Alright.
This week I'm sharing something we've been running since summer that's consistently outperforming our standard broad campaigns.
It's not complicated. But it works.
We're spending $50k+/month on one account using this method. Lower CPMs. Better engagement. Cheaper leads.
Watch the full breakdown here: [Loom link]
Below is the written version if you prefer to read or want to save it.
The Problem
Everyone's running the same broad, open-targeting campaigns. Same generic ads. Same tired hooks.
Meta's algorithm is good, but you're not feeding it properly.
You need to give it signals. Fast.
The Solution
Hyper-niche micro-audiences with open targeting.
Sounds contradictory. It's not.
Here's how it works:
The Method
1. Keep targeting open
No demographic restrictions on Meta
Let the algorithm do its job
Don't fight the system
2. Build micro-audience ad sets
Same campaign
Multiple ad sets, all open targeting
Each one represents a specific type of person in your niche
Example: Running credit repair?
One ad set for construction workers
One for realtors
One for unemployed people
One for single parents
One for travellers
You know your customers. Break them down.
3. Make hyper-specific creative for each
Every ad set gets its own ads
Written specifically for that person
Use their language, their problems, their terminology
If you're targeting realtors: mention broker fees, commission structures, CRM systems — whatever they actually talk about.
Make it so specific that when they see it, they stop scrolling.
How to Build the Ads
Use Claude (better for scripts than ChatGPT).
Prompt it properly:
"Give me ad scripts for realtors with bad credit"
"What problems do construction workers face with credit?"
"What language do single parents use when talking about debt?"
Layer in the terminology. Make it native to that audience.
Then create a mix:
Static images
AI UGC (massively underrated)
Real UGC if you've got it
Stock footage
Test everything. Find the winners. Scale them.
Why This Works
Lower CPMs
You're showing the right content to the right person
Meta rewards relevancy
We've seen CPMs drop significantly across accounts
Better engagement
People watch ads that feel like they're made for them
Likes, comments, and watch time send strong signals back to Meta
Algorithm learns faster
Cheaper cost per lead
Better engagement = better click-through rates
Better CTR = lower cost per click
Lower CPC = cheaper leads
Higher quality
Because leads are cheaper, you can add more friction
Extra qualifying questions specific to that audience
Better leads, same or lower cost
Real Results
We're spending £50k+/month on one account using this method.
CPMs averaging low-to-mid £20s (some ad sets even lower).
Cost per click around £3 or less.
Click-through rates consistently above 1%.
Cost per lead is cheaper than our generic broad campaigns, and quality is better because we're layering in more qualification.
The Process Going Forward
Every campaign I build now follows this structure:
Campaign level: Broad objective (leads, conversions, whatever)
Ad set level: 5–12 micro-audiences, all open targeting
Ad level: Hyper-specific creative for each audience type
Then it's just:
Test
Find winners
Scale
Repeat
Meta is a creative game now. Always has been, but especially now.
Pump in volume. Test fast. Kill the losers. Scale the winners.
This method just makes sure you're testing the right creative for the right people, even when targeting is wide open.
Bottom Line
Open targeting works. But you still need to feed the algorithm signals.
The fastest way to do that is by making ads so relevant that people engage immediately.
Hyper-niche creative does that.
Lower CPMs. Better engagement. Cheaper leads. Higher quality.
Simple.
Try this in your next campaign.
If you test it, reply and let me know how it goes.
More next week.
James