A lookalike audience is a group the ad platform builds by finding people who resemble your existing customers or leads, used to reach new prospects who behave like the people already buying from you.
A lookalike audience is a group the ad platform builds by finding people who resemble your existing customers or leads, used to reach new prospects who behave like the people already buying from you. You hand the platform a source list, say your past buyers or your email subscribers, and it goes looking across its user base for people who share the traits of that group.
The mechanics matter, because there is a common misconception that a lookalike is people who look like your customers on paper, same age, same suburb, same job. It is not really that. The platform models the behavioural and interest signals it holds on your source group, then finds others who pattern-match on those signals. You rarely see which signals it used, that part is a black box. What you control is the quality of the seed list and how tightly you want the match, usually expressed as a percentage of the population, where a smaller percentage means a closer, narrower match and a larger one means broader reach with looser resemblance.
For a business owner, the appeal is straightforward. It is a way to scale past the people who already know you without starting from scratch on cold targeting. The seed is the whole game, though. A lookalike built from a list of your best, highest-value customers will nearly always outperform one built from every lead you ever collected, because the platform is only ever as good as the example you give it. Feed it a messy or low-intent source and you get a bigger version of the same problem. It sits naturally alongside retargeting, which works your warm audience, while lookalikes go after fresh people who resemble them.
One honest caveat: a lookalike is a prospecting tool, not a magic audience. It gets you in front of plausible new people, but they have never heard of you, so the offer, the creative and the landing experience still have to do the convincing. I judge them on cost per acquisition, not on how clever the targeting sounds. This is core to how I run paid social.
Key points
- A lookalike audience finds new people who resemble your existing customers, built from a source list you provide.
- It matches on behavioural signals the platform holds, not just surface facts like age or location.
- The quality of the seed list decides everything: best-customer sources beat all-leads sources.
- A tighter match percentage means a narrower, closer audience; a broader one trades resemblance for reach.
- It is a prospecting tool for cold audiences, so the offer and creative still have to do the persuading.
- Pairs naturally with retargeting: lookalikes chase fresh prospects, retargeting works the warm ones.
Frequently asked questions
Common questions about lookalike audience.
Most platforms will let you build one from a fairly small seed. LinkedIn is the exception, it retired Lookalike Audiences in 2024 in favour of Predictive Audiences, but on the platforms that still offer it, small and technically-allowed are not the same as good. The model has more to learn from when the source is a few thousand people rather than a few hundred, and the resemblance it finds is more reliable. Just as important as size is quality: I would rather build a lookalike from a tight list of proven, high-value customers than a huge list of every low-intent lead. Give the platform a strong example and it finds strong matches.
No, and it helps to keep them separate in your head. Retargeting shows ads to people who already know you, past site visitors or engagers, a warm audience. A lookalike goes the other way: it finds brand-new people who have never heard of you but resemble your customers. They complement each other in a campaign, retargeting to convert the warm, lookalikes to fill the top with fresh prospects, but they are doing different jobs and you should measure them differently.
The percentage is how closely the audience matches your seed, so a 1% lookalike is the tightest resemblance and a 10% is much broader but looser. There is no universally correct answer. A tighter match usually converts better per person but runs out of reach quickly, while a broader one gives you scale at the cost of relevance. I tend to start tight, prove the economics work at a sensible cost per acquisition, then widen only if the numbers still hold. Widening for the sake of volume before you have proven the tight audience is a common way to waste budget.
Want to reach the right people on social?
Tell me who you are trying to reach and I will show you which platforms make sense.
Start a conversation