Facebook Lookalike Audiences in 2026: How to Build Them and When They Still Beat Broad Targeting

A Facebook lookalike audience is an audience Meta builds by finding people who resemble a source audience you choose, usually your best customers, sized from 1% to 10% of a country's population. In 2026 I build lookalikes from my highest value customers, with at least 1,000 people in the source, and start with 1% for the closest match. I use them mostly as a suggestion inside Advantage+ audience or as a test against broad targeting.

Last updated: 20 October 2026.

I've managed $35M+ in Meta ad spend across 50+ brands, and lookalike audiences are where performance marketing on Meta first showed me its power. Imagine taking your absolute best customers and telling Meta to find you more people exactly like them, which is what a lookalike does. What's changed is how much of that work the algorithm now does on its own.

This guide covers custom audiences, which every lookalike starts from, and how to choose the source and the size. It also shows how to create a lookalike step by step and how lookalikes fit into Meta's targeting in 2026. For the wider picture, read my guide to Facebook ads targeting in 2026 alongside this one.

Custom Audiences: Where Every Lookalike Starts

A lookalike always starts from a custom audience, so it helps to know the types. Custom audiences are people who already interacted with your business, and Meta lets you build them from almost anything they do.

Engagement audiences are often underrated. Someone who watched 75% of your video is valuable, someone who messaged your Page with questions is even more valuable, and someone who filled out a lead form but didn't convert is pure gold.

Customer lists are powerful too, as long as you don't just dump your whole database into Meta. Segment them first, so your high value customers, your repeat buyers and the newsletter subscribers who open every email can each become their own audience, which makes them far better lookalike sources. My guide to retargeting and catalog ads shows how I use custom audiences for warm campaigns.

How Lookalike Audiences Work in 2026

Meta compares the people in your source audience with everyone else in the country you choose, looking at their behaviors and characteristics, and builds a new audience of the most similar people. That's still how lookalikes work, and they're still useful.

What changed is everything around them. Today the algorithm finds lookalike people inside a campaign on its own. It studies who completes your optimization event and who responds to your creative, then shows the ad to more people like them. That's why broad targeting with a strong creative and the right conversion event often matches or beats a lookalike, and why Meta now starts new campaigns with Advantage+ audience.

When you use Advantage+ audience, you can still add a lookalike as an audience suggestion. Meta then delivers your ads to that audience first and expands beyond it as it gets more events and learnings. That makes a lookalike a useful starting signal for a new campaign or a young account.

Choosing the Source Audience

Here's where most advertisers go wrong. They create lookalikes from all their website visitors or their entire customer list, and the better your source audience, the better your lookalike will perform.

You need at least 1,000 people in your source audience for a lookalike to work well. If you don't have 1,000 customers yet, start with other targeting, like broad audiences or a few interests as suggestions, and come back to lookalikes once your data has grown.

Value based lookalikes

If your customer list includes a value for each customer, like their total spend, you can build a value based lookalike. Meta then gives more weight to people who resemble your most valuable customers, which suits brands with very different order values or long term customers worth far more than one time buyers.

Lookalike Audience Sizes: 1% vs 5% vs 10%

Meta lets you choose a size from 1% to 10% of a country's population. The 1% lookalike is your gold mine, because those are the people who most closely match your best customers. As you expand to 2%, 5% or 10%, you reach more people with less similarity.

Country size matters as well. In a small country, a 1% lookalike may be too small to spend a meaningful budget on, so you'll move to larger sizes sooner. In the US, a 1% lookalike already contains millions of people.

The fair test is your lookalike against broad targeting or Advantage+ audience, with the same creative, the same optimization event and the same budget. Let both run long enough to get meaningful data, and keep the one that delivers better results in your shop or CRM, not just in Ads Manager.

How to Create a Facebook Lookalike Audience

1. Open Audiences in Meta Ads Manager or in your business portfolio.
2. Click Create audience and choose Lookalike audience.
3. Select your source. Pick the best source you have. If you haven't uploaded a customer list yet, create it first as a custom audience, with columns like email, phone and country, which Meta matches in hashed form.
4. Choose the location. Lookalikes are built per country, because Meta finds similar people within that country's population, so build separate ones for each market.
5. Choose the size, starting with 1%, or create 1%, 3% and 5% versions from the same source to test.
6. Create the audience and wait until it's ready, which can take from a few minutes to several hours.
7. Use it in your ad set, either as an audience suggestion inside Advantage+ audience, or as a fixed audience if you deliberately switch to original audience options for a test.

Common errors are easy to fix. A source that's too small won't build, so aim for 1,000+ people from one country. An audience that shows a size of zero usually means the ad set targets a different country from the lookalike.

When Lookalikes Are Still the Right Tool

Lookalikes earn their place in a few situations.

• A new campaign or a young account. A lookalike from your best customers gives the algorithm a good starting point while it collects conversion data.
• High value or niche products. If you sell an expensive product to an unusual customer group, a 1% lookalike from your top customers can keep delivery close to the right people.
• Lead generation with a quality problem. A lookalike built from leads that became customers can be a better suggestion than any interest, especially together with a qualified lead custom conversion.
• Testing new markets. When you expand into a new country, a lookalike of your existing customers is a sensible first audience to test next to broad.

You can also test lookalikes the other way round. In campaigns where you can exclude custom audiences, some advertisers build a lookalike of disqualified leads or refunded orders and exclude it. That's worth testing if lead quality is your main problem, as long as you measure the effect on qualified leads in your CRM.

Common Lookalike Mistakes

• Using all website visitors as the source. Too noisy, so use purchasers, qualified leads or a segmented customer list instead.
• A source under 1,000 people. Meta may build it, but the signal is weak, so build up your data first.
• Stacking many lookalikes in one ad set. Several sizes from several sources in one ad set give Meta a fragmented pool, so pick one good source and test it.
• Never refreshing the source. Customer behavior changes, so rebuild your source every few months or after big changes in your products or markets.
• Expecting a lookalike to fix weak creative. Your creative and your optimization event still do most of the targeting work, so fix those first.

Frequently Asked Questions About Facebook Lookalike Audiences