Seed audience size significantly impacts lookalike quality. Meta recommends a seed of 1,000-50,000 people for optimal modeling; seeds under 100 people produce unreliable lookalikes. For most advertisers, the best seed sources ranked by quality are: (1) CRM list of closed-won customers, (2) purchase/conversion pixel audiences, (3) top-10-percentile website visitor audiences, (4) video 95% viewer audiences. Uploading hashed email, phone, first name, last name, and location data maximizes match rate to 60-80%.
Agency Perspective: Lookalike Expansion in Practice
MV3 implements lookalike expansion as a core prospecting layer for all paid social clients. We stack lookalikes in separate ad sets, 1%, 1-3%, 3-5%, and compare CPL and conversion quality across each. For B2B clients with small customer lists (under 500 customers), we build value-based lookalikes by uploading customer lifetime value alongside the email list, allowing Meta to model on the characteristics of the highest-revenue customers rather than treating all customers equally.
