Broad vs lookalike — the test that ends the debate

Broad vs lookalike — the test that ends the debate

Still building 1% lookalikes by hand? Compare honestly:

— Lookalike (audience modeled on your converters) gives a warm-ish start and predictable CPMs. Great when your pixel data is thin and you need guardrails.
— Broad (age/geo only, let the algo find them) scales further and often beats lookalikes once you've got volume — Meta's targeting got that good.

When each:
— Under ~50 conversions/week or new account → lookalike, give the algo a hint.
— Mature pixel, want to break a ceiling → broad, the lookalike pool's too small to scale.

Risk: broad burns budget early while it calibrates — don't kill it in 24h on a panic.

Run broad vs your best lookalike head to head, same creative, same budget.