How to structure the testing phase in Facebook Ads



23 February, 2021

Why is the Testing phase important in Facebook Ads?

The tests are a tool that allows us to discover what works best in our ad strategy, so that we can repeat it on future occasions to generate more income, increase the number of potential customers and obtain better results at a lower cost.

To give an example of the effectiveness of testing, various Rocket Digital customer accounts saw their CPA (cost per acquisition) reduced by up to 75%; all thanks to the previous implementation of the testing.

Often when we think that Facebook ads are not working, the problem is simply not having tested enough variables (or at least not the correct variables) to determine precisely what is working correctly and what, on the other hand, is there. to improve.

A common mistake is changing too many things at once, or losing sight of the specific setting that could make the biggest difference to your results.

In any case, it is a fact that the testing phase helps us to be strategic and systematic when testing our strategy on Facebook Ads.

Creation of a Testing Framework

The objective we pursue when implementing this technique is to create a testing and analysis framework that is consistent with our objectives.


To create a Testing system, we have to think about testing:

  • Target, making assumptions and then validating them.
  • Creativity, through an ex-ante analysis, where we do an in-depth analysis of the buyer persona, and an ex-post, reasoning in a data-driven way.

Before starting to create a testing system, the guidelines are:

  • Prepare many creativities
  • Create and study multiple audiences
  • Include Testing campaigns in the Media Plan
  • You use automatic rules in Testing campaigns,
  • Choose the winning Assets and include them in Scaling campaigns
  • Doing an A / B Test means creating different alternatives at the campaign, adset or ad level, putting them in opposition to each other and finally evaluating which option has obtained the best results.


We use the term A / B because the choice is usually made between a stable base and an alternative, although Facebook allows us to easily test more variables.

We can do Facebook Ads Testing in two ways:

  • Technically, taking advantage of the platform in a statistically perfect way
  • Manually, faster, less accurate, but just as useful
  1. How to implement a Technical A / B Test

There are several ways to create A / B tests on Facebook, depending on the variable you want to test and from where you start creating the A / B Test.

To create an A / B Test we have to go to the Ad Manager toolbar, which uses an existing ad campaign or an existing ad group as a template to do the test.

Next, we will explain how to use this toolbar:

  • Go to Ads Manager.
  • Check the box to the left of the campaign (s) or set of ads that you want to use for the A / B Test. In the toolbar at the top, click A / B Test.

After logging in, a window will appear in which you will have to choose whether you want to create a test on existing campaigns, or create variables for a single campaign.

The variables in this case can be:

  • Image: Try different images in creativity, as well as different focus angles.
  • Video: Try different videos in creativity.
  • Ad text: Change the ad text in an ad campaign, but keeping the image, audience, and ad placements.
  • Age and Gender: Set up audiences with similar interests, but different age groups or gender to compare performance by demographic group.
  • Saved Audience: Compare a saved custom audience to see how effective your ads are in reaching various demographics.

There are other methods for creating A / B tests in Business Manager, which can be found in the official Facebook guide.

All A / B tests are based on the same technology to compare different creatives, different audiences or placements and determine the winners. At the end of the test, you will find the results in Experiments, in the right column of the Business Manager


Pros and cons of the Technical A / B Test

Pros: It gives us statistically significant results because we will limit Overlapping between audiences

Cons: It takes a lot of budget for it to be of real value and most of all, it takes a long time.

  1. How to create a Manual A / B Test

In addition to using the official Facebook systems, there is another method to test creativity and the public, less precise but no less effective. It is an empirical method, so experience is needed to apply it, but it is a faster alternative to the previously explained technical test.

The method consists of 3 levels:

  • Creativity
  • Audience
  • Aim


Creativity Testing

The first test to be carried out is related to creativity. To do this, the following guidelines will be followed:

  1. Load very different creativities within the same adset (3-5, different formats, optimized for each placement)
  2. Expect 5-8000 impressions
  3. Evaluate which creativity will be most driven by Facebook, and evaluate the main KPIs
  4. Turn off all other non-winning creatives
  5. Duplicate the creative pool (excluding the winner) to evaluate other potential winners
  6. Wait another 5-8000 impressions
  7. Evaluate which of the creatives is performing better at the moment and keep it. If the results are not clear, prolong the test.


Audience Testing

Once a number of creatives with valid KPIs have been identified (for example, ctr or CPA), different audiences can be tested while continuing to test the creatives in other campaigns.

In this case, the guidelines to follow are the following:

  1. Select different target audiences.
  2. Insert the best performing creativities into all of them
  3. Expect at least 5-8,000 impressions for each creative
  4. Pick the best audiences and turn off the others.
  5. When creativities stop working, add more.
  6. Repeat the same method constantly to avoid Ad Blindness.


Objectives Testing

Once your creativity and audience have been tested, you should also consider testing different goals or optimizations:

  1. Use a campaign that is performing well.
  2. Duplicate the campaign by changing the objective or optimization (for example, go from Conversion Add to Cart to Purchase)
  3. Wait until it reaches a sufficient number of impressions (depends on the business, but never less than 8000 impressions)
  4. Disable lower performing campaigns at aggregate level


Pros and cons of the Manual A / B Test

Pros: Provides results in a short time and is suitable for businesses on a budget.

Cons: It is important to correctly configure the automatic rules (explained below) and requires experience.

How to optimize the manual testing phase

While at the AdSet level you can set a budget for each experiment, to ensure that each experiment receives a fairly uniform number of impressions, this is not possible at the ad level.

This often results in an uneven budget allocation with some experiments receiving many impressions and consuming most of the budget, leaving others with fewer impressions.

To make the manual testing phase more accurate, without relying solely on Facebook’s machine learning, Facebook’s automated rules can be used.

Let’s take the example of wanting to test creativities on a wide audience to give Facebook’s algorithm the freedom to show creativity to the right target, knowing that this will drive one creativity over another.

In this case, you will not have accurate and balanced data, because the results will be influenced by the number of impressions of each creativity. Therefore, Facebook’s automatic rules will help in this case. One suggestion would be to try inserting a rule that deactivates creatives once they reach 5/8000 impressions, and then evaluate the metrics in a balanced way.

To create this rule, you must go to the toolbar at the top, click on “Rules” and then “Create a new rule”. The following window will appear:

After naming the rule, you must choose which creativities the rule will apply to. To do so, select in “Action” the option “Turn Off Ads”.

The next step is to configure the condition, that is: the creatives will be turned off when that condition is met.

Therefore, you have to select “Impressions” “is greater than”, and in the third box enter the number of impressions that, once reached, deactivate the ads.

The last thing will be to choose the reference time range on which we will base ourselves to check the number of impressions.

In this case, if a campaign has just been created, the easiest step will be to choose “Lifetime”.

If your campaign and creatives already have previous data, you should try to choose a time frame that allows you to quickly reach 8,000 impressions, while taking into account the daily budget.

In conclusion, the model explained in this article is totally flexible, and must be adapted to the reality, strategy and capabilities of each business.

The main thing is to have a clear and structured reference model, which can be followed to test the different factors that influence the performance of Facebook Ads campaigns.

If you believe our team of specialists can help you consolidate or improve the digital strategy of your business, do not hesitate to contact us!


Keepup to datewith our news.

Subscribe to our newsletter and get out latest Rocket news, events, insights & open positions straight to your inbox.