# How to exit Learning limited on Meta ads

> To exit Learning limited, get the ad set to about 50 results in a week. Five fixes from Meta, the columns to add in Ads Manager and a worked example.
- **Author**: Marxx
- **Published**: 2026-10-24
- **Category**: Meta
- **URL**: https://marxx.ai/library-guides/how-to-exit-learning-limited

---

To exit Learning limited, get the ad set to about 50 results in a week, starting by combining ad sets. Meta's other fixes are to widen the audience, raise the budget, raise the bid or cost control, or pick an optimisation event that happens more often.

Meta says an ad set becomes learning limited when it is unlikely to get about 50 optimisation events in the week after its last significant edit. It calls this a signal, not a penalty. ([Meta's page](https://www.facebook.com/business/help/269269737396981), checked 2026-10-20)

## Find the cause first

In Ads Manager, the **Delivery** column reads **Learning limited**. Add two columns to see why.

1. Click **Columns**, then **Customise columns**, tick **Optimisation events** and click **Apply**.
2. Add **Last significant edit** the same way.

Meta lists these common causes: a small audience, a low budget, a low bid or cost control, high auction overlap, an infrequent optimisation event, and too many ads running at once.

## The five fixes

| Fix | Use it when | Watch out for |
|---|---|---|
| Combine ad sets and campaigns | Many ad sets each get few results | You get less control over each slice |
| Widen the audience | Targeting is narrow. See [broad targeting](/glossary/broad-targeting) | Quality can shift |
| Raise the budget | The budget is too small for your cost per result | You spend more |
| Raise the bid or cost control | A cap is below what the auction needs | Your cost per result may rise |
| Change the optimisation event | Purchases are rare. Meta suggests add to cart as an example | A weaker signal than a purchase |

If a learning limited ad set gets enough results since its last significant edit, Meta moves it back to Active.

## Steps

1. Filter your ad sets by Delivery status **Learning limited**.
2. Add the **Optimisation events** and **Last significant edit** columns.
3. Work out the gap between this week's results and 50.
4. Choose the lowest-risk fix. Combining ad sets is usually it.
5. Make all your edits in one batch, then leave the ad sets alone for a week. Each significant edit restarts the clock.

## A worked example

Say you run four ad sets at INR 1,500 a day each, and a purchase costs you INR 500.

- Each ad set spends INR 10,500 a week, so it gets about 21 purchases. That is far under 50.
- Combine all four into one ad set at INR 6,000 a day. It spends INR 42,000 a week and gets about 84 purchases. That is over 50.

If you must keep four ad sets, switch each to add to cart. At INR 100 an add to cart, INR 10,500 buys about 105 events. Each ad set clears 50, but add to cart is a weaker signal, so confirm results on purchases.

See [the learning phase](/glossary/learning-phase) for how it works, and [how to build a creative testing budget](/guides/how-to-build-a-creative-testing-budget) to size budgets for new tests.

## Mistakes to avoid

- **Editing every day.** Meta says unnecessary edits make ad sets re-enter learning.
- **Changing audience, budget and creative on the same day.**
- **Treating the label as a penalty.** It is a warning that results will be less stable.
- **Splitting a small budget across many ad sets.** Each gets too few results.

If many ad sets are stuck, the structure is probably the problem. Review the whole account with [how to audit a Meta ad account](/guides/how-to-audit-a-meta-ad-account). To decide where budget should sit, see [CBO vs ABO](/compare/cbo-vs-abo) and [Advantage+ vs manual campaigns](/compare/advantage-plus-vs-manual-campaigns).

## FAQ

### Is Learning limited bad?

Meta says it is not a penalty. It means the ad set is unlikely to get enough results to optimise well, so your budget is not working as hard as it could.

### How long does it take to exit?

Meta looks at the week after your last significant edit. If the ad set reaches about 50 results in that week, it exits.

### Should I turn off a learning limited ad set?

Not by default. Try the fixes first. If it still cannot get results at a cost you accept, merge it into another ad set or turn it off.

---
- [More Meta articles](https://marxx.ai/library-guides/category/meta)
- [All articles](https://marxx.ai/library-guides)