"Throw more budget at it to exit learning faster" — I bought my way into worse data

"Throw more budget at it to exit learning faster" — I bought my way into worse data

The advice for stuck campaigns: dump budget to hit the conversion threshold and exit the learning phase. I dumped, exited, and optimized toward garbage.

The setup
Lead offer stuck in learning, ~30 conversions, needed 50. I 3x'd the daily budget to rush it.

The move
It exited learning in two days. I celebrated, then watched cost-per-lead climb instead of fall.

The numbers
The rushed budget pulled in low-intent leads to hit the conversion count fast. The pixel "learned" from 50 cheap, junky conversions and optimized toward more of them. Cost per QUALIFIED lead went from $14 to $23. I exited learning and entered a worse equilibrium. About $2,100 to discover the threshold isn't the goal — the right conversions are.

The lesson
Exiting learning fast on low-quality conversions teaches the algorithm to find more low-quality conversions. Speed to threshold is not the same as quality of model.

What I'd do differently
I'd rather sit in learning an extra week with clean conversions than rush out on dirty ones. Optimize the input, not the timer. Numbers illustrative; the poisoned-pixel risk is real.