# How to build lookalike audiences that still work

> Build lookalike audiences from your best customers, with a source of 1,000 to 5,000 people, then test them against Advantage+ audience. Steps and an example.
- **Author**: Marxx
- **Published**: 2026-11-18
- **Category**: Meta
- **URL**: https://marxx.ai/library-guides/how-to-build-lookalike-audiences-that-still-work

---

Build lookalike audiences from your best customers, not from all of them, because Meta says that can give better results. Use a source of 1,000 to 5,000 people, which is what Meta generally recommends, with a minimum of 100. Then test the lookalike against Advantage+ audience, because Meta's system can now deliver beyond it.

## What changed

- **No location at creation.** You no longer pick a location when you create a lookalike. Meta uses the ad set's location and shows the estimated size inside the ad set ([Meta Help Centre](https://www.facebook.com/business/help/465262276878947), checked 2026-10-20).
- **Advantage+ lookalike is automatic.** For new, duplicated and draft campaigns and ad sets that use lookalikes built from mobile, website or customer list audiences, Meta uses your lookalike as a guide. It can deliver beyond the percentage you chose when it expects better results ([Meta Help Centre](https://www.facebook.com/business/help/1212225059146059), checked 2026-10-20).
- **Some ads are limited.** Lookalikes are limited or unavailable for some ads about financial products and services, employment or housing.

## Steps

1. **Pick a source of your best customers.** Meta says an audience built from your best customers can give better results than one that includes all of them. Repeat buyers and high spenders are good candidates.
2. **Check the size.** The source needs at least 100 people. Meta generally recommends 1,000 to 5,000. A saved audience cannot be a source.
3. **Create it.** Open **Audiences**, click **Create audience**, then **Lookalike audience**. Pick your source, set the size with the slider and click **Create audience** ([Meta Help Centre](https://www.facebook.com/business/help/465262276878947), checked 2026-10-20). A smaller percentage matches your source more closely. A larger one is broader ([Meta Help Centre](https://www.facebook.com/business/help/164749007013531), checked 2026-10-20).
4. **Do not wait for it to fill.** The status reads Populating for up to three days. You can run ads at once, but the audience grows as it fills. "Audience not created" can mean your source was too small.
5. **Add order value if you have it.** If your list has a value column, such as order value, pick that custom audience as the source for a value-based lookalike ([Meta Help Centre](https://www.facebook.com/business/help/458132024732845), checked 2026-10-20).
6. **Test against Advantage+ audience.** Meta recommends A/B testing Advantage+ audience for almost all campaign types, but not for retargeting ([Meta Help Centre](https://www.facebook.com/business/help/273363992030035), checked 2026-10-20). See [Advantage+ audience vs the original audience](/compare/advantage-plus-audience-vs-original-audience).

Start with a [custom audience](/glossary/custom-audience) of your customers. Then build the lookalike from it.

## A worked example

Say you have 12,000 past buyers. You sort them by lifetime spend and take the top 2,400, which is 20% and fits Meta's 1,000 to 5,000 range.

You A/B test two ad sets for 7 days, the minimum Meta recommends ([Meta Help Centre](https://www.facebook.com/business/help/290009911394576), checked 2026-10-20). Both use the same ad and INR 2,000 a day, so INR 14,000 each. The results are made up.

| Ad set | Spend | Purchases | Cost per purchase |
|---|---|---|---|
| Lookalike of top 2,400 buyers | INR 14,000 | 28 | INR 500 |
| Advantage+ audience, lookalike as a suggestion | INR 14,000 | 35 | INR 400 |

Cost per purchase is spend / purchases: 14,000 / 28 = INR 500, and 14,000 / 35 = INR 400. Here Advantage+ audience wins. Your test could go the other way, which is why you run it. Compare both with [broad targeting vs interest targeting](/compare/broad-targeting-vs-interest-targeting) too.

## Mistakes to avoid

- A source under 100 people, which can leave the audience uncreated
- A source of all buyers, including one-time discount buyers
- Editing the ad set during learning. See [how to exit learning limited](/guides/how-to-exit-learning-limited).
- Using a lookalike for ads in restricted categories

## FAQ

### How big should a lookalike audience be?

Meta says smaller percentages match your source more closely and larger ones give a broader audience. Pick by goal, and test two sizes.

### Do lookalike audiences still work with Advantage+?

They still work as a guide. Meta may deliver beyond your chosen percentage, so test the lookalike against Advantage+ audience instead of assuming either one wins.

### How many people do I need in the source?

At least 100. Meta generally recommends 1,000 to 5,000 (checked 2026-10-20).

---
- [More Meta articles](https://marxx.ai/library-guides/category/meta)
- [All articles](https://marxx.ai/library-guides)