The hardest part of scaling isn’t finding budget: it’s keeping your campaigns from sliding back into the learning phase every time you touch them. When campaigns reset, efficiency tanks, optimization restarts, and you lose the predictability you worked so hard to build.
That’s why the usual advice is:
“Don’t increase budgets by more than ~15% day-over-day, or you’ll reset the learning phase.”
It’s a solid rule of thumb. But on its own, it won’t get you far. The real key is building a scaling routine- a structured sequence of actions designed to grow budgets while keeping campaigns stable and out of learning.
Here’s the routine I use, step by step, with real-world cases to prove it works.
1. The 15% Rule: Guardrail, Not Playbook
The 15% day-over-day budget increase is the most widely cited method for avoiding the learning phase. It reduces the shock to the algorithm, letting it adapt gradually.
But it’s just a guardrail. If your campaigns aren’t stable, even a 15% bump can reset learning. And if they are very stable, you may have room to push further than 15% without issues.
Example: In the Pretty Farm Girl scale from $500/day to $5,000/day, Smart Marketer respected incremental increases but also refreshed creatives and cleaned up the account. That broader structure is what prevented resets ; not the 15% alone. (Smart Marketer)
2. Establish a Reliable Baseline
The fastest way to trigger learning is to scale an unstable campaign. If your CPA, ROAS, or conversion volume is swinging wildly, the system is already struggling. Add spend on top, and you’ll almost guarantee a reset.
To avoid that, let campaigns run untouched for 5–7 days. Define a tolerance band (e.g., CPA variance within ±10%). If metrics stay inside it, you have a baseline. Only then is scaling safe.
Example: David Vargas used a “high-velocity creative testing framework” to stabilize trial sign-ups before scaling Meta spend. Without that baseline, every budget increase would have triggered instability and resets. (RevenueCat)
3. Scale by Nudging + Duplication
Even stable campaigns can hit learning if you scale too aggressively. Nudging budgets upward in 10–15% increments allows the system to adjust without restarting.
Duplication adds another layer of protection. By duplicating your top performer, the original stays optimized while the duplicate takes on new budget. If the duplicate enters learning, your core campaign isn’t affected.
Example: Many marketers report that duplication is the safest way to scale without widespread resets. If one duplicate falters, you pause it and protect your original campaign. (Reddit case study)
4. Expand Structural Levers
Another trigger for the learning phase is budget pressure inside a narrow campaign structure. If you just keep raising budgets on one audience or creative set, the algorithm may reset as it searches for new delivery opportunities.
Expanding structurally: new audiences, new creatives, new bidding strategies; gives the algorithm space to grow without panic resets.
Example: Vargas’ framework fed a constant stream of new creatives into “testing” campaigns. Winners graduated into “BAU” campaigns. This gave the system fresh signals to scale with, rather than forcing resets on old assets. (RevenueCat)
5. Insert Cool-Off Periods
Scaling too frequently keeps campaigns bouncing back into learning. After two or three increases, pause for 3–7 days. This lets the algorithm stabilize at the new budget level.
During this cool-off, monitor whether CPA or ROAS hold steady. If they do, you’ve avoided resets and can push again. If not, you know the system needs more time.
Example: In Pretty Farm Girl’s scale, stabilization weeks were part of the routine. By holding steady, they gave Meta’s system time to lock in performance before resuming growth.
6. Automate Guardrails
Many campaigns enter learning again when spend runs unchecked and performance collapses. Guardrails prevent this.
Set automation to:
Automatic pausing of ads (first line of defense)
- Cuts wasted spend quickly
- Preserves stability inside performing ad sets
- Makes ad-set pauses a last resort, not a reflex
Ad-set safeguards (last resort)
- Pause ad sets only if CPA stays above baseline for a sustained window
- Resume automatically once metrics recover to target ranges
Budget pacing trims
- Auto-reduce budgets if daily/weekly pacing overshoots plan
- Gradual cuts (e.g., 10–15%) to avoid triggering learning
Real-time alerts (so you can intervene before resets)
- Notify when ROAS drops below target
- Notify on sudden CPM spikes with flat CTR
- Notify when frequency crosses your threshold for X days
Guardrails save money & go the extra mile of stopping campaigns from crashing and re-entering learning unnecessarily.
7. Tie Scaling Into Your Workflow
Another hidden trigger for resets is poor coordination. If creative teams can’t supply new assets, if reporting lags behind scaling, or if duplicate campaigns overlap too heavily, you force instability that can push campaigns back into learning.
By treating scaling as part of the larger workflow : with aligned creative, forecasting, and reporting: you minimize these risks and keep performance consistent.
8. Scaling Across Verticals
The ways campaigns fall back into learning differ by vertical, but the same routine applies.
- E-commerce / D2C: Frequent creative refresh and duplication keeps campaigns stable. Example: Seltzer Goods scaled revenue 785% in 30 days by pairing spend growth with creative innovation. (Inflow)
- Mobile apps: Retention and CPI are fragile. Scaling safely means anchoring these metrics before adding spend. Example: Vargas’ framework kept trial sign-ups steady before scaling.
- SaaS / B2B: Bottom-funnel campaigns saturate quickly. Scaling without resets means layering new funnels and geos rather than just pushing budgets.
- Cross-vertical principle: Top-of-funnel activity fuels bottom-of-funnel performance. Don’t rely on the same creatives for too long: monitor frequency closely and refresh as needed. Consistently feeding high-quality traffic keeps lower-funnel campaigns efficient and resilient to resets.
9. Where Prism Fits In
Scaling without resets requires constant monitoring. That’s where Prism strengthens the routine:
- Unified baselines: See whether campaigns are stable across all channels before scaling.
- Scaling insights: Identify which audiences and creatives can absorb more spend without resets.
- Overlap detection: Prevent duplicated campaigns from competing with each other.
- Guardrails: Apply pacing and efficiency rules across platforms, reducing reset risk.
- Cross-channel view: Understand how scaling on Meta impacts Google or TikTok, so one channel’s growth doesn’t destabilize another.
✨ The bottom line: the 15% rule is your starting guardrail. The scaling routine + Prism’s automation is what turns that guardrail into a sustainable growth engine.
Final Takeaway
Every step in this routine exists for one reason: to scale without triggering the learning phase.
- Baselines stop instability from resetting campaigns.
- Nudges and duplication let you increase budgets without shocking the algorithm.
- Structural expansion gives the system space to grow without panic.
- Cool-off periods lock in performance at new levels.
- Guardrails catch problems before resets occur.
- Workflow integration prevents missteps that force instability.
- Vertical adaptations ensure scaling strategies fit your market.
- Prism automates the monitoring so you don’t miss hidden reset triggers.
Scaling becomes less about fearing resets — and more about confident, predictable growth.