A lookalike audience is generated by an ad platform from a seed audience you supply. The platform looks for users with similar characteristics and behavior and lets you target them. Meta popularized the format. Other platforms have moved away from manually built lookalikes toward automated audience expansion, so what is available depends on the platform and the year.
Where they exist today
Availability has changed a lot, so check each platform’s current documentation.
- Meta: lookalike audiences are still offered, though Meta increasingly encourages Advantage+ audience, which treats your inputs as suggestions.
- Google Ads: similar audiences were retired in 2023 and replaced by optimized targeting and audience expansion.
- LinkedIn: lookalike audiences were retired in 2024; LinkedIn offers predictive audiences built from a seed list instead.
- Reddit, TikTok and others offer their own versions with varying controls.
Picking a good seed
The seed decides what the platform looks for. A seed of all website visitors produces a lookalike of casual browsers. A seed of paying customers, or customers on a higher plan, gives the model a stronger signal. Seeds need to be large enough for the platform to work with, and minimum sizes vary.
Fit for B2B
Consumer platforms model consumer behavior, which does not always map to job function or company size. B2B teams often find lookalikes useful for broad top-of-funnel reach and less precise than job-title targeting on LinkedIn. Judge them by qualified pipeline, not cost per click.
Example
Customer seed versus lead seed
Hypothetical company Dovecraft builds two Meta lookalikes: one seeded with all leads, one seeded only with customers who stayed beyond their first year. The second audience costs more per lead but produces more sales-accepted leads, so it gets the budget.
Common mistakes
Seeding with junk
If your lead list includes spam and students, the lookalike will find more of them. Clean the seed first.
Questions
You upload hashed data under the platform’s terms, and you never see who is in the resulting audience. You still need a lawful basis and appropriate notice for the uploaded data.
