What is a Lookalike Audience?

A lookalike audience (LAL) is a user segment the ad platform’s algorithm treats as similar to a seed audience: conversions, buyers, pixel visitors, or email/phone lists. The goal is to scale campaigns to cold users whose behavior resembles paying users.

Creation

On Meta, TikTok, VK, and others, you pick a source (custom audience on conversions, app events, CRM). The platform builds a model and outputs LAL at 1%–10% of a country’s population (1% is closest to seed; 10% is broader and cheaper). Seed quality is critical—noisy conversions produce weak lookalikes.

In media buying

Typical workflow: test cold broad → collect 50–100+ conversions → LAL 1–3% on purchases/leads → scale budget at stable CPA/ROAS. Compare against interests and broad in split tests. In the tracker, review CR and ROI by LAL campaign sub_id.

Seed size limits

Minimum seed size depends on the platform (often 100–1,000 users). On iOS after ATT, pixel-based seeds shrink—LALs from app events and server CAPI are more stable. LAL does not guarantee seed-level CR; matching is probabilistic.

Risks

Expanding LAL to 5–10% often drops CR. Prohibited verticals and policy violations in seed creatives risk account bans. Lookalikes trained on fraudulent conversions teach the algorithm on bots.