In the interactive ads that we did earlier, we gave Facebook some ad variables. And we left the responsibility on him to choose the appropriate text, image and title for the audience and display them based on the data he has. But the improvement that Facebook will make in this case will be limited only to the images and texts of the advertisement. But in the event that you want to experience a different audience, you want to know which image is better, which text is better among a wide range of texts. In this case, you can ask Facebook to conduct ad experiments, meaning that it does a test to see which image is better and which text is better to use in the ad.
How to conduct ad experiments
From the “Campaigns” page, from the left side of the “All Tools” page, we click on it.
Then we choose to choose the “Experiments” or experiences, as we see:
By clicking on this option, a screen appears stating that we have four tests as we can see:
We see here that the best of them is the “A/B test”, which is the candidate from Facebook, because by way we can test more than one copy of the advertisement, and determine which version gives better results than others.
The second option is the Campaign Budget Optimization Test. This is a test whose job it is to determine which is best to run the “CPO” nor does it mean improvement at the level of the advertising campaign at the time of creating the campaign itself. In the case of choosing this test, it will make it once activated and once inactive, and it will test which one is better.
The third test is the “Channel Conversion Optimization Test”. His job is to test the optimization campaign for both website and app conversions as opposed to testing the conversion optimization campaign only, to gauge the difference between them.
The fourth test is the “Brand Survey”. His job is to make surveys on my page by asking many questions on the page, and this is the easiest of them. We will learn how it works.
We click on “Get Started” as we see:
A page titled “Test Details” will appear. At the top of the page we find a message stating that we cannot do a “Brand Survey”. For this advertising account, because an amount of 32,000 pounds must have been spent during the past 90 days, as we see:
But if you use an ad account that has been spent, you will not see this message.
You will be asked if you want to do this test or survey on the whole count level, or on a specific campaign.
After that, it asks to select the page on which you want to work on the survey.
It then asks you to specify where this questionnaire will work. But you cannot specify a specific country, because Facebook here has specific options to choose from, as we can see:
Then he asks you about my field of work or the Vertical, choose your field of work as we see:
Then it asks you to set a timetable for doing this survey, as we see:
After that it asks you to specify the name of the test, you can write any name you want as we see:
The next step asks you to select the language of the questionnaire as we can see:
After that, it gives you a model of the questions that it will ask, which are questions that we cannot change, as we can see:
Ad Experiments
Finally, Facebook conducts ad experiments and publishes the survey and gives you its results.
We will return again to the Tests page. We choose the important test for me, which is the A/B Test.
We select the account I am working on from the top right of the page.
Then we click on “Get Started”, a page will appear entitled “A/B Test” as we see:
The first choice asks you to specify whether you want this test to be done on groups of campaigns, on specific campaigns, or on “Ad Sets”.
If you choose the “Campaigns” and then specify the names of the campaigns you want to compare, you can compare up to 5 campaigns.
Then you will be asked to set a start and end date for the experiment. But note that it is not less than a day and not more than a month, while the ideal period recommended by Facebook is 5 days for the experiment.
Then specify the name of the test so that you can know its results later.
After that, you are asked to choose a method by which you distinguish the winning test or the one that gives better results.
Then he needs to specify the basic measure of the success of the experiment, or “Key Metric” which is “Cost per result” as a default choice chosen by Facebook, meaning the cost of one result from the campaign, of course you can change it to any other choice but, but it is better to leave this default choice as it is because It is better to get the best results at the lowest possible cost.
Next we come to an item called “Estimated Test Power”. This item gives you the power of advertising and the expected results from it, and here Facebook determines it, based on the budget that you set for the campaigns, the duration of the test and the experiment itself. The higher the budget and the trial period, the higher the “Test Power”. Note that it is better that the “Test Power” should not be less than 80%, because less than 80% the results will not be correct.
If you get “Test Power” less than 80%. There are two actions you can take Ahmed from either to increase the duration of the test. Either you go back to the campaigns and increase the budget for the campaigns you have chosen.
It is very important to remember that for the success of conducting the ad experiments for the “A/B Test” test, the ads must be similar, there is only one variable inside them, which is either:
1- Location.
2- the target audience.
3- Ad content.
3- Pictures used.
Now you are ready for me to see the results of ad experiments, after the specified period of testing has passed, the results appear in the “Results” from here:
After conducting ad experiments, and the appearance of the test results and knowing which ad is better than the other, you can now spend the entire budget on the winning ad. This is one of the main advantages of conducting ad experiments.
