How to Use Lookalike Audiences in Meta Ads (2026 Guide)

Lookalike audiences in Meta Ads let you reach new people who share characteristics with your existing customers, website visitors, or engaged fans. When built correctly, they’re one of the most efficient prospecting tools available – delivering higher conversion rates than broad interest targeting because Meta’s algorithm starts with a clear signal of who your best customers are. When built incorrectly, they produce expensive traffic that doesn’t convert.

This guide covers how to build lookalike audiences properly, how to choose the right source, and how to integrate them into your overall Meta Ads strategy.

How Lookalike Audiences Work

Meta creates a lookalike audience by analysing the characteristics of a source audience – your customer list, website visitors, video viewers, or any other custom audience – and finding people in the broader population who match those patterns. The algorithm looks at hundreds of data points: demographics, interests, purchasing behaviour, online activity, and more.

You control the size of the lookalike using a percentage: 1% lookalike = the 1% of the selected country’s population that most closely resembles your source. 10% = the broadest version, with less similarity but much larger reach.

Source Audience Quality Is Everything

The single most important variable in any lookalike audience is the quality of the source. A lookalike is only as good as the data you feed it. Here’s how to rank your available sources:

Best: Customer purchase list – email addresses or phone numbers of actual buyers, uploaded as a custom audience. This is the strongest signal because it reflects who actually converts.
Very good: High-value customers – if you can segment your customer list by LTV and upload only your top 20% of buyers, your lookalike will skew toward higher-value customers.
Good: Pixel-based purchase events – customers who have purchased on your website in the last 90-180 days. Requires a well-configured Meta Pixel.
Decent: Website visitors or cart abandoners – lower purchase intent than buyers, but still a valid signal for businesses with limited customer data.
Weaker: Video viewers, page engagers – broad engagement without purchase intent; use only if you have no stronger source available.

Minimum source audience size: Meta recommends at least 100 people, but in practice, 500-1,000+ produces significantly better results. Below 100, the algorithm has too little to work with.

Choosing the Right Lookalike Percentage

The 1%-10% range controls the trade-off between similarity and reach:

1% lookalike: Tightest match, smallest audience. Best for limited budgets or when your source audience is very distinctive and well-defined. Typically delivers the highest conversion rates but saturates quickly.
2-5% lookalike: The sweet spot for most campaigns. Larger audience, slightly lower similarity, but usually enough volume to run at meaningful scale without saturating quickly.
5-10% lookalike: Broadest. Useful when scaling spend and you’ve exhausted the smaller percentages. Think of it as a midpoint between a standard lookalike and Advantage+ broad targeting.

A common mistake: running all three percentages in the same campaign with overlapping audiences. Meta may serve the same person from multiple ad sets, inflating effective frequency. Use stacked percentages to create exclusive tiers: 1%, 1-3%, 3-7%.

How to Set Up a Lookalike Audience in Meta Ads Manager

Go to Audiences in Meta Ads Manager
Click Create Audience → Lookalike Audience
Select your source audience (custom audience, page, or pixel event)
Select the target location (the country or region you want to reach)
Choose your audience size (1%-10%)
Click Create Audience

Lookalike audiences take 1-6 hours to populate and update approximately every 3-7 days as Meta refreshes the underlying data.

Layering Lookalikes with Additional Targeting

In 2026, Meta’s guidance is to use lookalike audiences with minimal additional layering – adding excessive interest or demographic filters on top of a lookalike often reduces audience size without meaningfully improving quality. The lookalike signal is already fairly precise.

The one exception: exclusions. Always exclude your existing customers and recent purchasers from prospecting lookalike campaigns. There’s no benefit in showing acquisition ads to people who already bought.

Lookalikes vs Advantage+ Audiences

With Meta’s shift toward Advantage+ Audiences as the default targeting method, some advertisers wonder whether lookalikes are still worth building manually. The answer: yes, with nuance.

Advantage+ Audiences can use your custom audiences and lookalikes as “suggestions” – soft signals that guide the algorithm without constraining it. Uploading a strong customer list and setting it as a lookalike suggestion within an Advantage+ campaign is often the best of both worlds: the precision of a seeded source with the flexibility of Meta’s broader AI-driven discovery.

Measuring Lookalike Performance

Compare your lookalike campaign’s performance against broad targeting on the same creative, budget, and time period. Look for differences in CPA, ROAS, and conversion rate. Well-built lookalikes from strong sources typically outperform broad audiences in CPL/CPA metrics, though broad may sometimes win on sheer volume at scale.

Update your source audience regularly – refresh your customer list upload every 30-60 days to ensure the lookalike reflects your most recent buyers rather than customers from 18 months ago.

Frequently Asked Questions

How large should my source audience be?

A larger source of real customers gives the model more to learn from. Quality matters too, so a list of purchasers beats a list of page visitors, and the Meta Pixel and Conversions API keep that list accurate.

Are lookalikes still worth using with Advantage+ Audiences?

Often, yes. Both can run in the same account, and the section above compares when each fits. Our guide to broad vs narrow targeting adds more.

Final Thoughts

Lookalike audiences in Meta Ads remain a core prospecting tool in 2026, especially for advertisers with strong first-party data. The key is the source: a high-quality customer purchase list will generate a lookalike that genuinely outperforms interest-based targeting. A weak source produces a weak audience, regardless of how many percentage tiers you run.

Build from your buyers, exclude your existing customers, and refresh regularly. That’s the formula that makes lookalikes work.

Marquee manages Meta Ads campaigns with full-funnel audience architecture – from custom audiences and lookalikes to retargeting. If your prospecting isn’t performing, let’s take a look at your audience setup.