/Advertising
Question
My campaigns keep resetting learning phase – am I editing too much?
Posted by •11/20/2025
Meta question (literally).
I’ve read that you shouldn’t touch your campaigns too often or they keep going back into the “learning phase”.
Thing is:
- I'm a beginner
- I keep wanting to tweak audiences, budgets, creatives
- Every time I do, it shows “learning limited” or “learning” again
Details:
- 3 ad sets, $10/day each
- Optimization event: Purchase
- I'm making changes almost every day (turning off ads, changing budgets, etc.)
Questions:
1. How often should a **beginner** realistically touch their campaigns?
2. Which changes reset the learning phase and which are “safe”?
3. Is it better to let a “bad looking” ad set run longer to exit learning, or cut it earlier to save money?
I’ve read that you shouldn’t touch your campaigns too often or they keep going back into the “learning phase”.
Thing is:
- I'm a beginner
- I keep wanting to tweak audiences, budgets, creatives
- Every time I do, it shows “learning limited” or “learning” again
Details:
- 3 ad sets, $10/day each
- Optimization event: Purchase
- I'm making changes almost every day (turning off ads, changing budgets, etc.)
Questions:
1. How often should a **beginner** realistically touch their campaigns?
2. Which changes reset the learning phase and which are “safe”?
3. Is it better to let a “bad looking” ad set run longer to exit learning, or cut it earlier to save money?
1 Reply
•11/20/2025
Basic rule: treat campaigns like a patient in a hospital. If you keep unplugging and replugging everything, nobody knows what actually works.
Edits that usually **impact learning**:
- Big budget changes (e.g. doubling/halving)
- Changing optimization event (ATC → Purchase, etc.)
- Large targeting edits
- Changing placements significantly
- Swapping lots of creatives at once
Edits that are mostly **safe**:
- Turning off 1 bad creative inside an ad set
- Small budget changes (e.g. +10–20%)
- Fixing typos in ad text
For beginners:
- Plan your test structure **before** launching
- Then let it run 3 full days without touching
- After that, make decisions once per day, not every few hours
If an ad set is clearly horrible (CTR < 0.5%, no ATC after decent spend), cut it even if it hasn’t left learning. No need to “respect the phase” if it’s obviously trash.
But don’t restart or heavily edit things that are getting:
- Decent CTR
- Some ATC/IC
You need data consistency more than “perfect” learning status.
Edits that usually **impact learning**:
- Big budget changes (e.g. doubling/halving)
- Changing optimization event (ATC → Purchase, etc.)
- Large targeting edits
- Changing placements significantly
- Swapping lots of creatives at once
Edits that are mostly **safe**:
- Turning off 1 bad creative inside an ad set
- Small budget changes (e.g. +10–20%)
- Fixing typos in ad text
For beginners:
- Plan your test structure **before** launching
- Then let it run 3 full days without touching
- After that, make decisions once per day, not every few hours
If an ad set is clearly horrible (CTR < 0.5%, no ATC after decent spend), cut it even if it hasn’t left learning. No need to “respect the phase” if it’s obviously trash.
But don’t restart or heavily edit things that are getting:
- Decent CTR
- Some ATC/IC
You need data consistency more than “perfect” learning status.