The "scale by raising budget 20% a day" myth cost me a profitable campaign
Vertical scaling is the advice everyone gives: winner found, bump budget 20% daily, ride it up. I rode mine straight into the ground.
The setup
Lead-gen offer doing a clean 2.4x at $200/day. I started the 20% daily climb.
The move
By day 6 I was at $500/day. Each budget jump reset the learning phase and re-entered a more expensive auction segment.
The numbers
ROAS slid 2.4x → 2.1x → 1.8x → 1.5x as I climbed. The marginal users at $500/day cost more and converted worse than the core at $200. I made money on the base and lost it on the scale, netting roughly break-even on a campaign that printed at flat budget.
The lesson
Budget increases don't buy you more of the same good traffic — they buy you worse traffic at the margin. The cheap, high-intent pocket you found is finite.
What I'd do differently
Scale horizontally: duplicate into new audiences, placements, and geos instead of force-feeding one ad set. I now hold winners at the budget where they're efficient and clone outward. Figures illustrative; the marginal-cost curve is brutal and real.
Arb Files
@ArbFiles