The Audience Size Teardown: Diagnosing an Over-Narrowed Ad Set
Objective: Given a real Ads Manager ad set structure export, diagnose why a heavily layered interest audience is starving the algorithm of data, and check whether the account's ad-set-to-conversion-volume ratio can support the learning phase at all.
You're a growth marketer at Nykaa auditing an underperforming beauty prospecting campaign that has been live for two weeks with a rising cost per lead.
Read the ad set's targeting stack and its estimated audience size, then check whether the account is splitting too little conversion volume across too many ad sets.
Before you start
What you'll need
Free path (everything below is enough to finish)
The audience estimate is only shown live in Ads Manager, free to view without spending
Free, no account access needed beyond the exported numbers
Paid upgrades (optional, faster/deeper)
Flags starved ad sets automatically instead of a manual weekly export
No access? The manual Google Sheets tally above catches the same pattern at low ad set counts
The process
2 steps
Step 01 of 02
The lesson's warning callout: targeting under 100,000 people is usually a mistake, the algorithm needs room to find your best buyers, and most campaigns should aim for 500,000-5 million.
This ad set stacks four interest layers. Given the estimated audience size shown, is the targeting structurally broken before you even look at creative?
Procedure
- Open the ad set and list every interest layer stacked with AND logic
- Read the estimated audience size shown in the Audience Definition panel
- Compare that number against the lesson's 500,000-5 million guidance
Ad Set: Beauty Prospecting, Women 18-24 Targeting stack (AND): Interest: Organic Skincare Interest: K-beauty Products Behavior: Engaged Shoppers Interest: Cruelty-Free Cosmetics Estimated audience size: 42,000 people
Healthy
The diagnosis correctly flags 42,000 as far below the 500,000-5 million range and traces it to four stacked AND layers, not to the creative or the offer.
Unhealthy
Concluding the ad set is 'fine' because 42,000 sounds like a reasonably large number in isolation, without checking it against the lesson's stated range.
What this means
Audience size is a hard delivery constraint the algorithm needs room to work within, not a targeting precision score.
So what do I do about it?
| Symptom | Action | Effort |
|---|---|---|
| CPL keeps rising on an ad set with strong creative and clear offer | Check the Audience Definition estimate before touching the creative | 5 min |
Step 02 of 02
The learning phase ends once an ad set collects 50 conversion events within a 7-day window; below that, the ad set is still stuck testing delivery.
This campaign has 6 ad sets sharing 80 total conversions per week. Which of them can actually clear the learning phase threshold?
Procedure
- List each ad set's weekly conversion count from the export
- Divide the account's 80 total weekly conversions across the 6 ad sets
- Flag any ad set below 50 conversions/week as unable to exit the learning phase
Weekly conversions by ad set Beauty Prospecting, 18-24 13 Beauty Prospecting, 25-34 11 Beauty Prospecting, 35-44 9 Skincare Retargeting 22 Haircare Prospecting 15 Haircare Retargeting 10 Total: 80/week across 6 ad sets
Healthy
All six ad sets are correctly flagged as stuck below the 50-conversion threshold, and the fix proposed is consolidating them, not adding a seventh.
Unhealthy
Recommending a 7th ad set to test a new audience segment while every existing ad set is already starved of conversion volume.
What this means
Splitting limited conversion volume across too many ad sets keeps every one of them permanently stuck in the learning phase.
So what do I do about it?
| Symptom | Action | Effort |
|---|---|---|
| Cost per result stays high indefinitely across several ad sets, none of them stabilize | Consolidate overlapping ad sets so each one can individually clear 50 conversions/week | 30 min |
Final deliverable
A short written diagnosis naming the audience size problem, the ad-set-count problem, and a consolidated ad set structure that fixes both.
See a reference example
Chewy, autoship prospecting ad set audit AUDIENCE SIZE: Ad set 'Dog Owners, Grain-Free Interest' stacks 3 AND layers, estimated audience 68,000, below the 500K-5M range. AD SET COUNT: 5 ad sets share 65 weekly conversions (13 avg each), all below the 50-conversion learning phase threshold. FIX: Merge the 5 narrow ad sets into 2 broader ones (Dogs, Cats) and let Advantage+ Audience find converters within each.
Success criteria
You're done when you can:
- Correctly identifies audience size below the lesson's 500K-5M range from the targeting stack
- Correctly divides total conversions across ad set count to check the 50-conversion threshold
- Proposes consolidation rather than more granular targeting