The period after you launch or significantly edit a Meta ad set, while the delivery system explores how best to deliver it. Meta says ad sets usually exit learning after about 50 results in the week after the last significant edit; performance is less stable until then.
Formula
Weekly results ≈ daily budget × 7 ÷ cost per result
Example
Meera's ad set optimises for website purchases at about ₹600 each with a ₹2,000 daily budget. That is ₹14,000 a week, or about 23 purchases, well short of the roughly 50 Meta looks for.
How you'll hear it
“Don't touch the ad set this week; every big edit sends it back into learning.”
What it helps you decide
Whether to leave an ad set alone, merge small ad sets, raise budget or optimise for a more frequent event. It also tells you not to judge an ad set on its first few days.
Where people misread it
Treating early, volatile results as final and editing the ad set daily. Each significant edit restarts learning, so results never settle. The other mistake is ignoring 'learning limited' on a budget too small for the chosen event.
The story
Meera splits ₹2,000 a day across one purchase-optimised ad set. Purchases cost about ₹600 each. Every two days she changes the creative or budget because results swing.
| Weekly budget (₹2,000 × 7) | ₹14,000 |
| Cost per purchase | ₹600 |
| Purchases in the week | about 23 |
|---|---|
| Results Meta mentions for exiting learning | about 50 |
| Daily budget for about 50 purchases | about ₹4,286 |
What looks right: frequent edits feel like active management.
What it hides: at about 23 purchases a week the ad set cannot settle, and each edit restarts learning. The swings she reacts to are partly caused by her own changes.
What to look at instead: whether the budget can produce enough of the chosen event. If not, combine ad sets, raise the budget or optimise for a more frequent step such as add-to-cart, then leave it alone for a week.
All figures for Meera's store are fictional and for illustration only.
Can this ad set get out of learning?
The example's numbers are filled in. Replace them with yours.
Unlikely to exit learning. Merge ad sets, raise budget, or optimise for a more frequent event.
Don't confuse it with
- Learning limited: Meta's status when an ad set is unlikely to get enough results to exit learning.
- Test period: a learning phase is the system's calibration, not your experiment.
Related terms
Want to use this at work? AI-led PPC Excellence is aFactor's 16-week program for paid ads.
Sources
- About the learning phase, Meta Business Help Center