Lookalike Audience
A lookalike audience is an audience created by an ad platform (Meta, Google, TikTok) from an original base of customers or interactors, finding people with similar profiles. In 2026, it is one of the most efficient ways to scale campaigns without losing quality, especially on Meta Ads.
What is a Lookalike Audience?
Lookalike audience is a type of audience automatically created by media platforms from a source base provided by the advertiser. The platform's intelligence identifies common patterns (interests, behavior, demographics, actions) and finds other people with similar profiles.
It is the most scalable way to extend performance beyond warm audiences without falling into overly generic targeting.
How it works
- You provide a source base: customer list, site visitors, leads, buyers
- The platform crosses it with its own data (interests, behavior, social circle)
- Creates an audience of similar people in the chosen country
- You use that audience in a campaign
Ideal size
- Lookalike 1%: the most similar — small audience, high quality
- Lookalike 2% to 3%: good balance
- Lookalike 5% to 10%: broad audience, lower quality — useful to scale a lot
2026 recommendation: start at 1% to 2%, scale gradually up to 5%. Above 5% quality drops faster.
Which bases to use
From BEST to less good:
- Buyers from the last 30/60 days (top tier)
- High-value buyers (LTV > X)
- Qualified leads
- Visitors who started checkout
- Site visitors
- Instagram/Facebook engagers
- Newsletter subscribers
The deeper in the funnel the base, the better the lookalike.
Practical example
A fashion e-commerce with 14,000 buyers in the last 90 days creates:
- LAL 1% (Buyers 90d): 2.3M people — CPA R$ 42
- LAL 3% (Buyers 90d): 6.8M people — CPA R$ 51
- LAL 1% (Visitors 30d): 2.3M — CPA R$ 68
- Interests (fashion + age): 4.1M — CPA R$ 89
The 1% buyer LAL performs 30% better than visitor LAL and almost 50% better than interests. As it scales (3%), CPA rises slightly but stays below interest — clear advantage.
Common mistakes
- Lookalike with a base under 500 people (poor quality — Meta recommends 1,000+)
- Lookalike of visitors instead of buyers when volume is available
- Global lookalike (choosing 10 countries). Always create by country
- Not refreshing bases. A lookalike of buyers from 2 years ago does not reflect today's audience
- Combining LAL with too many filters (age, interest) — defeats the point of starting from real behavior
Related terms
Want to see this in action?
Mais Social shows all your metrics in one place.
Try 14 days free