A few years ago, Facebook rolled out the “Learning Phase” which visibly shows up in the ad set’s view under the delivery status column. The term “learning” pertains to the learning activity of the Facebook AI bot that uses machine learning (ML) to deliver ads. Facebook has migrated to a very complicated self-learning algorithm that deliver ads to the right audience thereby optimising the advertiser’s goal but maintaining the experience of the users.
When does the learning phase start and when does it end?
The learning phase immediately starts when a campaign is newly set up or a new ad set is added to an existing campaign and continues up to 7 days based on the attribution window that is set. Recently, Facebook has reduced the default attribution setting from 30 days to 7 day click or 1-day view, because it believes that if a campaign achieves 50 optimisation events within this time frame it will simply not deliver the desired results. It is an indication that the advertiser should make relevant edits to the campaign for better results. An optimisation event in simple terms is a goal that your campaign wants to achieve. It could be clicks, landing page views, leads or conversions. The faster you complete the learning phase the better your campaign will perform in terms of optimisation events (goals set). Therefore, it is advised not to edit a campaign significantly during this phase.
Can I or should I edit a campaign during the learning phase?
The simple answer is that you shouldn’t alter any campaign settings, because if you do the learning may reset and it will start all over again thus wasting precious campaign time. If it’s a time sensitive campaign, it will impact your time to market.
There may be instances when an advertiser is tempted to edit the campaign during the learning phase such as the ones below,
- He simply cannot wait for 7 days to edit when he/she is assured that an important parameter was missed out during the setup phase which in most cases is the audience targeting.
- CPAs are running at an extremely high cost.
- The learning phase is painfully slow.
Small edits are fine but significant changes should not be done. To know what Facebook considers as significant edits read on.
How should I edit the campaign during the learning phase without affecting the learning progress of the campaign?
It is always safer to clone the existing ad set in the original campaign and make whatever edits need to be done and run both the ad sets alongside and monitor the costs.
How to exit the learning phase quickly?
As discussed previously, the faster you move out of the learning phase the quicker your campaign will be built for delivery. The creative and target audience are the two most important factors at play. Once they resonate with one another, the Facebook AI bot takes into consideration your bid and budget and works with a multiplier effect. It is for this very reason that the budget that you set should be in multiples of the CPA. It is always advisable to front load your budgets in the learning phase so that you can the learning phase faster. For example, if you set your cost control at an amount “A” then your budget should be “n” times A. The ideal budget would be 20 X A. Once the learning phase is successfully exited, the budget can be strategically reduced to adjust with your business targets. Please note that adjusting bids and budgets are not considered as significant changes and will not reset the learning phase provided the reduction is done in smaller proportions. Up to +/-5-10% is fine but +/-50 or 100% would be significant.
What factors will impact the campaign’s learning phase?
The following significant edits can impact the learning phase:
- Any change to targeting
- Any change to ad creative
- Any change to optimisation event
- Adding a new ad to your ad set
- Pausing your ad set for seven days or longer (the ad set re-enters the learning phase once you unpause the ad set)
- Changing bid strategy
Note: When using campaign budget optimisation, switching your campaign bid strategy might cause multiple ad sets within the campaign to re-enter the learning phase.
A change to any of the following areas may or may not be significant, depending on the magnitude of the change:
- Ad set spending limit amount
- Bid control, cost control or ROAS control amount
- Budget amount
Note: When using campaign budget optimisation, adjusting your campaign budget might cause multiple ad sets within the campaign to re-enter the learning phase.
What is learning limited and how do I fix it?
When the Facebook AI bot cannot achieve the 50 optimisation events it will hint as “learning limited” under the delivery column. Here are some simple hacks to get rid of “learning limited”
- Combine ads sets and campaigns
- Expand your audience
- Raise your budget
- Raise your bid or cost control
- Change your optimisation event
So, we hope to have helped you from a lot of guess work when dealing with the Facebook learning phase and saved those wasted go to market hours. Let me know in the comments below if you have questions or share your experiences with your Facebook campaign’s learning phase.