A lookalike audience is a group of people Meta builds because they share characteristics with an audience you already have, such as your customers, your website visitors or people who have engaged with your Instagram account. You supply the starting list, called the source or seed audience, and Meta finds others in your chosen country who resemble it.
How a lookalike audience works
The seed is a custom audience: an uploaded customer list, website visitors tracked by the Meta Pixel, people who watched your videos, or people who opened a lead form. At the time of writing (October 2026), Meta needs at least 100 people from one country in the seed and suggests that larger seeds, in the low thousands or more, work better.
You then choose a location, such as the United Kingdom, and a size from 1% to 10% of that location’s population. A 1% lookalike is the closest match to your seed and the smallest. Larger percentages reach more people who resemble the seed less closely. If your customer list includes a value for each customer, a value-based lookalike weights its search towards people who resemble your higher spenders.
How lookalikes behave depends on the rest of the ad set. With Advantage+ audience switched on, Meta treats your lookalike as a suggestion and can deliver beyond it if it finds better results elsewhere. With it switched off, the lookalike acts as a firmer boundary.
Why it matters
For years, lookalikes were the standard way to find new customers on Facebook. Today Meta’s delivery system often finds buyers just as well through broad targeting, so lookalikes are no longer automatically the best choice. They still earn their place in some accounts, particularly where the seed is high quality and the product is niche, and they are worth testing rather than assuming either way.
For UK businesses, customer list uploads carry data protection duties. Meta hashes the list before matching, but you are still sharing personal data for advertising, so under UK GDPR you need a lawful basis and a privacy notice that tells customers you do this. The ICO’s direct marketing guidance covers custom audiences, and it is worth reading before your first upload. Website-visitor seeds depend on the pixel, which on a compliant site only fires after cookie consent, so they will be smaller than your real visitor numbers.
Common mistakes
- Seeding from all website visitors, including job seekers and competitors, instead of buyers.
- Using a seed so small that the lookalike is close to random.
- Running 1%, 2%, 3% and 5% lookalikes in separate ad sets, so they compete for the same people.
- Leaving existing customers in prospecting campaigns because no audience exclusion was added.
- Uploading a customer list without checking the privacy notice or lawful basis.
- Letting customer lists go stale for a year.
How to act on it
Start with the seed. Export your best customers from your ecommerce platform or CRM: repeat buyers, high-value clients or customers from your most profitable service. Include customer value if you have it. Refresh the list monthly, or connect it so it updates automatically.
Build a 1% UK lookalike and test it against broad targeting with the same budget, creative and dates, ideally using Meta’s A/B test tool so the comparison is fair. Exclude current customers from both. Judge on cost per purchase or lead over a few weeks, not on click-through rate.
Before uploading anything, read your privacy notice and confirm it covers advertising use of customer data. Setting up seeds, exclusions and fair tests is routine work in Facebook and Instagram ads management.