A Lookalike Audience is an expansion method where Meta analyzes a source audience and automatically finds new users who show similar behavior and characteristics. It’s one of the most structured ways to reach a cold audience.
The quality of this method depends on the input. The more clearly the source audience represents a specific behavior — like purchasers — the more precise the resulting lookalike audience will be.
Meta’s Lookalike Audience documentation states that the source (seed) audience can be an existing custom audience with at least 100 members, campaign conversions, or page likes. The same documentation explains that the similarity ratio can be manually adjusted from 1% to 20% in 1% increments, where 1% is the most precise and narrowest match, and the audience grows larger — with less precision — as the percentage increases.
How Does Lookalike Work?
Meta analyzes the common characteristics of the users in the source audience and finds users showing similar traits in the same country.
Minimum source audience size is a deciding factor
The source audience needs at least 100 members. Audiences below that number may not give Meta enough to detect a meaningful common pattern; in that case, the resulting lookalike can end up closer to a random group than a genuine pattern.
Geographic scope is country-based
A lookalike audience is created for a specific country. If the source audience doesn’t have enough members in that country, granting international source permission may allow members from other countries to be included; this setting can change over time, so current options should be checked at setup.
How Should the Source Audience Be Chosen?
The quality of the source audience directly determines the quality of the resulting lookalike audience.
Purchasers are usually the strongest source
A source audience made up of users who purchased or completed a high-value conversion — not just site visitors — generally produces a more precise lookalike; the real difference shows up in the quality of the source. For example, a lookalike built from the last 500 purchasers typically produces a higher conversion rate than one built from the last 500 visitors, because Meta learns the common traits of actual purchase behavior from the first group.
A small but high-quality audience can beat a large but weak one
A broad but low-intent source audience (like all site visitors) can produce a less precise lookalike than a narrow but high-intent source (purchasers).
How Should You Choose the Similarity Percentage?
The percentage you choose determines the balance between audience size and precision.
A lower percentage produces a narrow, precise audience
1% is the closest and narrowest match to the source audience. For small-budget campaigns, this lower range can help direct the budget toward the most precise users; the real risk is choosing a wide percentage when the budget is small.
Gradual expansion can be tested
Starting at 1% and expanding to 3% and 5% based on performance makes it possible to build a gradual balance between precision and reach; this incremental path is safer than jumping straight to a wide percentage.
Does Lookalike Replace Broad Targeting?
Meta’s automatic broad targeting options (like Advantage+ audience) work on different logic than lookalike; neither directly replaces the other.
Over-restricting during the learning phase is risky
Limiting the audience to a very narrow lookalike before the campaign has gathered enough data can slow down the system’s learning. In some cases, starting with broad targeting and narrowing over time can be a more measured approach; the real mistake is repeatedly changing the audience before learning has completed.
Pre-Launch Check
Before putting a Lookalike audience into use:
- Does the source audience have at least 100 members, and does it represent a clear behavior?
- Does the similarity percentage match the size of the budget?
- Does the geographic scope align with the market the campaign is targeting?
- Is the source audience current, or is it based on an outdated dataset?
Common Mistakes
The most common mistake is using all site visitors as the source audience without separating out a high-intent behavior like purchasing. In that case, the resulting lookalike can end up resembling users who visited the site once, rather than people genuinely interested in the brand.
Summary
A Lookalike Audience is an expansion method that analyzes a source audience to find similar users. The quality of the source audience and the similarity percentage you choose determine both the size and precision of the resulting audience. A lower percentage produces a narrower, more precise audience; a higher percentage produces a broader, less precise one.