1% for best CPA. Expand to 3-5% when 1% frequency exceeds 3.0. 10% barely outperforms broad targeting.

100 in one country, 1K+ for stability. Quality over quantity — 500 high-value buyers beat 5K random leads.

Still effective in 2026?

Yes, less dominant post-iOS 14.5 but still 15-25% better than interest targeting. Combine with Advantage+ for best results.

← Back to AutoAdy

How to Use Lookalike Audiences on Meta Ads

Last updated: April 5, 2026

Lookalike audiences find people resembling your customers. 1% lookalike targets top 1% most similar (~2.1M in US) with lowest CPA. Source quality beats size — 500 high-value customers outperform 5,000 random leads.

Key Takeaways

✓1% = lowest CPA (2.1M in US); 5% = 20-30% higher CPA; 10% = near-broad
✓500 high-LTV customers outperform 5,000 random leads as seed audiences
✓Stack 1%, 1-3%, 3-5% in separate ad sets for controlled scaling

1% vs 5% vs 10%

1%: 2.1M in US, best CPA. 5%: 10.5M, 20-30% higher CPA. 10%: 21M, 40-60% higher CPA. Start 1%, go broader only when saturated (frequency above 3.0).

Source Quality

Best: highest-LTV buyers > all buyers > high-intent leads > all leads > visitors. 500 top buyers beat 10K email lists. Min 100 people, 1K+ for stability.

Stacking Technique

Three ad sets: 1%, 1-3% (excludes 1%), 3-5% (excludes 1-3%). Budget: 50/30/20 split. Shift budget as tiers saturate. Scales without diluting best audience.

Data From AutoAdy

•1% lookalikes deliver 35% lower CPA than 5% across AutoAdy accounts
•Top 20% LTV sources produce 42% better results than all-customer sources
•AutoAdy detects saturation and recommends next-tier expansion

Sources

Meta Business Help Center: About Lookalike Audiences
Revealbot Lookalike Performance Study 2025 (n=8,000 campaigns)

FAQ

Common questions

How to A/B Test Facebook Ads

Ready to try AI Meta Ads optimization that actually works?

Start free. Connect your Meta account. See your wasted spend in 60 seconds.

Start free See how it works