Lookalike Audiences Today
Lookalikes were the crown jewel of Facebook targeting for a decade: show Meta your best customers, and it finds more people like them. They still work — but their role in 2026 is different, and knowing that saves you from misusing them.
By the end of this lesson you'll know
- How a lookalike is built
- Why the source list decides everything
- What audience size to choose
- Their honest role in 2026
How they work
You give Meta a source audience — your customers, leads, or website converters — and Meta finds new people who resemble them across its platforms. It's how you scale beyond people who already know you, without hand-picking interests.
Source quality is everything
A lookalike is only as good as the list it's built from. Feed it a random list and you get a random result. Feed it your actual buyers — better yet, your high-value buyers — and Meta hunts for more of your best customers. Aim for a source of at least 1,000–ish people for a meaningful pattern.
If you send purchase value with your events (Module 3), you can build a lookalike from your highest-spending customers. Instead of “more buyers,” Meta looks for more high-value buyers — a meaningful upgrade for many businesses.
Choosing the size
Lookalikes range from 1% (the closest match to your source, narrowest) up to 10% (much broader, less similar). Start on the tighter end — around 1–3% — when your source is strong, and widen only if you need more scale.
Their honest role in 2026
Here's the update most guides skip. Like interests, lookalikes are now usually treated as a suggestion — for many common goals you can't hard-restrict delivery to them. And with Meta's AI so capable, a strong Custom Audience fed straight into Advantage+ often outperforms a manually built lookalike.
So lookalikes haven't died — they've been demoted from “the strategy” to “a useful signal.” They're most helpful for newer accounts where Meta lacks data, or as one more seed layered into a broad campaign.
Build one lookalike from your best customers and treat it as a starting signal, not a fence. Don't agonise over 1% vs 2% vs 3% — source quality matters far more than the percentage.
A lookalike built from a tiny, stale, or irrelevant source is worse than no lookalike — it points Meta at the wrong people with false confidence. Fix the source before you blame the tool.
Key takeaways
- Lookalikes find new people who resemble a source you provide.
- Source quality decides everything — use your best (or highest-value) customers.
- Sizes run 1% (closest) to 10% (broadest); start tight.
- In 2026 they're a signal, not the strategy — strong custom audiences often win.