How the Facebook Ads Delivery System Works
As many of you in the digital marketing world already know, the Facebook ads delivery system uses machine learning to optimize for your results. Basically, each time a campaign ad is served or displayed, the delivery system learns more about the best people to target, times of day to show the ad, and placements and creatives to use. The more an ad is shown, the better the delivery system becomes at optimizing the ad’s performance.
For that reason, it’s important to understand what Facebook calls “The Learning Phase.” In a nutshell, the learning phase is the period when Facebook’s delivery system is still learning and gathering information about an ad set. During the learning phase, the delivery system (also referred to as Facebook’s algorithm) is understanding the best way to deliver a particular ad based on a number of parameters, and therefore, the ad performance is not yet optimized. What is important to understand is that the learning phase not only affects NEW ads but it may also affect EXISTING ads or ad sets that go through significant edits.
Some of the consequences of the lack of ad stability during the learning phase include ads having a higher CPA and overall campaign costs being higher across the board. Obviously, all advertisers want their ads to exit the learning phase as soon as possible. According to Facebook, “performance stabilizes after an ad set receives around 50 optimization events within a 7-day period.” Note: This is assuming that you have a 7 day conversion window.
One strategy to reduce spend in the learning phase is to avoid editing an ad set or ad until it has exited the learning phase. This ensures your optimization decision is based on results that are more indicative of future performance. If you have multiple edits to make, batching them all at once ensures that learning is only reset one time.
So the next logical question is what type of edits may trigger (again) the learning phase on ads and ad sets that have already been running for some time after existing the initial learning phase. Here’s a quick list of factors:
Frequent Edits: Manual tweaks and optimizations (edits) are the main reason why ad sets will remain under the learning phase longer than usual.
Campaign Level Adjustments: Budget and Bid cap amount changes (depending on magnitude) and a change in bid strategy.
Ad Changes of any type: This is clear. You change your ad, the system needs to learn again.
Ad Set changes to targeting, placement, optimization events, adding new creative, as well as a change in bid amount or budget (depending on magnitude) will trigger the learning phase again.
Adding a new ad to your ad set. Obviously, this will cause Facebook to start testing the new ad within the set, causing the entire set to go into learning phase again.
Having too many ad sets: This will decrease the frequency of delivery of each ad set, therefore, taking longer for Facebook to optimize each ad set for performance. The result: more budget spent before all of the ad sets exit the learning phase.
Ad Set Pauses: Pausing your ad sets for more than 7 days, will trigger the learning phase again once you re-start using your ad sets.
The one factor that is a bit more nebulous is the budget change and the bid cap change. Facebook states that it has to be a significant change for it to cause the campaign or ad set to go into learning phase again. However, there’s no explanation about how much of a change is “significant” in their system. We typically limit budget changes to a 10-15% range to be safe. Or, you can also use CBO Ad sets if you want to have a bit more latitude with your budget. In the case of CBO, Facebook seems to be much more flexible with ad set budget changes without causing any resets of the learning phase.
As much as we advocate testing different ads and ad sets to ensure the best ROAS, it is important to understand the rules of the game to do it without backfiring and adding additional CPA to our campaigns. Experience is key and knowing how to navigate any type of change without resetting the learning phase is a sure way to keep your campaign costs under control while getting the most out of your individual ads.
As always, if you need any assistance with your eCommerce marketing, our MGR Team will be happy to chat with you one-on-one. Use this link to contact us and set up your free consultation.
Thank you for reading. Until next time, this is Manuel Gil del Real (MGR).