What is the Meta learning phase, and what restarts it?

You launch an ad, and Ads Manager says "Learning". Results jump around. Costs look high. That is the learning phase. This page explains it in plain words, using Meta's own Help Center pages. We read them on October 12, 2026.

The short answer

Here is how Meta's learning phase page puts it. It is "the period when the delivery system still needs to learn about how an ad set may deliver and perform."

It starts when you "create a new ad or ad set or make a significant edit to an existing one."

While it runs, the Delivery column reads "Learning".

Ad sets leave the learning phase when they "can deliver stably." Meta says when that tends to happen.

"This usually occurs after about 50 results in the week after the ad set's last significant edit."

Why it matters

Meta is clear about the cost. Meta says that while learning, "ad sets are less stable and usually have a higher CPA." CPA means cost per action, like cost per purchase.

So every time an ad set goes back into learning, you pay for some days of shaky results. That is fine when you are testing something new. It hurts when you only wanted to change a picture for a sale.

What restarts the learning phase

Meta's significant edits page says: "Only a significant edit causes an ad set to reenter the learning phase." It lists these as significant:

• "Any change to targeting"
• "Any change to ad creative"
• "Any change to optimization event"
• "Adding a new ad to your ad set"
• "Pausing your ad set for 7 days or longer"
• "Changing bid strategy"

Some changes may or may not count, "depending on the magnitude of the change."

These are the spending limit, the bid or cost goal, and the budget. Meta's example: a budget move from $100 to $101 "isn't likely" to restart learning, but $100 to $1000 may.

Learning vs Learning limited

Meta says the status reads "Learning limited" when an ad set "isn't getting enough results to exit the learning phase."

How to check if an ad set is learning

1. Open Ads Manager and pick the ad set.
2. Look at the Delivery column. It reads "Learning" or "Learning limited" during the phase.

   

3. Add the Last significant edit column if you do not see it. Click Columns to add it.
4. Check the Results column to see how many results came in since that edit.

Meta's advice during learning

Meta's learning phase page gives four tips:

• "Wait to edit your ad set until it's out of the learning phase."
• "Avoid unnecessary edits that cause ad sets to re-enter the learning phase."
• "Avoid high ad volumes." Many ads and ad sets split what the system learns.
• "Use realistic budgets." And avoid frequent budget changes.

Meta also says you "shouldn't try to avoid the learning phase completely." Testing new ads is how you find better ones. The goal is to avoid needless resets, not all learning.

Needless resets: the promo problem

Here is where many stores lose money. Every promo (Black Friday, Boxing Day, New Year) gets a new ad with new pictures. Each new ad starts in learning. A short promo can end before the ad settles.

For catalog ads (also called dynamic product ads, or DPA) there is another option. The pictures come from your catalog, so you can change them in the catalog instead of making a new ad. Read how to change catalog ad images without new ads for the steps. With Convinci you make each promo's look ahead of time and your catalog switches to it. You launch no new ad and edit nothing in Ads Manager. On the Pro plan each look can start and end on dates you pick.

For the narrower question of catalog pictures, read does changing catalog images reset learning.