What Is a Lookalike Audience on Meta?

A Lookalike Audience on Meta is a new audience the platform builds by analyzing an existing seed audience, typically a Custom Audience, and finding new users within a chosen country who share similar characteristics and behavior patterns, letting an advertiser expand reach toward people who resemble its best existing customers.

How Meta Builds One

An advertiser selects a seed source, commonly a high-value Custom Audience like past purchasers or top-tier customers, along with a target country and a similarity setting expressed as a percentage from 1% to 10%. Meta's algorithm then analyzes shared traits across the seed audience and searches its user base in that country for people matching those patterns.

The similarity percentage controls a direct trade-off between precision and scale: 1% produces a smaller audience most closely matched to the seed's characteristics, while 10% produces a much larger audience with a looser degree of similarity, most advertisers start narrow and expand only if the narrower audience's performance justifies scaling up.

Lookalike Audiences at Work

  • A skincare brand builds a 1% Lookalike Audience from its highest-lifetime-value customers, prioritizing precision over reach for a premium product line with a smaller ideal customer base.

  • A mass-market retailer builds both a 1% and a 5% Lookalike Audience from the same seed list, comparing conversion rate and CPA between the two to decide where the reach-versus-precision trade-off actually pays off for its budget.

  • A subscription business rebuilds its Lookalike Audience quarterly from an updated seed of its most recent high-retention customers, rather than relying on a static Lookalike built once at launch.

The Similarity Percentage, Explained

Meta doesn't expose the exact similarity-matching formula, but the resulting audience size scales roughly with the percentage chosen, a 1% Lookalike in a large market like India typically represents several million people, while a 10% Lookalike in the same market can represent tens of millions, the percentage is relative to that country's eligible Meta user base, not a fixed audience size.

How This Compares to Custom Audiences

A Lookalike Audience is built from a Custom Audience but serves a different purpose, a Custom Audience targets people already known to the business directly; a Lookalike Audience uses that same data only as a pattern to find entirely new people who've never interacted with the brand before, extending reach rather than re-engaging.

Choosing the Right Similarity Setting

Lookalike performance should always be read in light of seed quality, a Lookalike built from a broad, low-value seed (every website visitor) will typically underperform one built from a narrow, high-value seed (top-tier repeat customers), since the resulting audience can only be as relevant as the pattern it was modeled on.

Why Advertisers Rely on This So Heavily

Lookalike Audiences let a business scale prospecting beyond its existing reach while still grounding that expansion in real data about who its best customers actually are, generally producing more relevant, better-converting cold traffic than broad, purely interest-based targeting alone.

Frequently Asked Questions

What's the minimum seed size Meta recommends for a Lookalike?
Meta generally recommends at least 100 people in the seed source, though performance and reliability tend to improve meaningfully with larger, more well-defined seed audiences.
Should a Lookalike Audience be rebuilt periodically?
Yes, refreshing it from an updated, current seed (recent high-value customers rather than a static list from months ago) generally keeps it more accurately matched to who's actually converting now.
Can a Lookalike Audience target multiple countries at once?
Meta allows building Lookalikes for specific countries or regions, a business targeting multiple markets typically builds separate Lookalikes tailored to each one rather than one blended global audience.
Is a smaller percentage Lookalike always better?
Not universally, it's more precise but reaches fewer people, whether that trade-off is worth it depends on the account's budget, product margin, and how much volume is actually needed.
Does a Lookalike Audience overlap with the original seed audience?
No, by definition a Lookalike is built from people outside the seed audience, new prospects who merely resemble it, not the seed members themselves.

See NYX in action.

Book a demo