A Meta Ads custom audience lets you reach groups of users who have already engaged with your brand. Because these users already know the brand, they can often produce a lower conversion cost than a cold audience.
There are three main sources: website visitors, customer lists, and engagement data. Each draws from a different data source and requires a different setup.
Meta’s custom audience documentation states that when a customer list is uploaded, data such as email, phone number, and name are encrypted with SHA256, and that responsibility for data management sits with the advertiser. The same documentation explains that the hashed data is matched against Meta’s own database across multiple keys (email, phone, name, etc.), and that using multiple keys improves the match rate.
Website Custom Audience
This audience is built from visitor behavior on the site, captured through the Meta Pixel or the Conversions API.
URL-based segmentation is possible
All visitors can be pooled into a single audience, or you can build separate audiences — for example, only product-page visitors, or people who added to cart but didn’t purchase. This separation lets you personalize the ad message based on intent.
Duration should be chosen based on intent
Audience duration (say, 7, 30, or 90 days) should be chosen based on how strong the visitor’s intent is. A short-duration audience may suit someone who abandoned a cart, while a longer duration may suit a general visitor. For example, showing the same ad to a cart-abandoner for 90 days can keep spending budget inefficiently after that user has already purchased elsewhere; a 7-14 day window is usually more accurate for this audience.
Customer List
This audience is built by uploading an email or phone list exported from a CRM or e-commerce system to Meta.
Match rate depends on data quality
When multiple keys (email and phone together) are uploaded, match rate can improve. Old or inaccurate data can make the audience size appear smaller than the actual customer count.
Data responsibility sits with the advertiser
Obtaining consent to collect and use the uploaded data is the advertiser’s responsibility. For more on this, see What Is Email Marketing Consent?.
Engagement Audiences
This audience is based on behavioral data such as mobile app usage, Canvas/Instant Experience interactions, or offline actions (like store visits).
Can be segmented by video watch percentage
Users who watched 25%, 50%, 75%, or 95% of a video can be defined as separate audiences. Users who watched most of the video are generally considered a more engaged audience.
How Should the Segmentation Strategy Be Built?
Audience structure should be built around where the user sits in the funnel.
Audiences are layered by funnel stage
General visitors, product-page visitors, cart-adders, and buyers each represent a different level of intent. A separate ad message can be considered for each layer.
Audience size also serves as the base for lookalikes
These audiences are also used as the source audience when building a Lookalike Audience. The quality of the source audience directly affects the quality of the resulting lookalike.
Pre-Launch Checklist
Before putting a custom audience into use:
- Does the audience source (website, customer list, engagement) match the campaign objective?
- Is the customer list current, or does it contain old, outdated data?
- Does the audience duration match the strength of visitor intent?
- Is consent for collecting and using the data documented?
Common Mistake
The most common mistake is pooling all website visitors into one audience and showing them all the same ad. A user who only visited the homepage doesn’t carry the same intent as one who abandoned a cart; audiences built without this distinction can waste ad budget.
Summary
Meta Ads custom audiences can be built from three sources: website, customer list, and engagement. Match rate depends on the quality of the uploaded data and how many keys are used. These audiences aren’t just used for direct targeting — they also serve as the source for building lookalike audiences.