When you launch or significantly change an ad set, Meta’s system needs time to learn who is most likely to give you results. This is the learning phase.
What happens during learning
- Costs can be higher and less stable.
- Delivery is exploring different audiences and placements.
- The ad set shows “Learning” in the Delivery column.
How to exit it
Meta has generally said an ad set leaves the learning phase after about 50 results, such as leads or purchases. These need to happen within about a week of your last big edit. Check current guidance in Ads Manager.
“Learning limited”
This means the ad set is unlikely to get enough results to exit learning. Common causes:
- Budget too low for your cost per result.
- Audience too small.
- Too many ad sets splitting the budget.
Fixes
- Combine similar ad sets.
- Widen the audience.
- Increase the budget so the expected results are achievable; work it out with our daily budget calculator.
- Optimise for a more frequent event, such as leads instead of purchases, if purchases are rare.
Avoid resetting learning
Large changes to budget, targeting, creative or bid can restart learning. Make changes less often and in moderate steps.
Judge results fairly
Give new ad sets several days before deciding. Track cost per result with our CPA calculator.
Want the full picture? This article is part of Meta ads: the complete guide, our in-depth guide with everything in one place.
