> Quick answer: Use a Custom Audience of 1,000 to 5,000 high-value customers as your seed. Build lookalikes at 1%, 3%, and 5%. Test each as a separate ad set. The 1% delivers the closest match. The 5% delivers more scale. Pick the winner with data, not guesses.
Lookalike audiences are one of the most effective targeting tools on Facebook. They let you reach new people who share real traits with your best customers. Done right, they can expand your reach without sacrificing relevance.
What is a Lookalike Audience?
Lookalike audiences take a group you already know and find millions of new people who resemble them.
Definition and core concept
A lookalike audience is a group of Facebook users who resemble your existing customers. You provide a "seed" group. Facebook finds new people who share similar behaviors, interests, and demographics. You reach a fresh audience with proven buying signals built in.
How Facebook builds lookalike audiences
Per the Meta Business Help Center, Facebook uses demographics, interests, and behaviors from your source audience to find similar people. The system builds the audience within 1 to 6 hours of creation. Once active in a campaign, it refreshes every 3 days automatically to stay current.
When to use lookalikes in your strategy
Use lookalikes when interest-based targeting stops scaling. Use them when you have a proven customer base and want more of the same. They work best when your seed reflects your highest-value buyers, not just any site visitor or Page follower.
Choose Your Source Audience
The quality of your seed audience determines the quality of your lookalike.
Types of source audiences you can use
Three seed sources work for lookalike creation. First, Custom Audiences built from customer lists, website traffic, or app activity. Second, Page fans from your Facebook Page. Third, conversions from a past campaign or ad set. Each source has different data density and different strengths. Customer list seeds tend to produce the strongest results.
Minimum size requirements
Meta requires a minimum of 100 people in the source audience. But 100 is a floor, not a target. Meta recommends 1,000 to 5,000 people for better predictive results. More diversity in your seed means more signal for Facebook's matching model. Below 1,000 people, the model has too little to work with.
Quality matters more than quantity
A list of 2,000 recent purchasers beats a list of 20,000 cold leads every time. Use audiences that represent your actual best customers. Recent converters, high-spend buyers, and loyal repeat customers all make strong seeds. Broad or low-intent audiences weaken the model and produce weaker matches.
Best practices for source selection
Upload a customer list from your CRM if you have one. Segment by high lifetime value where possible. If you use website traffic, filter to purchase or checkout events, not homepage visits. The more precisely your seed reflects real buyers, the stronger the lookalike Facebook builds from it.
Create a Lookalike Audience in Ads Manager
Creating a lookalike takes under five minutes in Ads Manager.
Step-by-step creation process
- Go to Ads Manager and open the Audiences section.
- Click Create Audience and select Lookalike Audience.
- Choose your source audience from the dropdown.
- Select the country or countries you want to target.
- Set your audience size using the percentage slider.
- Click Create Audience. Facebook populates the audience within 1 to 6 hours.
Selecting similarity percentage
The percentage slider controls how closely the lookalike matches your source. Per the Meta Business Help Center, lower percentages (1-2%) more closely match your source audience. Higher percentages (5-10%) create larger, broader audiences with less precision. Start at 1%. Expand once you know what performs.
Geographic targeting options
You choose the country or countries at creation time. For multi-country targeting, Meta recommends at least 100 seed audience members per country. Distribution across countries will not always be even. Facebook targets each location independently, finding the best match in each market separately.
Creating variations at different percentages
Build at least two or three lookalikes at different percentages. A 1%, 3%, and 5% version covers precision, mid-range, and scale. Run each as a separate ad set with identical creatives. Compare cost per result across each. Real data beats assumptions every time.
Lookalike Audiences from Campaign Conversions
Conversion-based lookalikes draw directly from people who took action inside Facebook.
Why use conversion-based lookalikes
No external list required. Facebook tracks who converted in your past campaigns and builds the lookalike from that group. This is especially useful if you don't have a strong CRM list but have run conversion campaigns for several months.
Minimum conversion requirements
You need at least 100 unique conversions to use a campaign as a seed. Meta recommends 200 or more for stronger results. Below 100, the data is too thin to produce a reliable audience. Check your conversion volume before choosing this seed type.
How Facebook uses conversion data
Per Meta's developer documentation, Facebook uses up to 180 days of past conversion data to build conversion-based lookalike audiences. Recency and volume both matter. More recent conversions produce more relevant matches. If your campaign ran years ago, a customer list seed is probably more reliable.
Key Settings and Best Practices
Small setup decisions create big differences in targeting quality.
Percentage selection for your goals
Start at 1% for prospecting campaigns where relevance is the priority. Move to 3-5% when scaling ad spend or testing new markets. Use 5-10% for broad awareness campaigns where volume matters more than precision. Match the percentage to your current goal, not a default.
Avoiding audience overlap
Running multiple lookalike ad sets at the same time creates overlap. That means you're competing against yourself in the auction. Use the audience overlap tool in Ads Manager to check. Separate ad sets by percentage range. Consider excluding lower-percentage lookalikes from higher-percentage ad sets to reduce internal competition.
Testing multiple variations
Never run a single lookalike and call it done. Test at least two percentage variations simultaneously. Test different seed audiences too. A purchase-based seed versus a high-lifetime-value seed may produce noticeably different results worth comparing.
Monitoring performance
Check cost per result by ad set at least once a week. If a 1% consistently outperforms a 5%, precision matters more for your product. If both perform similarly, scale the broader one for more reach. Track results over time and refresh your seed audiences every few months to keep the data current.
Common Mistakes to Avoid
Most lookalike problems trace back to a few predictable errors.
Using too small a source audience
A 200-person list meets the minimum but limits Facebook's ability to find diverse patterns. Push toward 1,000 to 5,000 people before building. If you don't have that yet, grow your Custom Audience first through retargeting campaigns or customer uploads.
Over-narrowing with additional targeting
Stacking detailed interests on top of a 1% lookalike creates a tiny audience. The lookalike already contains the signal. Let it do the work. Age, gender, and broad location are usually enough. Extra targeting layers often hurt delivery without improving relevance.
Picking the wrong conversion window
If you track 7-day click conversions, Facebook seeds the lookalike from 7-day click converters. Match the conversion window to your typical buying cycle. A mismatch produces a seed that doesn't reflect your real customers and weakens the model from the start.
Neglecting to test different percentages
Many advertisers set one lookalike percentage and never revisit it. Run 1%, 3%, and 5% versions in parallel. Review performance after two weeks of significant spend. Invest budget in what the data shows, not what feels right.
> Heads up: Beginning September 2, 2025, Meta will restrict Custom Audiences and Lookalike Audiences that imply specific health conditions or financial status. Review your source audiences to make sure they don't fall into those categories before that date.
Scale Faster with Coinis
How Coinis streamlines audience and campaign setup
Building the audience in Meta is only half the job. You still need compelling creatives and sharp copy to make that audience convert. The Campaign Launcher in Coinis walks you through audience targeting, creative selection, and budget setup in one guided flow. The audience targeting step keeps everything organized without bouncing between Ads Manager tabs. Your Brand Profile feeds the creative and copy generation automatically, so every ad reflects your brand voice from the first draft. Once your lookalike is ready in Meta, use Coinis to launch a full campaign against it without the usual back-and-forth setup time.
Or let Coinis do it.
From a product URL to a live Meta campaign. AI-generated creatives. On-brand copy. Direct publish to Facebook and Instagram. Real performance reporting. All in one platform.
Start free. Upgrade when you're ready.
15 AI tokens a month. No credit card.
Frequently Asked Questions
How many people do I need in my source audience to create a lookalike?
Meta requires a minimum of 100 people, but recommends 1,000 to 5,000 for better results. Smaller seeds limit the diversity of patterns Facebook can detect, which weakens the match quality.
What percentage should I use for my Facebook Lookalike Audience?
Start at 1-2% for precision targeting closest to your source audience. Use 5-10% for broader reach and scale. Test multiple percentages in separate ad sets and invest in the one that delivers the best cost per result.
How long does it take for a Facebook Lookalike Audience to populate?
Between 1 and 6 hours, per Meta's documentation. Once the audience is active in a campaign, it refreshes automatically every 3 days.
Can I use campaign conversions instead of a customer list as my seed?
Yes. You need at least 100 unique conversions, with 200 or more recommended for stronger results. Facebook uses up to 180 days of past conversion data to build the audience.