How to create lookalike audiences in Meta Ads in 2026: quality seed audiences, 1%-10% sizes, Advantage+ Audiences, and best practices to find new customers.
Lookalike audiences remain one of the most powerful prospecting levers in Meta Ads — but how to use them changed radically between 2021 and 2026. The impact of iOS 14.5, the launch of Advantage+ Audiences, and the migration from Facebook Business Manager to Meta Business Suite transformed both the mechanics and the optimal strategy.
If your knowledge of lookalike audiences comes from tutorials written before 2022, this guide is for you.
A lookalike audience is a targeting type in Meta Ads that lets Meta’s algorithm find people with similar characteristics to a seed audience you define.
The principle is simple: you tell Meta “my best customers are these people” and the system analyses their demographic, behavioural, and interest attributes within its platform to find other users with a similar profile — people who don’t yet know your brand but statistically have a higher likelihood of converting.
The seed audience can be:
The quality of the seed audience directly determines the quality of the resulting lookalike audience. Garbage in, garbage out.
Since 2023, Meta has actively pushed toward Advantage+ Audiences, its AI-based targeting system. Here’s what each option actually does before you pick one.
Advantage+ Audiences uses the conversion signals from your account (pixel, CAPI, campaign history) to automatically find the audience most likely to convert, without you having to manually define interests, behaviours, or custom audiences.
You can provide a suggested audience (equivalent to your seed audience) as a starting point, but the system has the freedom to expand beyond it if it identifies conversion opportunities.
| Situation | Recommendation |
|---|---|
| Account with little conversion data (<50/week) | Manual lookalike — more control |
| Established account with good conversion volume | Advantage+ Audiences |
| New prospecting campaign without history | Start with 1–3% lookalike, migrate to Advantage+ as data accumulates |
| eCommerce with catalogue and many buyers | Advantage+ Shopping Campaigns (full automation) |
| Very specific niche where interests don’t work | Lookalike audience based on customer list |
On accounts with enough conversion signal, Advantage+ Audiences outperforms manual lookalike audiences — I’ve seen this on virtually every account where I’ve been able to run a side-by-side comparison. But manual lookalike audiences remain the best entry point when you’re starting out or when you want precise control over the prospecting profile.
The seed audience is the core of the system. A poor seed produces a poor lookalike audience, regardless of budget.
From highest to lowest quality:
Meta requires a minimum of 100 people in the seed audience to create a lookalike. But the practical minimum for quality results is:
If your list has fewer than 1,000 records, consider using a behaviour-based source (website visits, Instagram interactions) that has more volume.
In April 2021, Apple launched iOS 14.5 with the App Tracking Transparency (ATT) framework, which required apps (including Facebook and Instagram) to ask users for explicit permission before tracking their activity. Most users declined tracking.
The effect on lookalike audiences was direct: pixel signal was significantly reduced, especially for conversion events on Apple devices. This affects:
The right answer is the Meta Conversions API (CAPI). While the Meta pixel works from the user’s browser (and is therefore subject to iOS blocks), CAPI sends events directly from your server to Meta, without depending on the user’s device or browser.
With CAPI correctly configured:
If you don’t have CAPI implemented, it’s the first technical optimisation you should prioritise before scaling budget on prospecting with lookalike audiences.
The process is done from Meta Ads Manager (ads.manager.facebook.com).
Before creating a lookalike audience you need a custom audience as a seed. Go to Audiences within Ads Manager and create a new custom audience from:
With your custom audience ready:
Meta will take between 1 and 24 hours to process and make the lookalike audience available.
When creating or editing an ad set, go to the “Audience” section, click “Custom Audiences” and select the lookalike audience you just created.
Important: Don’t combine lookalike audiences with detailed interest or behaviour targeting unless you have a specific reason. The lookalike audience is already built on behavioural attributes; adding interest filters reduces it unnecessarily.
The percentage you choose when creating a lookalike audience determines the size and precision of the resulting audience.
Always start with 1%. When the cost per result starts rising because the audience is saturating (high frequency, growing CPM), move to 2–3%. Don’t jump directly to 10% without having validated the message with more qualified audiences.
In countries with large populations (USA, Brazil, Mexico), 1% can represent several million people — more than enough for months of campaign before needing to scale.
One of the least used but most powerful features of Meta Ads is the value-based lookalike audience.
Instead of treating all clients on your list identically, you can assign a numerical value to each (their LTV, average ticket, order value) and upload that information to Meta. The system will prioritise the attributes of your highest-value customers when building the lookalike audience, resulting in prospection oriented not just towards “people similar to your customers” but “people similar to your best customers”.
To use this feature, your CSV must include a value column (labelled value or LTV). Meta will automatically detect the column and offer the option to create a value-based lookalike audience during the upload process.
This technique is especially effective in eCommerce where there is significant variability in value per customer.
Without the Conversions API, your event signal is incomplete due to iOS 14.5+. Prioritise implementation before investing significant budget in prospection.
A lookalike audience based on 500 real buyers consistently outperforms one based on 10,000 newsletter subscribers. Source quality matters more than volume.
Lookalike audiences are created at a specific point in time and don’t update automatically. If your buyer list grows, create a new updated seed audience and generate a new lookalike. Many advertisers work with seed audiences that are months old without realising it.
When using lookalike audiences for prospection, exclude your custom audience of existing buyers from the ad set. Paying to show acquisition ads to people who are already customers is one of the most common waste points in Meta Ads.
Once you have sufficient conversion history (more than 50 weekly conversions), consider using your customer list as a suggested audience within Advantage+ Audiences instead of as a manual restriction. The system has freedom to find conversions beyond 1% or 3%, and with sufficient signal it will do so at a competitive CPA.
Review weekly for prospecting campaigns with lookalike audiences:
Using page fans as your seed audience. A Facebook page’s followers are the worst starting point for a lookalike audience: they’re heterogeneous, many may be bots or passive followers, and they have no purchase signal. Always use sources based on conversion behaviours.
Creating lookalike audiences with lists that are too small. With fewer than 300–500 people in the seed, the resulting lookalike has poor statistical precision. Meta builds the profile on an insufficient sample.
Overlapping 1%, 3%, and 5% lookalike audiences in the same campaign. If you have multiple lookalike audiences active simultaneously without exclusions between them, Meta bids against itself, inflates CPM, and confuses campaign learning. Use exclusions or consolidate into separate ad sets with separate budgets.
Ignoring audience exhaustion. A 1% lookalike in a small market can exhaust itself in weeks. Monitor frequency and prepare the next layer (2–3%) before the cost per result spikes.
Not updating seed audiences. Custom audiences and their derived lookalikes are static. If your customer base grows, you need to regenerate the seed audience and create a new lookalike to capture the most recent patterns.
If you want to analyse how your prospecting audiences are performing in Meta or review the complete structure of your account, book a free audit. I review advertising costs, audience quality, and technical setup in a 30-minute session.
30 minutes to review your situation and tell you exactly what I would change. No pitch, no sales proposal.