Definition Lookalike builds from a seed list. The list can be customer emails, website visitors, app users, or converted users exported from CRM.
How it works The platform extracts the behavioral fingerprint of seed users: interests, demographics, behavior patterns, purchase signals. The algorithm searches other platform users for the most similar profiles. A similarity percentage is chosen (1-10% on Meta): a lower percentage means tighter similarity and a smaller pool. The ad publishes to the pool. Seed list quality and size directly impact lookalike performance.
Where you see it in Scope Trends The **Audience Management** tab holds lookalike source lists and performance comparison. The **Meta Ad Command Center** suggests lookalike seeds.
Frequently asked questions **How large should the lookalike seed list be?** Meta requires a minimum of 100 users; 1,000-50,000 is ideal. Very small seeds yield weak models.
**Did iOS 14.5 hurt lookalike quality?** Yes, the conversion signal contraction impacted quality. Modeled conversion partially recovered it.