Over the years we heard a lot of advertisers say the learning phase negatively affects the cost-per-action (CPA) and overall ad performance. But hey, we don’t blame them or you for thinking so. Facebook actually tells you the same thing.
“During the learning phase, the delivery system is exploring the best way to deliver your ad set – so performance is less stable and cost-per-action (CPA) is usually worse.”
But is that really the case?
Let’s go through the data and discuss everything you need to know…
What is the learning phase?
The learning phase is the period that starts when you launch a new ad set or make a significant edit to an existing one.
During the learning phase, Facebook is essentially learning what is the best way of optimizing your ads based on your optimization goals, creative, audience, etc. So, as you start getting more and more conversions, Facebook is becoming better at knowing when and where someone is most likely to convert.
The question, though, is whether your CPA is becoming better (decreasing) as well?
How long does the learning phase last?
Facebook usually needs around 30-50 conversion actions within 7 days to exit the learning phase.
So, if you’re optimizing for landing page views you’ll need 30-50 landing pages views. While, if you’re optimizing for purchases, you’ll need to generate 30-50 purchases to exit the learning phase.
After those 7 days, your ad sets can either go from “learning” to “active” or “learning limited”. Later in the article, we’ll explain what “learning limited” means and whether you should be concerned about it.
Since we mentioned optimization, we always recommend optimizing for conversions, as you want to find people who will purchase from you. Additionally, we’ve found that optimizing for conversions not only decreases your CPA but also significantly increases the conversion rate (CR).
When does the learning phase restart?
According to Facebook, the learning phase restarts as soon as you make any changes to the following:
- Budget (increasing it more than 20%)
- Bid amount
- Bid strategy
- Optimization event
- Adding a new ad
- Bid strategy
- Bid amount
- Budget (increasing it more than 20%)
- Pausing your ad set for over 7 days or longer
- Any change to the ad creative (visual, ad copy, headline, description, call to action, landing page url)
What is Learning Limited?
So basically, if you don’t hit those 30-50 conversions within 7 days, your ad sets will go from “learning” to “learning limited”.
Facebook recommends you take a couple of steps to go from learning limited to active, but we found that following any of those will decrease your performance even more.
For example, one of their tips includes changing your optimization event to an action that occurs more frequently. So basically, switching between the “purchase” and “add-to-cart” event.
But why shouldn’t you do so?
While you might get those 30-50 conversions much quickly when optimizing for add-to-carts, the quality of this audience is not going to be as good as when you optimize for purchases.
By optimizing for add-to-carts, you’re telling Facebook to find people that are most likely to add an item to the cart, but not necessarily purchase from you – which is what you ultimately want.
The bottom line is that the “Learning Limited” phase doesn’t change the way your ads are being optimized. If you want to target an audience that will actually convert, don’t blindly follow Facebook’s recommendations and always optimize for purchases.
Doing anything other than that will only decrease performance even more than if you were in learning limited.
What we recommend
Simplify your ad account structure
The golden rule for both Facebook and Google advertising is to avoid having an overly complicated ad account. Having 10 ad sets with tons of ads in it will neither help you nor Facebook.
Your account is going to be overloaded with data you can’t draw conclusions from, while Facebook will not truly know what to optimize for.
Additionally, exiting the learning phase will not be that easy and quick.
For example, if you have 10 completely different ads in one ad set, Facebook will need to test out all the 10 options to see which one will perform the best on your target audience.
So you’ll basically need 500 conversions (10 ad sets x 50 conversions) to know what is the best performing ad set.
So yeah, having a simplified account structure is quite beneficial. What we found works best for the ad setup is having 3 to max 5 ad sets with 1 ad per ad set. In case you’d like to learn more about the best practices we recommend when setting up your account, check out this post.
Have a realistic ad budget
As we mentioned before, to exit the learning phase, you’ll need to generate 30-50 conversions within 7 days. Therefore, you’ll need to have a realistic daily ad budget. And when we say realistic, we mean, not too small.
Say your daily budget is $15. To generate those 50 conversions and exit the learning phase, your CPA should be around $2.1. You might even hit those 50 conversions with that budget when optimizing for landing page views, but since we always want to optimize for purchases, you might need to increase your budget.
The point is you don’t want to start with just $10 or $20 a day. See how much money you can actually spend to acquire a new customer and multiply that by 50-60 conversions to get an idea of what should be your daily budget.
Increase your audience size
Lastly, keep an eye on your audience size. When targeting a smaller audience, your frequency will go up much faster, resulting in ad fatigue. So, as a result of that, your chances of exiting the learning phase within those 7 days will not be that high.
Most of our customers nowadays use broad targeting anyway, as it happens to give the best results. So if you’re still focusing on interest targeting or lookalikes, we’d definitely recommend expanding your audience and going as broad as possible.
Results during learning phase vs. after
Now let’s look at the actual data to see if the CPA and overall results really improved once ad sets exited the learning phase.
Note: The table below includes aggregated data from multiple ad accounts and ad sets.
Ad set status
In the learning phase
Exited learning phase
*CAC= customer acquisition cost (same as CPA)
Surprisingly, as we can see from the results, ad sets that were still in the learning phase had a 10% lower CAC.
Similarly, looking at the example below (smaller sample size) we can see that the CAC was again much lower on ad sets that were still in the learning phase.
Ad set status
In the learning
This just proves that, once you exit the learning phase, the CPA doesn’t magically become better – which is the exact opposite of what Facebook tells us.
However, what you always want to measure is whether the conversion rate increase will be higher than the expected cost per click increase. More often than not, we see quite the opposite. Ads perform better during the learning phase due to a lower CPC and Facebook entering less competitive auctions.
Audience saturation and the learning phase
The audience saturation graph is part of the Facebook Ad Inspect Tool that helps you understand the relationship between your ad frequency and CPA.
Looking at the graph below, you can see that as ad sets exit the learning phase, the audience saturation starts to increase. This especially happens if you’re targeting a lookalike or interest-based audience that is generally small.
What happens is that your frequency starts to increase faster than usual, resulting in ad fatigue and audience saturation. So in conclusion really, exiting the learning phase isn’t a necessity since sooner or later you’re going to need to expand your audience.
Note: the yellow area of the graph represents the learning phase.
What about you?
Have you been seeing the same results across your campaigns? Have any questions or thoughts on this topic? Let us know in the comments below or drop us a line at [email protected]. We are always happy to help or discuss the best way of optimizing your ads.