Hey.
Everyone talks about their winning ads.
Nobody talks about the 99% that failed.
Here's how a $3.3M ad spend in meta actually looked like:
28 campaigns launched
871 ad sets tested
3,579 individual ads created
Only 1 ad spent more than $500k, 2 others spent $100k+
That's a 0.08% UNICORN! rate.
The other 3,576? Dead on arrival or killed within days.
Most marketers would call this failure. I call it the cost of finding what actually works.
The uncomfortable truth about scaling
When people see a brand at $100k/$200k or even $300k+ per month in revenue, they imagine some genius marketer who "cracked the code"
The reality?
It's a graveyard of failed tests with a few zombies that refused to die (YES, I LOVE THE WALKING DEAD!)
Here's the monthly massacre from one account:
Month 1-2 (Foundation):
Ads launched: ~50
Survivors: 2
MRR: $5k
Month 3:
Ads launched: ~300
Survivors: 1 (but what a survivor)
MRR: $20k
Month 4-6 (Creative Mining):
Ads launched: ~1,200
Survivors: 0
MRR: $36k → $60k (riding Month 3's winner)
Month 7-8 (Scale or Die):
Ads launched: ~1,200
Survivors: 0
MRR: $210k (still riding that ONE winner)
Here's what we actually tested:
Week 1-2: Messaging mining
100+ ads launched in traffic campaigns
Kill everything after 400 impressions
Document what wins and loses
Week 3-4: Creative format testing
Take surviving messages
Test across 5 formats (static, video, UGC, carousel, etc.)
Kill anything that sucks!
Week 5-8: Audience expansion
Take format winners
Test against 20+ audience segments
Look for unexpected pockets of performance
Month 3+: Scale testing
Only scale ads that maintained metrics at scale!
Most "winners" died here
The kill rate? 99.2%
But that 0.8% that survived? Pure gold.
What the data taught me
Lesson 1: Your hit rate will be terrible
Accept it. Budget for it. Most of your ads will fail immediately. That's not incompetence - it's the cost of finding outliers.
Lesson 2: One breakthrough > 1,000 iterations
We spent months iterating on ads. Sometimes you need a complete paradigm shift, not a copy or format tweak.
Lesson 3: Month-over-month compounds
Month 1: Learn who you're selling to
Month 2: Learn what message resonates
Month 3: Find your breakthrough format
Month 4+: Ride the wave while testing the next one
Lesson 4: Winners emerge, they aren't created
You can't force a winning ad (or at least that's what I think)... You can only create conditions for winners to reveal themselves.
aaand the framework you can steal...
The 99/1 Testing Protocol:
Accept the 99% fail rate. Budget assuming 99 out of 100 ads will fail. If you can't afford this, you can't afford to scale.
Test in waves, not drips. Launch ads weekly in batches (or at once). Not 5 today, 3 tomorrow, 7 on Sunday...
Kill fast, scale slow. I usually recommend ads to spend at least 3-5x the CPA or 72-hours (I have an exact calculator for this - let me know if you're interested and I may write a full piece on this)
Document the corpses. Every failed ad teaches you something. Track why things died, not just what lived.
Build investment engines. Stop optimizing for quick conversions. Build funnels that provide additional learnings (like quizzes, they're DATA RICH!)
Your move
What's your current test velocity?
How many ads did you launch this week?
Hit reply and tell me your biggest testing constraint. Budget? Creative resources? Approval bottlenecks?
I read everything and might solve your specific problem in the next newsletter.
Kris