Great analysis and we anecdotally see about the same thing.
The problem is that humans need to be in control and tinker, so most people really struggle with trusting someone/something else.
Most people I speak with still feel the need to control budget allocation at a very granular level. Even if the results are worse and they are shown data to prove it they still don’t want to hand over control.
An analogy I like to use is your the coach of a basketball team and you think one of your players is the best and keep wanting to give him the ball despite him playing like crap and another player absolutely killing it.
Meta picks a winner within 72H and is right 67% of the time
I went digging into how fast Meta actually decides which ad to back, and how often that early call sticks.
A lot of ad accounts, millions in spend, hundreds of batches of ad launches analyzed.
A launch is a batch of fresh ads dropped into one ad set on the same day.
Mostly lowest-cost bidding, a mix of CBO and ABO.
The average launch ran about 7 new ads, 4 of them real spenders.
Inside each batch the budget, audience and timing are identical, so the only thing that moves is velocity.
When budget rushes into a single ad early, Meta is telling you it's confident in that angle, that persona, right now.
Here's how often that early confidence is right.
Take the ad leading on spend at day 1.
It stays the top-funded ad across its entire run 45% of the time.
By day 3, 67%.
By day 7, 75%.
A blind guess inside a batch is about 1 in 5.
What that 75% means: in three of four launches, the ad Meta was spending the most on by day 7 was the same ad that finished with the most total spend over its whole life, from first delivery until it stopped getting budget.
The early leader becomes the lifetime leader.
And we didn't manufacture that by killing the rest.
About 70% of the batch was still delivering past day 7.
The other ads weren't force-paused on day 1, they faded on their own while the winner pulled ahead in a live field. The read is fair.
So what do you do with a one-week head start?
1. Iterate faster
If an ad is pulling spend by day 5, build the next version off it now.
Same angle, same persona, fresh executions.
You're feeding the flywheel off a signal that holds three times in four.
2. Scale sooner
Real performance concentrating in the first few days is usually enough to start adding budget.
You're following the algorithm's own money.
Of course ads don't perform in a vacuum.
They run in sequence across the customer journey, so the longer your journey and the more ads you have live, the more you weight the batch over any single unit.
Read the signal in context
And the ad that spends hard but sits under your CPA?
The data is blunt: efficiency barely moves between day 7 and day 14.
If your top spender is below threshold after a week, more time probably won't rescue it.
Meta found the volume.
The margin is on you, through creative quality and spend discipline at least on lowest cost.
Meta commits fast, inside a week, and it's right far more often than not.
Trust the velocity EARLY, iterate and scale on it to increase your creative flywheel, thus increasing the overall business growth rate.