What is a custom audience in Meta ads?

A custom audience is an audience Meta (Facebook and Instagram) builds from data you already own. Uploaded emails or phone numbers, pixel events like ViewContent or Purchase (a pixel is a small snippet of code on your site), app activity, video views, and Instagram engagement can all become one. You then target or exclude it. Let's take a drink brand as an example. They sell a low-sugar sports drink for active women over 50, mostly from their own website. The drink brand builds a custom audience of people who visited the website in the last 14 days, and another of people who already bought, used as an exclusion.

Is a custom audience the same as a lookalike audience?

No. A custom audience contains people who already interacted with you. A lookalike audience is a new, cold audience modelled on a custom audience (people who resemble your seed list but have never engaged). The custom audience is the seed. The lookalike is the output.

Glossary

Custom audience

A custom audience in Meta ads is a group of people you already know that you can show Facebook and Instagram ads to. You build it from your own data: a customer file, people who visited your site, people who used your app, or people who liked or watched your posts. It is a list of people who already showed up, not a lookalike (new people who only resemble them).

Theodor Lindfors, Founding Marketer · September 1, 2026

What a custom audience is

Let's take a drink brand as an example. They sell a low-sugar sports drink for active women over 50, mostly from their own website. The drink brand advertises on Meta (Facebook and Instagram), TikTok, and YouTube. A custom audience is simply a list of people who already did something with the drink brand, so Meta can show them ads (or keep them out of ads).

In Ads Manager the drink brand keeps three lists. Website visitors, last 14 days. People who added a case to cart but did not buy, last 7 days. People who already bought, last 180 days. The first two are who the drink brand shows ads to again (retargeting: ads to people who already showed up). The third is who the drink brand keeps out of ads meant for new customers.

Meta (Facebook and Instagram) groups those lists by where the names came from. A customer file audience is an email or phone list the drink brand uploads. Meta hashes it first (turns each address into a fingerprint, so Meta is not storing the raw email in the ad tool). A website audience comes from the pixel (a small snippet of code on the drink brand's website) or the Conversions API (the same events sent from the drink brand's server). An app audience comes from an SDK (a small kit inside an app). Engagement audiences come from people who watched a drink brand video, opened a lead form, or interacted with the Instagram or Facebook page. Every one of them is a warm list.

Why custom audiences matter

Custom audiences are the two levers you actually control on Meta (Facebook and Instagram) now: who to chase again, and who to keep out. Excluding recent buyers from a prospecting campaign (ads to people who have never engaged) is a custom audience decision.

Custom audiences are also the seed for lookalike audiences (new people who resemble the list), so their quality sets a ceiling on the cold targeting built on top.

How to read custom audience performance

Check size and match rate first. Pixel audiences (lists built from site tracking code) shrink with browser restrictions, and an uploaded file that matches poorly leaves you targeting a fraction of the list you thought you had.

Let's take a drink brand as an example. They sell a low-sugar sports drink for active women over 50, mostly from their own website. The drink brand exports 8,000 customer emails. Meta (Facebook and Instagram) matches 6,000 of those 8,000 emails to accounts, a 75% match rate. The remaining 2,000 emails never become targetable. If the drink brand had uploaded emails plus phone numbers, the match rate would likely have been higher. The 14-day visitor audience shows 12,000 people. The 180-day visitor audience shows 48,000 people. Averaging results across both windows hides that the 14-day list of 12,000 converts and the 180-day list of 48,000 mostly annoys people.

Mind the retention window. A 30-day site visitor audience behaves nothing like a 180-day one, and averaging both into a single result hides which window is doing the work.

Common custom audience mistakes

Uploading a stale customer file and blaming the algorithm for the match rate.
Building overlapping audiences that compete in the same auction.
Forgetting exclusions, so buyers keep seeing acquisition offers.

Lemonado

How Lemonado helps with custom audiences

Lemonado reads campaign results next to the revenue from those lists when the accounts are connected, so a retargeting audience is judged on sales, not list size.

See marketing analytics