Lookalike Audiences (LAL)

What is Lookalike Audiences (LAL)?

Lookalike Audiences (LAL) are Meta-generated prospecting audiences built from a source Custom Audience — known as the "seed." Meta analyses the shared characteristics of people in your seed (demographics, interests, online behaviour) and identifies new users across its platform who share similar patterns. The size of the lookalike is expressed as a percentage of a country's total Meta user base: 1% is the most similar (and smallest), 10% is the broadest.

The quality of a Lookalike Audience is entirely dependent on the quality of its seed. A lookalike built from your top 1% of customers by lifetime value will consistently outperform one built from all purchasers, because Meta has a cleaner signal of exactly who your most profitable customers are.

How to Detect Issues with Lookalike Audiences (LAL)

• Lookalike ROAS declining over 60–90 days without creative or offer changes — Meta's model is working from an aging seed; the audience characteristics of your best customers may have evolved, particularly if you've changed product lines or customer acquisition channels
• 1% LAL significantly overlapping with 2–5% LALs at high spend — as you scale into broader lookalikes, overlap between tiers increases, driving self-competition in the auction
• Seed audience size below 1,000 matched users — Meta requires at minimum 100 people, but meaningful pattern recognition requires at least 1,000 to produce a high-quality lookalike; small seeds produce noisy audiences
• iOS 14+ degrading Pixel-based seed quality — if your seed is built from Pixel purchase events, Apple's ATT restrictions reduce the signal available; server-side Conversions API seeding produces more complete audiences
• Lookalike audience estimated size shrinking in Audience Manager — usually caused by seed audience shrinkage; the lookalike can only be as large as Meta's model allows based on the seed

How AdEvolver Handles Lookalike Audiences (LAL)

AdEvolver automates the monitoring and optimization of Lookalike Audience performance:

1. 24/7 Monitoring: AdEvolver tracks ROAS and CPA for each Lookalike Audience on a rolling quarterly basis, comparing current period performance to the same audience's historical baseline — identifying model aging before a manual review would catch it.
2. Slack Alerts: When a Lookalike Audience's ROAS drops more than 20% versus its 30-day average, a Slack notification names the audience, the performance delta, and when the seed was last refreshed — so you know whether a seed update is overdue.
3. One-Click Fixes: AdEvolver surfaces the seed audience's current estimated size and match rate. When the seed has shrunk below the quality threshold, the fix initiates a seed refresh workflow using your most recent customer data.

Related Articles

• How to Detect Audience Overlap in Meta Ads
• Scaling Meta Ads Without Tanking ROAS