Lookalike audience: definition and creation
Updated on February 22, 2026
Quick definition
A lookalike audience is an ad audience segment created algorithmically from a source audience (your best customers, your converters) to identify new users with similar characteristics. Lookalike audiences let you extend a campaign's reach to qualified prospects without having their direct contact details.
How it works
To create a lookalike audience, you upload a source audience to the ad platform: a list of customer emails, a list of converting visitors, or a retargeting audience from a pixel.
The platform's algorithm (Meta, Google, TikTok, LinkedIn) analyses common characteristics of source-audience members and scans its user base to identify statistically similar profiles.
On Meta Ads, you can create lookalike audiences from 1% to 10% of a target country's population:
- 1% lookalike: very similar to the source audience, more restricted
- 10% lookalike: broader but less precise
Lookalike quality depends directly on the source audience: ideally at least 1,000 homogeneous profiles. The best sources are high-value customers (high LTV) or long-term active users.
Why it matters
Lookalike audiences are one of the most effective acquisition tools because they allow you to prospect at scale by relying on the characteristics of your best existing customers, without manual interest- or keyword-based targeting.
They are particularly powerful in campaign scaling phases: once an offer converts well on a retargeting audience, the lookalike lets you test that offer on a cold but qualified audience.
How to improve or use it
- 1Segment your source audience by customer value: build a lookalike based on your top-20%-LTV customers rather than all customers.
- 2Refresh your source audiences regularly (every 30 to 90 days) to include new converters.
- 3Test multiple source audiences in parallel (buyers, converted free trials, engaged newsletter subscribers).
- 4Compare the CPAs obtained from your different lookalike sources.
- 5Combine lookalike with interest targeting to refine qualification further.
With Sublim
Sublim helps you build high-quality source audiences for your lookalikes by precisely identifying the behaviours and journeys of high-value users. By exporting these segments via the Sublim API, you feed your ad platforms with reliable first-party data, with no third-party cookie dependency.
Frequently asked questions
What is the minimum size for a lookalike source audience?
Most platforms recommend a source audience of at least 1,000 to 2,000 people to generate a statistically reliable lookalike. Below 100 profiles, algorithm precision is very limited. Meta recommends between 1,000 and 50,000 people for optimal performance.
Do lookalike audiences work on every platform?
Yes, most major ad platforms offer lookalike audiences: Meta Ads (lookalike audiences), Google Ads (similar audiences), LinkedIn Ads (lookalike audiences), TikTok Ads (lookalike audiences) and Pinterest Ads. Quality varies based on user-base size and the quality of behavioural signals available on each platform.
How do I measure the performance of a lookalike audience?
Measure standard performance KPIs: CPA (cost per acquisition), ROAS (return on ad spend), post-click conversion rate and LTV of acquired customers. Compare your lookalike's results with your cold-audience interest-targeted campaigns and your retargeting campaigns to position the lookalike in your mix.
Related terms
Retargeting is an advertising technique that targets users who have al…
Audience analytics covers all the data and analyses related to the cha…
First-party data is all the data collected directly by an organisation…
LTV (Lifetime Value), or customer lifetime value, is the total revenue…