<b>Scale in multipliers your algo can digest</b>

<b>Scale in multipliers your algo can digest</b>
+20% budget feels safe but on big spend that's a huge absolute jump → re-learning. On tiny spend it's nothing.
Think in <u>absolute dollars the algo already handles</u>:
— If it spends $300/day clean, adding $60 is a non-event → fine daily
— Adding $300 (a full duplicate of its known volume) → shock, learning resets
Rule of thumb: never add more in a day than the campaign already proved it can spend efficiently. Scale at the speed of proof.