On this page
What a lookalike audience is
A lookalike audience is built by Meta, not by you. You give it a source, a group of people you already know are valuable, and it finds people in a chosen country who share patterns with that group. You never see who is in it, and it refreshes as the source changes.
It exists to solve a targeting problem: you know who buys but cannot describe them in interests. It helps most when you have a good source and a narrow product, and least when your offer already appeals to a wide audience. Meta’s newer Advantage+ audience options compete with it, so test rather than assume.
Lookalikes are not the same as interest targeting. Interests are labels Meta attaches to people. A lookalike is a statistical match to your own customers, which makes it more specific to your business and less dependent on whether Meta has a label for what you sell.
Choose the source audience carefully
The source decides everything. A list of high-value customers, repeat buyers or qualified leads teaches Meta to find more of them. A list of all website visitors or everyone who liked a post teaches it to find people who like clicking. Use the most valuable group you can build that has enough people.
Meta’s guidance is a source of at least 100 people from one country, and more is better, with a few thousand giving stronger results. Upload a customer list (email and phone, hashed on upload) or use your Pixel’s Purchase or Lead events. Value-based lookalikes, which weight customers by lifetime value, are available when you include a value column.
Clean the list before uploading. Remove refunds, fraud, test orders and people who bought once and never returned if you can tell them apart. Where you can, use a source with a value column, so a customer who spent a lot counts for more than one who spent a little.
Want this checked in your Meta account?
Reads your campaigns, ad sets, ads and pixel health. Every change waits for your approval.
Create the audience in Ads Manager
In Ads Manager, open Audiences, choose Create audience, then Lookalike audience. Pick your source, choose the country, and set the audience size. The slider runs from 1% to 10% of the country’s population, where 1% is the people most similar to the source.
You can create several sizes in one step and name them clearly, such as “LAL 1% US, purchasers 180d”. Allow up to a day for an audience to fill. Details on building the underlying custom audience are in the custom audience guide.
Pick the right country. The lookalike is built from people in that country only, so a US-wide audience built from a global customer list will lean toward your US customers. For several countries, create one lookalike per country rather than a mixed one.
Do not edit a lookalike once it is running. If you want a different size or source, create a new one and compare, so you can tell which change made the difference.
Pick a size and use it in an ad set
Start at 1%. It is the tightest match and usually the strongest, but in a small country it can be too small to spend a budget. If delivery stalls or cost climbs, add 1 to 3% rather than jumping to 10%, where the audience loosens toward the general population.
Use the lookalike in an ad set and exclude your existing customers and recent purchasers. Keep the ad set structure simple and give it enough budget to get results, as the learning phase guide describes. A lookalike is a targeting input, so it does not make weak creative work.
Check overlap before you run several lookalikes at once. Two 1% audiences built from similar sources overlap heavily, and the ad sets then bid against each other. The Audience overlap tool in Audiences shows how much.
Test it against broad and keep what wins
Meta’s delivery has gotten good at finding buyers without help, and many accounts now do as well with broad targeting and strong creative. Run the lookalike beside a broad ad set with the same ads and budget, and compare cost per qualified result after each has enough data.
If the lookalike wins, refresh the source every few months so it reflects current customers. If broad wins, put the effort into creative instead. Sprites can read your audiences and results across ad sets on a schedule and propose a budget or audience change as an approval card you edit and approve first.
Look at more than click cost. A lookalike that gets cheap clicks and few purchases is the wrong audience. Compare cost per purchase or per qualified lead, and check results by placement and age as well, since each can hide a different story.
Write down the result for next time. Whether lookalike or broad won, with the date, the source and the cost per result, is the evidence you will want when Meta changes its targeting options again.
Questions
What does lookalike audience mean?
It is a group of people Meta selects because they resemble a source audience you provide, such as your customers. The match is based on patterns in their profile and behavior, and Meta does not reveal who is in it.
How big should my source audience be?
Meta requires a minimum of 100 people from one country, and a source of a few thousand quality customers tends to work better. Quality still matters more than size.
Which audience is best for Facebook ads?
It depends on your data. Retargeting audiences convert best but are small. Lookalikes of your best customers help prospecting. Broad targeting with good creative often matches them. Test all three against the same ads.
Why is my CPC so high on Facebook?
High CPC usually comes from low ad relevance, a small or overlapping audience, or heavy competition. Check frequency, creative and audience size before blaming the lookalike. See Facebook ads high CPM.
Do lookalike audiences still work?
Yes, when the source is strong. They are one option among several, and Meta’s broad delivery can match them, so keep testing.