Learning limited on Meta: what it means and how to fix it
Learning limited means the ad set is unlikely to get about 50 results a week. The budget arithmetic, and the fixes in the order Meta suggests.
Not spending or delivering Updated 9 November 2026 Checked against Meta’s Help Center
The short answer
An ad set is "learning limited" when Meta expects it to get fewer than about 50 optimisation events in the week after its last significant edit. It is not a penalty, but costs are usually higher and less stable. Fix it by combining ad sets, widening the audience, raising the budget to about 50 times your cost per result per week, or optimising for a more frequent event.
This is probably you if
- The Delivery column reads "Learning limited"
- Cost per result swings a lot from day to day
- The ad set gets a handful of results a week
Rather see it in your own account? Free, read-only audit. No sign-up.
Check my accountCauses, most likely first
Work down the list: rule each cause out or mark the one you find
- 1
The budget cannot buy 50 results a week
At a $30 cost per purchase, 50 purchases a week needs about $1,500 a week, or about $215 a day.
- How to check
- Multiply your cost per result by 50 and compare it with seven days of budget.
- Fix
- Raise the budget toward that figure, or move budget from other ad sets into this one.
Is this the cause?
- 2
Too many ad sets split the budget
Five ad sets on $50 a day each learn five times slower than one ad set on $250.
- How to check
- Count the ad sets in the campaign that target similar people.
- Fix
- Combine them. Meta says combining ad sets also combines what they learned.
Is this the cause?
- 3
The audience is too narrow
A small audience gives the system fewer chances to find people who convert.
- How to check
- Look at the audience size estimate when editing the ad set.
- Fix
- Widen targeting or use Advantage+ audience.
Is this the cause?
- 4
The event is too rare
If purchases are rare, no realistic budget gets 50 a week.
- How to check
- Compare weekly purchases with weekly add to carts or initiated checkouts.
- Fix
- Optimise for an earlier event, such as add to cart, until volume allows purchases.
Is this the cause?
- 5
Frequent edits restart learning
Each significant edit, such as large budget changes, new ads or targeting changes, restarts the week.
- How to check
- Open the ad set history and count edits in the past week.
- Fix
- Make changes in batches, at most about once a week.
Is this the cause?
Keep it from coming back
- Budget for 50 results a week before launching an ad set.
- Prefer fewer, broader ad sets.
- Batch edits so learning restarts once, not five times.
Did this guide help?
Terms in this guide
Frequently asked questions
Still unsure? Ask us anything. Talk to us
- Is learning limited bad?
- It is not a penalty, but it means Meta cannot optimise delivery well, which usually costs more per result.
- How long does the learning phase last?
- Until the ad set gets about 50 results in the week after its last significant edit, which is usually within a week if the budget allows it.
Related problems
- Meta ads stopped working after an edit You changed the budget, the audience or an ad and performance fell apart. Which edits restart learning, and how to recover.
- Meta costs went up after raising the budget You doubled the budget and the CPA got worse. Why scaling raises costs, and how to scale Meta ads without losing efficiency.
See what your Meta ads are really doing
Connect Meta in two minutes and see your first check while you watch. 14 days free, no card, and a flat price per ad account after that.