The New Campaign Learning Phase

Red Hand At Crosswalk

When you start a new marketing campaign it’s extremely easy to be anxious for results. You are spending money after all! While certain campaigns types for certain verticals can produce relatively instant results, this is by no means the norm. Nor, does that mean that those campaigns can not be improved. In this article we’re going to cover the Facebook learning phase and the learning phase in general for new campaigns. In launching a new campaign, the age old adage holds more than normal, “patience is a virtue.”

Where Does The Learning Phase Start?

Facebook and Google are much smarter than us. Or maybe it’s better to say they are more knowledgeable whereas the savvy marketer is more wise. That knowledge comes from aggregating massive amounts of personal and behavioral user data, amassed from multiple sources. With this data, social and search platforms will take into account your inputted variables, desired outcome, and build momentum towards achieving it. Otherwise known as an algorithm. Any time you launch a new campaign or strive towards a desired outcome on one of these platforms, you are in fact leveraging the algorithm. Whether you are doing it according to best practices might be another story…so keep reading!

What Is The Learning Phase And Why Is It So Important?

So why should we care about algorithmic learning and what does it have to do with our campaigns? Well, because Facebook, Google, Microsoft, etc. are combining their data with your inputs and desired outcome, the trajectory and early performance of your paid campaigns rely heavily on trusting this algorithm to work towards your benefit. In simpler terms, it’s better for you to work with the platform rather than against it. Or, suffice it to say that you don’t have to shoulder the burden of all the optimizations that a marketer might have used to do in the past.

This means, you need to be cognisant of the learning phase. The learning phase is the time period it takes each platform to achieve a base level of momentum towards successfully achieving your campaign’s desired outcome. Disrupting the learning phase can mean you’re negatively impacting the platform’s ability to assess the marketscape and optimize for better performance.

How Long Is The Learning Phase?

In general, for most platforms, the learning phase comes in around 2 weeks. During this time period it is crucial that you don’t make any major changes to your campaigns.

How Do I Set Myself Up For Success?

When you start a new marketing campaign, it’s all too easy to look at the performance hourly or daily. But this isn’t constructive. It can actually drive you a little nuts. So, what can you do? You can rely on the campaign structure you initially built. So building a structurally sound campaign to begin with is square one. Here are some factors in doing so:

Always be testing!

Make sure you have at least 2 ads per ad set/group. This ensures that the platform can drive performance to the top performer and not sacrifice budget on low-performing ads. At a maximum, I wouldn’t recommend more than 5-6 ads per ad set/group. It’s also a good idea to vary the format of the ads where possible (i.e. video, responsive, etc.).

Pick your campaign objective wisely

If you pick traffic as your objective rather than conversions, you better bet that Facebook is optimizing to get you more traffic, not to get you more sales/leads.

Set A Reasonable Budget

It can change and isn’t always explicit, but as of now Facebook’s learning phase wants you to get about 50 results per week towards your desired campaign objective. Falling short of this number means that you aren’t providing Facebook with enough data to properly “exit” the learning phase. One of the biggest factors limiting potential results is of course, budget! If your budget is too low, you may not be getting in front of enough people to get the desired number of results during the learning period.

In practice, this is actually much easier said than done. Not all campaigns are going to have a big enough budget. Some companies because of their profit margins, might also be satisfied with a lower number of conversions within that 1-2 week time period anyway. So, this is where a marketer’s wisdom needs to come into play. For that, we’ll keep a few of our secrets to ourselves.

The other important element of budget other than simply being big enough to get results, is allocating enough of it to gather data. If you are running lots of ads per ad set/group and many ad sets/groups per campaign and all of these are splitting the same budget, you’re going to end up with incomplete data and will effectively be throttling the algorithm from finding the right people. So, make sure you aren’t stretching your budget too thin by trying to test too many things at once.

Now, some people might ask how Campaign Budget Optimization (CBO on Facebook) comes into play here. Well, fortunately, it is still not required as of writing this article. The premise of CBO is that you might not be stretching yourself too thin by trying to test multiple ad sets because Facebook will inevitably find the top performers and drive the majority (if not all) of the budget towards those ad sets. I’ve seen this work wonders. I’ve also seen this be a major blunder. We’ll cover this more in-depth another time.

What Changes Should I Avoid During The Learning Phase?

  1. Do not make changes to the budget unless you absolutely have to.
  2. Do not make changes to your targeting (audiences, interests, demographics, etc.)
  3. Do not make major changes within your ads
  4. Do not call it quits too early on a campaign that you set up well!

What Changes Are Ok During The Learning Phase?

  1. If you need to remove something because it was off brand or had a typo, do it!
  2. For Google and Microsoft Ads, adding things like negative keywords can actually benefit the algorithm’s learning if it was way off base.

What Happens After The Learning Phase

After the learning phase you’re free to make changes. But, we do advise caution. Drastic budget changes can easily reset learning, so you should scale incrementally if possible. If a campaign is performing well for a specific audience and you want to turn that off and test a new one, it’s a better idea to try it out in a new campaign. Likewise, if you want to try a different optimization event, don’t adjust the original campaign as this will drastically skew the campaign performance data and might not reset the learning phase properly. Because the learning phase is so important, it means you should NEVER delete a campaign that historically had good results.

How To Handle The Learning Phase For Campaigns Shorter Than Two Weeks

So, a holiday is coming up and you want to go ham on it with budget. We’ve tested numerous different ways of building campaigns for increased holiday expenditure and have consistently found that building into previously existing campaigns will get you the best performance. This does contradict some of the advice mentioned previously, but c’est la vie.

The way we see it, you have three main options:

  1. Build into current campaigns and increase budget.
  2. Build new scheduled campaigns with a large budget and turn off or decrease budget of old campaigns.
  3. Focus on scaling the retargeting budget only and leave current campaigns as are.

#1 is the winner. However, if a new segment isn’t within the audiences you already have built into your previously existing campaigns, then it’s perfectly reasonable to build a new one. If you can start this in advance of the holiday to get some data, great! If not, make sure you’re leveraging top performing ad copy to give you a better chance at success. Because you’re making such a huge budget change during such a small window, the “learning” is somewhat out the window, so in this instance it’s ok to make certain game-time decisions like changing out ads. The last word of caution on this note, is being aware of the conversion window and data lag. Performance data does not come in real time.


Once you’ve set up a campaign well and dedicated a budget, don’t make major changes to it during the first two weeks. Letting the algorithm do it’s thing will help you find out whether a campaign is a winner or not. If it is one, you’ll have a good foundation of momentum that you can then optimize off of with new ad copy, incrementally increasing the budget, and making slight adjustments to the targeting.


Useful, relevant content that you won’t want to archive.