Skip to content
Adrythm
Ads and paid search

Learning phase

learning stage / learning limited / still learning / learning period

In short

The learning phase is the stretch after a change when an ad platform is still working out who to show your ads to, so results swing around. Meta reports it as learning, success or fail. Meta publishes no number of conversions needed to leave it. Google does state a figure for its own bid strategies.

It is still in the learning phase. That is the most common explanation you will hear for a disappointing first few weeks. Sometimes it is true. The problem is that most people have no way to tell a real one from a convenient one, so it is worth knowing exactly what each platform commits to in writing.

Meta reports the state of an ad set in three words. Learning means the ad set is still learning. Success means it exited the learning phase. Fail means the ad set is not generating enough results to exit. That third word is the one that matters, because the interface usually shows it as learning limited, which sounds like a phase rather than a verdict.

Now the part worth pinning down. Almost every article about this quotes a threshold of 50 conversions. Meta's own reference does not state it. That page documents a field called a conversions threshold for dynamic learning phase status, and it publishes no value for that field. It also documents a conversions count, and a timestamp of the last significant edit that caused the ad set to re-enter the learning phase. So Meta tells you a threshold exists, and does not tell you what it is.

The figure everybody quotes does appear in documentation, but it is Google's, and it is about something else. Google says it can take up to around 50 conversion events or 3 conversion cycles for the bid strategy to calibrate. Google also lists what puts a strategy back into learning: creating or reactivating a strategy, changing bid strategy settings, and adding or removing campaigns, ad groups or keywords. And it adds a line worth remembering, that its algorithms continue to learn even when the status no longer shows learning.

In practice

An ad set has run for three weeks and still says learning limited. Under Meta's own wording that is not a phase it will grow out of by waiting. It means the ad set is not producing enough results to exit, which usually points at the budget being spread too thin or the conversion being too rare. Waiting another month is the one response the status does not support.

Not the same as

A new campaign settling down
Any campaign is noisy at first. The learning phase is a specific reported state with a specific exit condition.
A bad campaign
Real instability early is normal. The question is whether the state ever resolves.

Why it matters to you

This is the single most common thing said to justify a poor month, and until now you had no way to test it. You do. Ask which state the ad sets are actually in, and when the last significant edit was. If someone changed the budget last week, the clock restarted, and repeated restarts are their own problem rather than an explanation for one.

What to ask or check

  1. 01Are the ad sets showing learning, success, or learning limited right now?
  2. 02When was the last significant edit, and what was it?
  3. 03If we are told a conversion threshold, where is that number published?

What people get wrong

That Meta requires 50 conversions in 7 days to exit the learning phase. Meta's own reference names a conversions threshold field and publishes no value for it. The 50 figure appears in Google's documentation, about calibrating a bid strategy, which is a different thing on a different platform.

Red flags

  • An ad set stuck at learning limited for weeks, described as a phase that needs more time.
  • Frequent budget or targeting edits, each of which restarts the clock, followed by the learning phase being offered as the reason for poor results.

Where you will see it

As a delivery status on the ad set in Meta, and as a bid strategy status in Google Ads.

Conversion rate

Conversion rate is the share of ad clicks that turned into something you count as a result, like a phone call or a filled in form. Google works it out as conversions divided by ad interactions. Microsoft divides by clicks. The two are not the same number, so compare platforms carefully.

Cost per click

Cost per click is what you pay each time someone clicks your ad. You set a maximum you are willing to pay, and the auction usually charges less: Google says advertisers are often charged less, sometimes much less, than that maximum. It is the unit your ad invoice is built from.

Smart Bidding

Smart Bidding sets a bid for every individual auction rather than per keyword. Google calls this auction-time bidding and says some of its signals are exclusive to it. Microsoft is blunt about the trade: with auto-bidding you can only set a bid at campaign level, and the model computes the rest.

Quality Score

Quality Score is a 1 to 10 diagnostic number reported against each keyword, estimating how relevant your ad, keyword and landing page are. Both Google and Microsoft state plainly that the score itself is not used in the auction. Ad quality does affect what you pay. The number in your account does not.

Click-through rate

Click-through rate is the share of people who saw your ad and clicked it, worked out as clicks divided by impressions. Google gives the example of 5 clicks from 100 impressions being a 5 percent rate. It measures interest in the ad, not whether the click was worth anything.

Search terms report

The search terms report lists the actual things people typed before your ad appeared, next to what each one cost. It is where you find out whether you are paying for searches you would never want. Google omits low volume terms from it for privacy, so it is not a complete list.

Want this explained against your own numbers?

Twenty minutes, a straight answer, and no follow-up sequence if you decide not to work with us.