The learning phase in Meta ads: what resets it and how long it takes

This guide is built from the Meta Business Help Center: About the learning phase, Significant edits and learning phase, About learning limited, Optimization events, Results, About managing ad volume, Combine ad sets and campaigns, Understand fluctuations in ad performance and Your campaign hasn't delivered enough since it started. Checked on 10 November 2026. Meta changes these pages from time to time, so check them before you rely on a detail. How Ads Manager is organised and how to set up a campaign is in our Meta ads basics guide; this page covers the learning phase only.

What the learning phase is

Meta defines the learning phase as "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 when you make a significant edit to an existing one. During that time the system is "exploring the best way to deliver your ad set", and the Delivery column in Ads Manager reads "Learning".

Learning happens at the ad set level. Each ad set learns on its own, based on the optimization event you chose for it: the outcome Meta bids on in the auction, which "can be different from your campaign objective". You can, for example, run a sales campaign with an ad set optimized for link clicks.

Meta is open about the cost of this period: "During the learning phase, ad sets are less stable and usually have a higher CPA." It also says you should not try to avoid learning altogether, because "the learning phase is necessary to help the delivery system best optimize ads". New creative and new tests will always trigger it.

How long it takes: about 50 results in a week

There is no fixed number of days. Ad sets "exit the learning phase as soon as they can deliver stably", and Meta says this "usually occurs after about 50 results in the week after the ad set's last significant edit." The Optimization events metric page puts it the same way: "We recommend around 50 optimization events, but some ad sets stabilize earlier."

Two things follow from that rule:

• The clock is tied to the last significant edit, not to when the ad set was created. Edit it in a way that counts, and the count starts again.
• What counts is your optimization event. An ad set optimized for purchases needs about 50 purchases in a week; one optimized for landing page views needs about 50 of those. A rare, expensive event takes more budget to reach the same count.

Shops ads have their own threshold. For them, Meta says "you need a minimum of 17 purchases through your website and 5 through Meta for the learning phase to complete."

Results are counted with your attribution setting. Meta's default is "7-day click, 1-day view and 1-day engagement conversion", and a different setting changes what gets counted. Our Meta attribution settings guide explains the options.

How to see where an ad set stands

The Delivery column shows the status: Learning, Learning limited or Active. To see how close an ad set is to leaving learning, Meta suggests two reporting columns:

1. In Ads Manager, select the ad set and look for the Last significant edit column. It "shows when you last made a change that could reset the learning phase." If it is missing, click Columns and add it.
2. Check the Results column for the number of results since then. For the event the ad set is optimized for, you can also add the Optimization events column (Columns, Customize columns, tick Optimization events, Apply).

Meta notes that it is "useful to evaluate the number of optimization events you've received since your last significant edit", especially before you use those costs to set a cost per result goal.

What resets it: significant edits

Not every change sends an ad set back. Meta says every edit "has some effect on delivery, but not every edit causes the ad set to reenter the learning phase. Only a significant edit causes an ad set to reenter the learning phase."

These always count as significant edits:

• Any change to targeting
• Any change to ad creative
• Any change to the optimization event
• Adding a new ad to the ad set
• Pausing the ad set for 7 days or longer (it re-enters learning when you unpause it)
• Changing the bid strategy

These may or may not count, "depending on the magnitude of the change":

• The ad set spending limit amount
• The bid control, cost per result goal or ROAS goal amount
• The budget amount

Meta's own example: raising a budget from $100 to $101 "isn't likely" to reset learning, while going from $100 to $1000 means "one or more ad sets may reenter the learning phase." The page gives no exact threshold. For ads about social issues, elections or politics, changing the disclaimer on the ad is also a significant edit.

How the bid strategies themselves work, and Meta's advice on changing goal amounts in small steps, is in our Meta bid strategies guide.

Advantage+ campaign budget and learning

With Advantage+ campaign budget (the budget set at campaign level), some campaign-level changes reach every ad set. Meta says that switching the campaign bid strategy, or adjusting the campaign budget, "might cause multiple ad sets within the campaign to reenter the learning phase."

Meta's answers to the common questions about it:

• Initial learning takes "the same time as ad set budgets."
• Moving budget between ad sets does not reset learning: "ad sets within the campaign won't reenter the learning phase as budget is distributed."
• A significant edit made at the ad set level only affects that ad set. Other ad sets in the campaign do not re-enter learning.
• Adding a new ad set to the campaign does not send the existing ad sets back into learning.

More on campaign-level budgets and Meta's automated setups is in our Advantage+ campaigns guide.

What to do while an ad set is learning

Meta's best practices for this period come down to four points:

• Wait before editing. Results during learning "aren't necessarily indicative of future performance", and editing an ad, ad set or campaign during learning resets it and delays optimization.
• Skip unnecessary edits. "Only edit your ads or ad set when you have reason to believe that the edit should improve performance."
• Avoid high ad volumes. With many ads and ad sets, "the delivery system learns less about each ad and ad set". Combining similar ad sets also combines what they learn.
• Use realistic budgets. A very small or inflated budget gives the system an inaccurate picture. Set one large enough to get enough results and avoid frequent budget changes.

On ad volume, Meta explains the mechanism: when you run too many ads at once, "each ad delivers less often. This means that fewer ads exit the learning phase, and more budget is spent before the delivery system can optimize performance." Its suggested alternatives include fewer ads per ad set with several creative assets in each (up to 10 per ad), multiple text optimization instead of separate ads for text variants, and Advantage+ placements with asset customization instead of one ad set per placement.

Learning limited and how to fix it

An ad set shows "Learning limited" when it "is unlikely to receive about 50 optimization events in the week after your last significant edit." This can be flagged any time after you create the ad set or make a significant edit. For Shops ads, it happens when the ad set has not generated 17 website purchases and 5 through Meta after 7 days.

Meta stresses that it "isn't a penalty". It means your budget "isn't being spent effectively because the ad delivery system can't optimize performance with your current setup." The usual causes, per Meta:

• A small audience
• A low budget
• A low bid or cost control (hover over the status to see a tooltip if this is the cause)
• High auction overlap, where several of your ad sets compete for the same people
• An optimization event that happens rarely
• Running too many ads at the same time

Meta's fixes match those causes:

• Combine ad sets and campaigns. Meta's example is three ad sets with the same objective and creative, split by interest in dogs, dog toys and dog collars; combining them into one broader ad set "could improve performance". Use reporting breakdowns (age, gender, country, region and others) instead of separate ad sets when you only want reporting by segment.
• Expand the audience. More people means more chances to complete the event.
• Raise the budget. Elsewhere Meta gives a general rule that a daily budget "should be at least 10 times the average cost of your performance goal" ($50 a day for $5 link clicks, in its example).
• Raise the bid or cost control if it is set too low to win enough results.
• Choose a more frequent optimization event, for example "move from purchases to add to cart."

Once an ad set "receives enough optimization events since your last significant edit, it will move from Learning limited to Active."

Reading results during learning

Because delivery is still being explored, early numbers are noisy. Meta's general guidance on performance fluctuations applies here:

• Analyse results for the whole campaign if you use Advantage+ campaign budget, or for the ad set if you use ad set budgets.
• Be careful with hourly breakdowns, where swings "average out over time". Meta suggests looking at "at least a full week".
• Judge the cost of your optimization event, not CPM. When optimizing for conversions, the system may take more expensive impressions if that gives cheaper conversions.
• Costs can exceed a cost per result goal or bid cap, "especially during the learning phase."

If you want to compare two setups fairly instead of reading before-and-after numbers, use a split test; see our Meta A/B testing guide.

How we handle it

This is our own practice for client accounts, not a Meta rule:

• Before launch, we check that the budget can plausibly buy about 50 optimization events a week at the expected cost. If it cannot, we pick a more frequent event or combine ad sets rather than launch something that will sit in Learning limited.
• We batch changes. Creative swaps, targeting changes and budget changes go in together, once a week at most, so each ad set gets one reset rather than several.
• We add the Last significant edit and Optimization events columns to every saved view, so the whole team can see why an ad set is still learning.
• We check conversion tracking first. If events are missing or counted twice, the ad set learns from the wrong signal; see our Meta Pixel and Conversions API guide.

Frequently asked questions

How long does the Meta learning phase last?

There is no fixed duration. An ad set usually leaves learning after about 50 results for its optimization event in the week after its last significant edit, and some stabilise earlier.

Does increasing the budget reset the learning phase?

It depends on the size of the change. Meta says a small change such as $100 to $101 is unlikely to, while a large one such as $100 to $1000 may send one or more ad sets back into learning.

Does adding a new ad restart learning?

Yes. Adding a new ad to an ad set is on Meta's list of significant edits, as is any change to ad creative or targeting.

Does pausing an ad set reset learning?

Only if it stays paused for 7 days or longer. In that case the ad set re-enters the learning phase when you turn it back on.

Is Learning limited bad for my ads?

Meta says it is not a penalty. It means the ad set is unlikely to get about 50 optimization events a week with its current setup, so the budget is not used as effectively as it could be.

For more on Meta advertising, see our guides to Meta ads basics, bid strategies, Advantage+ campaigns, A/B testing, lookalike audiences and attribution settings. We set up the conversion tracking the learning phase depends on as part of our website development work. See all our services, or browse more guides.