How to Set Up Meta Audience Segments - Before It's Too Late
ยท Articles

Phil here, Director of Paid Social. 13 of the 76 accounts in our recent analysis had no Audience Segments defined.
For those 13 accounts, everything we found in that data was invisible. The Engaged Audience frequency running at 13 times a month. The Existing Customer frequency at 18. The $1.86M per month drifting toward warm audiences that nobody consciously allocated. The CPMr gap between prospecting at $81 and existing customers at $607.
It was all happening. There was just no way to know.
If you haven't set up Audience Segments yet, you're in the same position. Not because your account is broken, but because you don't have the visibility to know whether it is or not.
We tracked Audience Segments across 76 Meta accounts. Here's what the blended numbers were hiding ๐
Why this is more urgent than most Meta settings
Audience Segments are only forward looking. The data is only available from the date you define them. The diagnostic history starts the day you set them up, and not a day before.
The accounts in our analysis that had segments defined for the full January to May period could see exactly when Engaged Audience spend started climbing. They could pinpoint the month frequency jumped. They could track the drift.
The 13 accounts without segments could not. They were managing one blended number across all three audience types and calling it account performance.
Every week you wait is a week of data that will not exist when you need it.
What are Audience Segments?
Audience Segments is a setting in Meta Ads Manager that classifies your reach into three buckets: New Audiences, Engaged Audiences, Existing Customers.
Once defined, these labels appear as a breakdown option in your reporting. You can split any metric: spend, reach, frequency, CPMr, CVR, CPA by audience type. Instead of one blended number, you get three.
The segments don't change what Meta does. They make what Meta is already doing visible.
Meta is already serving your ads to a mix of new, warm, and existing audiences. It has always done this, it's a feature. The algorithm finds the cheapest conversions and gravitates toward them. Existing customers convert at nearly three times the rate of cold audiences and at half the CPA. The algorithm rewards itself by going back to them.
Without segments, that behaviour is invisible inside your blended metrics. With segments, you can see exactly how much of your budget is going where - and whether the split is something you'd consciously choose.
How to setup Audience Segments
Go to Meta Ads Manager. Open the All Tools menu and find Advertising Settings. Scroll to the Audience Segments section.
You'll see two configurable buckets: Engaged Audiences and Existing Customers. New Audiences is calculated automatically, it's everyone not in your other segments.
Defining your Engaged Audience
Add every custom audience that represents people who have interacted with your brand without purchasing. Website visitors (all pages, 30 to 180 days depending on traffic volume). Email and SMS subscribers who haven't purchased. App activity audiences. Lead form submissions. Video viewers at meaningful thresholds.
The goal is to capture everyone who knows you exist but hasn't converted. Be generous here. If in doubt, include it.
Defining your Existing Customers
Add purchase-based audiences. Your customer list uploaded as a custom audience. A website custom audience built from the Purchase event. Any audience that definitively signals a completed transaction.
Keep this tight. It should represent people who have actually bought, not people who are likely to buy.
One thing most setups miss
Social media engagers, people who have liked, commented on, or shared your posts on Facebook or Instagram, cannot be added to your Engaged Audience segment. Meta always classifies them as New Audience regardless of your segment settings.
You can exclude them from specific ad sets. But they will never appear in your Engaged Audience breakdown. They will always show up under New Audiences.
This matters more than it sounds. For accounts with large organic followings or active Partnership Ads programmes, a meaningful chunk of what appears as New Audience spend is actually going to people who already know the brand well. The New Audience figures will look healthier than they are. Worth knowing when you're interpreting the data.
What to track once they're defined
Add the Audience Segments breakdown in Ads Manager. Under Breakdowns, select By Delivery, then Audience Segments. This splits every metric in your selected date range by segment.
The three things to check on a weekly basis:
Spend percentage by segment. What proportion of total spend is going to New Audiences, Engaged Audiences, and Existing Customers? In the 76-account dataset, healthy accounts had prospecting at 75% or above. When engaged and existing combined start creeping above 20%, the drift is underway.
Frequency by segment. Not blended frequency. By segment. If blended frequency is 7 and your Existing Customer frequency is 18, those are two different situations requiring two different responses. You cannot see that without the breakdown.
CPMr by segment. Which segment is driving your account-level CPMr up? Across our data, Existing Customer CPMr ran at 7.4 times the cost of New Audience CPMr. If your account-level CPMr is climbing and you don't know which segment is responsible, you're fixing the wrong thing.
What to do when the numbers look wrong
If Engaged Audience or Existing Customer spend is running higher than you'd expect, the first interventions are always the same.
Exclusions first. Purchasers, email list, website visitors, social page engagers - all should be excluded from prospecting campaigns. Audit your exclusion stack before anything else. These audiences should have no place in cold prospecting, and if they're not excluded, the algorithm will find them because they're the cheapest conversions.
Then campaign structure. Overlapping campaigns with no exclusions bid against each other and create the conditions for the algorithm to concentrate spend on warm audiences. Consolidate first, then check whether the segment split improves.
The 1% distraction approach works well for accounts where Existing Customer spend keeps climbing despite exclusions. Run a dedicated retention ad set with a hard budget cap - around 1% of total spend. This gives the algorithm a controlled place to serve existing customers. Without it, Meta finds them in your prospecting campaigns instead. Give it somewhere small to go so it doesn't invade everywhere else.
The accounts that can't see any of this
In our dataset, the 13 accounts without segments defined were not obviously underperforming. Their blended metrics looked similar to accounts with segments. Blended ROAS was within normal range. Blended frequency was manageable.
That's the point. The blended numbers don't tell you whether the performance is coming from new customers or existing ones. They don't tell you whether Meta is quietly growing your customer base or quietly harvesting the one you already have.
The accounts with segments defined could see the difference. The accounts without could not.
Setup takes ten minutes. The data starts accumulating from today. Come back in eight weeks and you'll have two months of segment-level diagnostics that don't exist in any account that didn't set this up.
That gap between what you can see and what you can't is what this is actually about.
Read more about CPMr:
READ THE FULL CPMR BREAKDOWN
Book a call:https://taikundigital.com/linkbios/phil