Facebook ads Learning Phase: 9 Important Questions Answered!
Are you aware that there is a machine learning algorithm behind every Facebook ad that runs?
It helps you fine-tune your ads/ad sets and optimizes them for the correct delivery.
But initially, when you run your ad, this algorithm is clueless regarding how to target the right audience for you.
Thus you may observe that :
Your ad performance is less stable, and
CPA (cost-per-action) is worse.
These effects last for a short time and are a learning curve for the algorithm.
This helps optimize the algorithm according to Facebook standards.
But during this span, many of the new marketers tend to panic as they are unaware of :
Facebook Ads Learning Phase,
Facebook ad learning limited, and
Ways to quickly exit the learning phase
That is why I shortlisted 9 important questions, the answers to which will give you a thorough understanding of the Facebook ads learning phase.
It will also help you make better decisions about your campaigns and optimize them.
Let us dive in right away.
It is the initial time where Facebook is yet to learn about your ad set and explore ways to deliver it in the best possible way.
During this phase, Facebook figures out how to optimize your ads based on your goals, audience, budget, etc.
To check if your ad is in the learning phase, visit your ads manager and check the delivery column for status displayed as “Learning”.
If not in the learning stage, the status will be either “active” or “learning limited”.
8 Million+ Ready Automations
For 750+ Apps
The learning phase starts when you launch a new ad/adset or make a significant change to an existing one.
Here, significant changes include:
Changes to targeting/ ad creative/ Optimization event
Adding a new ad to your ad set
Pausing your ad set for 7 days or longer
Changing your bid strategy
Depending on their magnitude, the following changes may also be treated significantly:
Ad set spending – limit amount
Bid control, cost control, or ROAS(Return on ad spend) control amount
Each time the delivery system displays your ad, it
Learns about the best people and places to show your ad
And as you get conversions, Facebook improves in delivering your ads to the right audience.
Well, the delivery system never stops learning.
It keeps on improving and finding better ways to deliver your ads.
However, your ad set exits the learning phase once the performance stabilizes.
It usually happens after around 50 optimization events (in a week) since its last significant edit.
Wondering what an optimization event is?
It is nothing but a metric that shows how well your bid strategy meets your goals (based on your current optimization criteria).
For example, if you optimize your ad with a goal of purchase, then your adset comes out of the learning phase after generating around 50 purchases in a week.
Now, this 50 is not a fixed value always.
It is an estimated value that varies according to:
Characteristics of the ad,
Market conditions at that time
Once your adset comes out of the learning phase, it goes into either of the following states:
Active: It is the state where your ad is up and running with improved accuracy
Learning Limited: It is the state where the delivery system could not finish its learning within a week and hence cannot exit the learning phase.
This generally happens if :
Your ad did not get the required optimization events, or
The delivery system predicts that your ad will not receive enough optimization events in the future
Your learning phase comes to an end once you enter the active state.
After this, you can use the available data to make an informed decision about your ad set as below:
Satisfied – Keep your ad running or increase its budget
Unsatisfied – Edit your ad set and try to improve its performance or pause it
But what if you enter the Learning Limited state?
Seeing your ad in a Learning Limited state is not what you wish!
You mostly end up in Learning Limited state because of :
small audience size,
low bid or cost control,
high auction overlap,
infrequent optimization event, or
running too many ads at the same time
You can come out of the ‘Learning Limited’ state by trying out the following options:
Combining ads sets and campaigns: When you have multiple ad sets under a campaign, your budget spreads across them, and it becomes difficult to achieve the recommended 50 optimization events in a week. You can combine such ad sets to reduce their number and get stable results to come out of the phase.
Expanding your audience: Expanding your target audience set helps your ads reach more people. It further increases your optimization events and helps you get out of the state.
Raising your budget: If your budget is too low to receive 50 optimization events, then increasing it may help out.
Changing your optimization event: You can change your optimization event to the one that occurs frequently or the one people prefer.
For example, you can change your optimization event from ‘purchases’ to ‘adding to the cart’ (because people most likely opt for adding to the cart rather than purchases). But changing this may conflict with your goal.
Like in the above example, you want the person to purchase your items. If you change it to adding to cart, that may increase your optimization events. You may also come out of the ‘Learning Limited’ state, but you will land up with people adding products to their cart and not buying them! So you need to think about the results as well.
After trying out the above options, if a Learning Limited ad set receives 50 optimization events since your last significant edit, it will move from Learning Limited to Active state.
‘Learning Limited’ is not a penalty, nor does it mean that your ad has stopped running.
So relax for now!
It is more like a warning from Facebook to the advertisers.
It alerts you to make changes to your ad and optimize it according to Facebook standards.
To answer about the negative impact, I can say that you may not get the expected results as your ad fails to optimize according to Facebook standards in this state.
Or, you may end up spending more without much ROI.
But that again depends on your ad, target audience, and goal.
For example: If you are an automobile company running an ad for your SUV product, your target audience may be small, and your ad may enter Learning Limited state.
But besides being in the Limited Learning state, you may get the right leads from the narrow audience that you set, and that too with a good ROI!
If you wish to avoid entering the Learning Limited stage or exit the Learning phase quickly, you need to follow these best practices:
Do not edit your ad set when in the learning phase: Initially, the ad performance is less stable and does not always indicate future results. It may tempt you to edit your ad, but wait till the learning phase ends! Why so? That’s because once you edit your ad/adset/campaign it resets the learning of the algorithm and delays the optimization. So everything starts over again!
Avoid high ad volumes: It is easy for the delivery system to learn more when you have few ads and ad sets. Increasing them reduces its efficiency.
Use realistic budgets: Your budget must be sufficient enough to get at least 50 optimization events in a week. Working with small budgets fails to get them.
The learning phase indeed is a vital part of your ad lifecycle. It is during this time that your ads optimize for performance and delivery.
While it seems like a trial and error process at first, the overall experience ensures that you get optimal results.
So, if you’re running several different campaigns, you can expect more stable performance as your ads/ad sets mature. All thanks to the learning phase!
Keep learning, Keep optimizing!
Yes, you can come out of the learning phase by :