Meta Lookalike Audiences · Creation, Seeding & Scaling

A Lookalike Audience is created by Meta's AI analysing the demographic and behavioural characteristics of a seed audience (a Custom Audience you specify) and finding new users who share those patterns. The result is a new audience of people who have never interacted with your brand but statistically resemble those who have.

Meta's similarity model analyses hundreds of signals: what content users engage with, pages they follow, websites they visit, purchases they make, videos they watch, and demographic characteristics. It creates a weighted profile of your seed audience's characteristics and finds the closest matches in the target country's user population.

The seed audience is the most critical variable in Lookalike performance. Meta finds people like your seed — so the more your seed represents your highest-value customers, the more valuable the resulting Lookalike will be.

Seed audiences ranked by typical quality

Seed quality matters more than seed size — above the 1,000-person threshold, a list of 2,000 high-value purchasers produces a better Lookalike than a list of 50,000 all-time website visitors. Meta's model can identify patterns from a smaller, cleaner seed more accurately than from a large, noisy one.

When creating a Lookalike, you select a percentage of the target country's Facebook users to include — from 1% (most similar) to 10% (least similar but largest). In a country with 30 million Facebook users, a 1% Lookalike contains approximately 300,000 people; a 10% Lookalike contains 3 million.

Lookalike Audiences are created in Ads Manager under Audiences → Create Audience → Lookalike Audience:

1. Select the source (your Custom Audience seed)
2. Select the target country (where you want to find similar users)
3. Select the audience size (1%–10%)
4. Create Audience — Meta typically takes 1–6 hours to populate a new Lookalike

You can create multiple Lookalike sizes from the same seed simultaneously — a 1%, 2%, and 3% from your purchaser list — for testing. Lookalike Audiences automatically update as the seed audience updates; you do not need to manually refresh them when new purchasers are added to the seed list.

Stacked lookalikes use different percentage ranges as separate Ad Sets within the same campaign — allowing Meta's CBO to allocate budget to the best-performing size:

Campaign (CBO): Prospecting — Lookalike Test
  Ad Set 1: Purchasers LAL 1% (most similar)
  Ad Set 2: Purchasers LAL 1–3%
  Ad Set 3: Purchasers LAL 3–5%
  [Each Ad Set has its own creative; budget distributed by CBO based on performance]

When stacking, ensure audiences are mutually exclusive — a 1–3% lookalike should exclude the 1% lookalike to prevent overlap and internal auction competition. Use the Audience Overlap tool to verify before launching.

Advantage Lookalike is Meta's AI-expanded version of standard Lookalike targeting. When enabled, Meta can serve ads beyond the specified Lookalike percentage to additional users it predicts will respond well — similar to how Advantage+ Audience expands Core Audience targeting.

Advantage Lookalike is increasingly the default in Meta's recommendations. It can improve delivery and reduce CPM by giving Meta more audience flexibility, but reduces your control over who sees your ads. For performance campaigns with strong conversion data, Advantage Lookalike typically produces better ROAS than manually restricted Lookalikes because Meta's AI can identify patterns the percentage-based model misses.

Lookalike Audiences lose accuracy over time through audience saturation — you show ads to the same users repeatedly until they stop responding. Signs of Lookalike decay: rising CPM, falling CTR, rising CPA on campaigns that previously performed well.

Refreshing Lookalike strategy:

• Refresh the seed audience regularly — a purchaser list with 6 months of new buyers creates a stronger signal than a static list from 2 years ago
• Rotate creative to re-engage saturated audiences — new creative can revive performance from an otherwise fatigued Lookalike
• Expand to larger Lookalike percentages when 1% saturates — moving from 1% to 2–3% reaches new users without abandoning the Lookalike approach
• Test Lookalikes from different seed audiences — a video viewer Lookalike reaches different users than a purchaser Lookalike at the same percentage

Authentic Sources