CPMs Spiked 33%. We Cut Cost Per Buyer by $810.
Paying a third more for media, we filled 115 more seats on $3,478 less and cut cost per buyer by $810.
Here's what my AI media buying and conversion teams found, what they recommended, and the decisions I made to turn a difficult flight around.
This is a real client account. I'm keeping the client anonymous, but I'll show you the numbers and the work behind them.
1. The Numbers Got Better While Media Got More Expensive
Going into the second half of October, ad costs jumped.
One flight hit a $1,893 cost per buyer, one of the highest CPAs we'd seen that year.
The next flight faced even higher media costs. CPM spiked from $40.71 to $53.95.
But the business results moved in the opposite direction.
Results at a Glance
| Metric | Previous Flight | This Flight |
|---|---|---|
| CPM | $40.71 | $53.95 |
| Cost per buyer | $1,893 | $1,083 |
| Buyers | 19 | 30 |
| Cost per booked seat | $35.23 | $28.60 |
| Seat fill | 85.9% | 95.6% |
| Ad spend | $35,968 | $32,490 |
The changes that mattered:
Clicks actually dropped 31%.
But seats per click improved from 5.3% to 8.5%, a 61% increase.
We weren't buying cheaper attention. We were getting more value from the attention we paid for.
2. First, the AI Team Found Tracking That Was Blocked
The first problem showed up in a split test.
It was reporting zero conversions.
Instead of assuming the page wasn't working, my AI team traced the tracking.
They found that the Meta pixel wasn't loading on two pages because of code someone had added.
I identified the code causing the problem and moved the pixel.
We restored tracking our team found blocked.
That matters because you can't make good decisions when your conversion data is incomplete.
I'm not crediting the pixel fix with the CPA improvement. It restored our ability to measure what was happening.
3. Then We Tightened the Rules That Protect Spend
Our automated cut rules were checking performance once a day.
That left a problem.
An ad could cross its spending limit and keep running overnight before the next check.
My AI team recommended checking every 30 minutes using the rules I'd already established.
I approved the change.
During this flight, the system only needed to pause two ads, each after spending about $50 without generating an application.
That wasn't where the $810 savings came from.
It was a practical improvement that helped prevent wasted spend from drifting between checks.
4. The Biggest Lever We Could See Was Retargeting
This is where the more meaningful performance change happened.
My AI media buying team flagged that retargeting wasn't getting enough of the budget.
Its report recommended increasing the allocation.
I made the call to move it.
Retargeting went from 16.7% to 26.1% of total spend.
We also expanded retargeting into three additional countries and removed its weakest ads.
During the new flight, retargeting registrations cost about $10.49 each.
The budget shift, audience expansion, and ad cleanup were likely the biggest drivers we could see behind the improvement.
I can't isolate one change and claim it caused the entire result. Several things changed together.
What I can say is that the AI team identified the opportunity, and I approved the moves.
5. The Conversion Team Is Still Testing
The AI conversion team also flagged the opt-in page for a free guide as a place worth testing.
It wrote three alternative headlines.
I selected an outcome-focused headline and approved the test.
Here's where the test stands:
| Headline | Opt-in Rate |
|---|---|
| Original | 26.81% |
| Challenger | 25.81% |
The original is slightly ahead.
There is no winner yet.
The test is still running, and we haven't rolled out the challenger.
That's part of the process I want people to see.
The AI team doesn't get to declare a win because a new idea sounds better. We let the evidence develop before making the next decision.
6. The AI Team Found It and Built It. I Made the Calls.
The result wasn't one magic optimization.
It was a sequence of work:
- Finding and restoring blocked tracking
- Tightening spend-protection checks
- Identifying a better budget allocation
- Cleaning up and expanding retargeting
- Opening a conversion test without calling a premature winner
The AI media buying and conversion teams handled the investigation, analysis, and recommendations.
I set the rules, approved the changes, and made the decisions.
That's how I want AI working inside my advertising operation.
Not replacing judgment. Giving me better information and more execution capacity so I can make the calls that matter.
Want to See How These AI Teams Work?
I use these AI media buying and conversion teams inside my own advertising operation.
You can learn how they work inside the free AoA Skool community.
Eligible members can also try the same AI workforce for $1 for the first 30 days, with a paid membership afterward unless canceled. AI usage and other third-party costs are separate.
Join the Free AoA SkoolSee the process. Ask questions. Decide whether it fits your business.
This case study describes one client's results, not a promise of what your business will achieve.