Scaling What Works
You've found a winner. Now the dangerous part: pouring more money in without blowing it up. Scaling done wrong resets learning and tanks performance; done right, it grows results smoothly. Here are the two ways to do it.
By the end of this lesson you'll know
- The difference between vertical and horizontal scaling
- The 20% rule that protects your learning
- How to scale horizontally by duplicating winners
- Why profit, not ROAS, is the real ceiling
Two directions to grow
| Type | What it means | Best for |
|---|---|---|
| Vertical | Increase the budget on a winning ad set/campaign | Smooth, steady growth |
| Horizontal | Duplicate the winner into new audiences, geos or more creatives | Bigger, faster scaling |
Vertical scaling — the 20% rule
The instinct is to double the budget the moment an ad works. Don't. A big, sudden budget jump throws the ad set back into the learning phase and can wreck the performance you were trying to grow. The safe rule: increase the budget by roughly 20% every few days, letting it re-stabilise between steps.
Gentle, incremental increases preserve the learning the campaign has already done. You're nudging the system to spend more on what it already understands — not forcing it to relearn from scratch every time you get excited.
Horizontal scaling — duplicate winners
When you want to grow faster than ~20% steps allow, scale sideways. Take your winning ad or concept and:
- Launch it to a new audience (a different custom audience, lookalike, or interest).
- Expand into a new location or market.
- Add more creative variations on the winning angle.
Because each duplicate is its own ad set, you add scale without violently disturbing the original winner's learning.
As you scale, costs usually creep up — you're reaching beyond your easiest buyers. A campaign that looks great in Ads Manager can quietly turn unprofitable once you factor in product cost, shipping and fees. Track profit on ad spend, not just the ROAS number on screen.
Resist scaling by spawning dozens of tiny ad sets — that's just fragmentation again. Feed your winners well and keep the structure tight. One strong, well-funded campaign beats a sprawl of starved ones, at every size.
Key takeaways
- Vertical = raise budget on a winner; horizontal = duplicate it into new audiences/geos/creatives.
- Scale vertically by ~20% every few days to protect learning.
- Scale horizontally for bigger jumps without disturbing the original.
- Track profit, not just ROAS, and keep the structure consolidated.