Finding new customers is the hardest part of advertising. Lookalikes use the platform's data to find people who share patterns with your existing customers, giving prospecting campaigns a strong starting point. The Academy's glossary explains lookalike audiences.
How lookalikes work
- Choose a seed: a custom audience such as customers or website purchasers.
- Choose a location: the country or region to find people in.
- Choose a size: a percentage of the population, from closest match to broader.
- The platform builds the audience from people similar to the seed.
Meta, TikTok, LinkedIn and Pinterest offer lookalike or similar audience features, with different names and options.
Choosing a good seed
| Seed | Strength |
|---|---|
| Purchasers or high-value customers | Strongest signal of buying intent |
| Leads that became customers | Good for lead generation |
| Website visitors who added to cart | Useful when purchases are few |
| Video viewers or engagers | Larger but weaker signal |
| All website visitors | Broad and mixed quality |
Quality matters more than size, but very small seeds can produce weak lookalikes. Many advertisers aim for at least a few hundred to a few thousand people in the seed. The Academy's guide to custom audiences explains how to build seeds.
Value-based lookalikes
Some platforms let you upload customer value, such as total spend, so lookalikes focus on people similar to your most valuable customers rather than all customers. This can improve return on ad spend for businesses with varied order values.
Audience size
Smaller percentages, such as one percent of a country's population, are closest to the seed. Larger percentages reach more people with less similarity. Test a small and a larger lookalike to see which delivers better results at your budget.
Lookalikes versus broad targeting
Modern ad systems often find good customers with broad targeting, using conversion data from your pixel. In some accounts, broad targeting now matches or beats lookalikes, especially with strong tracking and enough budget. In others, lookalikes still help, particularly for new accounts or niche products. Test both side by side. Meta's Advantage+ audience options blend these approaches. The Academy's glossary explains Advantage+.
Using lookalikes in campaign structure
Place lookalikes in prospecting campaigns alongside or instead of interest targeting. Exclude existing customers and recent website visitors so prospecting budgets reach new people. Retarget the warm audiences separately. The Academy's guide to campaign structure shows a simple setup.
Creative for lookalike audiences
Lookalike audiences are cold: they do not know you yet. Use introductory creative such as a strong hook, a clear benefit and social proof. TikTok's documented organic signals are completion rate and rewatches, shares, comments, likes, interest signals from watch behaviour✓, and ads that hold attention from the first second help cold audiences engage.
Privacy considerations
Seeds built from customer lists or website data must be collected lawfully, with appropriate consent for tracking. The lookalike itself does not share your customers' identities with you, but the seed data must still be handled properly. The Academy's guide to custom audience GDPR explains what to check.
Refreshing lookalikes
Lookalikes based on customer lists need refreshed seeds as your customer base grows. Lookalikes based on pixel audiences update automatically. Review performance monthly and rebuild if results fade. When your product range or customer base changes significantly, such as after a rebrand or a move into a new market, create fresh seeds from the customers you now want more of rather than relying on older lists that reflect a different business.
Lookalikes on LinkedIn
LinkedIn's equivalent uses professional data, which suits B2B advertisers. Seeds from website visitors or contact lists help reach similar professionals, though LinkedIn's own targeting by job title, seniority and industry often does much of this work already, so test whether a lookalike adds anything before relying on it.
Measuring lookalike performance
Judge lookalikes on the result that matters, such as cost per purchase or cost per qualified lead, rather than clicks. Compare each lookalike against broad targeting over the same period with similar creative and budget. Small differences over a few days are usually noise; look for consistent gaps over two to four weeks before deciding. Record which seeds and sizes worked, because the best options can change as your customer base and tracking improve.
When lookalikes struggle
Lookalikes struggle when the seed is too small, too mixed or built from low-intent actions such as page likes. They also weaken when tracking is broken, because the platform cannot see who converts. If results are poor, check tracking first, then try a higher-quality seed or switch to broad targeting with strong creative.
A worked example
An online tea shop creates a one percent lookalike from its last six months of purchasers and a three percent lookalike from the same seed. Both run alongside a broad audience in one prospecting campaign, excluding existing customers. After three weeks, the one percent lookalike has the lowest cost per purchase, the broad audience is close behind and the three percent lookalike trails. The shop keeps the one percent lookalike and broad audience, and refreshes the seed monthly.
Common mistakes
- Seeds built from low-quality actions such as page likes.
- Seeds too small to model well.
- Not excluding existing customers.
- Never testing lookalikes against broad targeting.
- Using retargeting-style creative for cold lookalikes.
Learning targeting
The Academy's paid social module in the third level covers lookalikes, custom audiences, broad targeting and testing. The free assessment shows where you would start.
