Facebook Custom Audiences: How to Create, Sync and Exclude Them (2026)
Every Custom Audience type, the exact setup steps, why match rates decide performance, and how enriched Klaviyo segments make better seeds than raw CSV exports.
What a Facebook Custom Audience actually is
A Custom Audience is your data, matched to Meta's users. You supply a list - customers, subscribers, site visitors - Meta hashes it, finds the matching profiles, and lets you target or exclude those people across Facebook and Instagram. It's the bridge between what you know about your customers and what Meta can do with it.
There are four sources, and they behave very differently:
- Customer list - a CSV or live sync of emails, phone numbers, names. Your highest-value source, and the one this guide focuses on.
- Website - Pixel-based visitors. Updates continuously, but post-iOS14 coverage is partial and shrinking.
- App activity - SDK events, if you have an app.
- Engagement - people who watched videos, opened lead forms, interacted with your page. Useful but shallow signal.
How to create a Custom Audience from your customer list
- In Ads Manager, go to Audiences and choose Create Audience → Custom Audience → Customer list.
- Prepare a CSV: email as the primary identifier, plus first name, last name, country and postcode columns where you have them. More fields means a better match rate.
- Upload. Meta hashes the file on upload - the raw data doesn't sit in Ads Manager.
- Name it by segment and date ("vip-buyers-2026-08", not "audience 1"). You'll thank yourself in six months.
- Wait 24-48 hours for the audience to populate before judging its size.
Why the list underneath decides performance
The upload mechanics take ten minutes. What separates a Custom Audience that performs from one that doesn't is the quality of what you uploaded:
- Match rates. A raw email export typically matches at 50-70%. Adding name and postcode fields pushes that up; stale or role-based emails drag it down.
- Segmentation. "All customers" is the least useful list you own. A top-decile buyer segment, an at-risk lapsed segment, or a recent-mover segment each give Meta a sharper signal - and give you a cleaner test.
- Freshness. A CSV is a snapshot. Your 90-day buyers list stops being your 90-day buyers list the day after you export it.
Enriching the list before upload is the single biggest upgrade available: profiles carrying household attributes (purchasing power band, family stage, move stage) let you build segments like "top-decile affluent recent movers" that a raw order export can't express. The enrichment guide covers how.
Upload Custom Audiences Meta can actually learn from
Outra enriches your Klaviyo segments with household attributes - purchasing power, family stage, move stage - and syncs them live into Meta. Household-level suppression stops you paying to re-acquire existing customers.
What Outra knows about every UK household
All attributes and segments are tied to a single identifier: the household, not an email, not a cookie.
- Employment stage
- Move stage
- Age band
- Family stage
- Occupancy status
- Property type
- Bedrooms
- Property size
- Property value band (£)
- Garden
- Garage
- Parking / Driveway
- Area type
- Region
- Purchasing power band
- Household income band
- Credit score band
- UHW (ultra-high wealth)
Do Custom Audiences update automatically?
Website and app audiences update continuously. Customer list audiences do not - they're frozen at upload. Two fixes: re-upload on a schedule (monthly at minimum, weekly for fast-moving segments), or wire a live sync from your ESP so profiles are added and removed as they move in and out of the segment. The sync version is the one that keeps suppression honest.
Exclusions: the highest-leverage use
The most valuable Custom Audience in most accounts is the one used to exclude. Build an "all customers + 90-day subscribers" audience, then exclude it from every prospecting ad set. Without that step, a meaningful share of your prospecting budget is spent re-acquiring people who already bought - at acquisition CPMs. Suppression is also where household-level resolution earns its keep: one person with three email addresses gets excluded once, at the household level, instead of slipping through two of the three.
Custom Audiences as Advantage+ and lookalike seeds
Every lookalike and every Advantage+ audience is modelled from a seed, and the seed is a Custom Audience. The algorithm can only extrapolate from what you feed it - a 5,000-profile enriched seed of your best buyers produces a materially better modelled pool than a 50,000-row "everyone who ever ordered" dump. The full head-to-head on seeds versus lookalikes is in household audiences vs lookalike audiences.
Common failure modes
- "Audience too small." Meta needs roughly 1,000 matched users. If a 1,500-row CSV produces a sub-1,000 audience, the problem is input quality, not the upload.
- One giant list. Undifferentiated uploads prevent you from learning anything. Segment first, upload second.
- Set-and-forget. A quarterly-refreshed exclusion list is a leak in every prospecting campaign you run.
The bottom line
Custom Audiences are the highest-control surface Meta still offers. The mechanics are trivial; the leverage is all in list quality, segmentation, freshness and suppression. Treat the audience as a product - enriched, segmented, synced - and every campaign downstream gets sharper. For the targeting layer above this, see how to target a specific audience on Facebook ads.
Read next
Upload Custom Audiences Meta can actually learn from
Outra enriches your Klaviyo segments with household attributes - purchasing power, family stage, move stage - and syncs them live into Meta. Household-level suppression stops you paying to re-acquire existing customers.
