Facebook Ads Domination Course

Best Practices in Split Testing

NB: There is no video for this segment.  Please read the information included below or you can view it within Guide 7: Split Testing Your Ads For Optimum Performance.  You can also refer back to the video in segment 1 of this module where this topic is discussed.


There is an infinite number of possibilities with which you can carry out your split testing.  However, for all split tests, there are some things in common which usually lead to successful tests.  Here are some best practices for split testing:

1.  Start testing vastly different elements at the beginning of your campaign.

At the start, you don’t want to test one element’s variation only.  What you want to do is to try and cover all possible bases, so you can get an overall sense of what’s working for your audience.  Start with broad variables first and for each successful split test where one advert clearly comes out on top, use that element to narrow down your ad sets and do further split testing.

For instance, if you’re looking for the optimum age group to target, you may want to test first an age group of 18-30-year-olds and then 30-50-year-olds.  Once you have the winner for this test, you can then refine your experiment to look for a narrower age range.  If, for example, the winner was the 18-30-year-old group, then in your next split test, you can target users from 18-22, 23-26, and 27-30. 

This method not only saves you money, but it also saves you time as testing multiple specific variations at once may take you anywhere from several days to a few weeks to get reliable data.   

2.  Change only one element and keep everything else the same.

This information is important because if you change many elements or variables in a split test, then you won’t know which variable caused your ad to fail or succeed.  The process of elimination is helpful as you go through each element and cross out those non-performing or non-converting variables.

As an example, you can run the same advert to 2 different audiences.  You’ll use the same image, headline, ad copy, and call to action, however, the first audience will be 18-25-year-old women in relationships while the second audience will be 18-25-year-old single women.  If one of the audiences come out on top, then you know which audience group to target in your next ad campaign. 

3.   Run your split test ads for at least 3-4 days.

Facebook recommends letting your split tests run between 3 and 14 days.  Running tests for only 1-2 days may not produce data sufficient enough to determine a clear winner, and a 2-week test is just far too long and inefficient for your budget.  The benefit of letting ads run for a few days is you can see which particular times of day your audience is engaging with your ad.  You can then keep this factor in mind when optimizing your ad. 

4.  Have a sufficient budget in mind.

Before you begin your tests, you should have a defined budget in mind otherwise you run the risk of losing a fortune on Facebook ads.  The bigger the audience size, the bigger the budget you need to allocate.  The important thing to note here is that your budget should allow Facebook to conduct the split test effectively and produce a convincing winner.

When setting up your ads, Facebook will give you a suggested budget, but oftentimes, this can be quite expensive.  Most Facebook ads experts say you can start with a £5/day budget for each ad set. 

However, if you want to see results fast enough, then you need to increase your daily budget.  Moreover, you should know that you can also choose to split the budget evenly or weight one more heavily than others.