Facebook Split Test Ads 101: How Split Testing Works

Split Testing is a crucial part of creating Facebook campaigns. The goal is to test ONE variable difference at a time in Facebook campaigns to identify the best combination of variables! This can mean testing a different image, ad copy, call to action button, audience/targeting, placement, etc. 

Just make sure to test only ONE variable at a time otherwise if you test multiple variations in a new campaign and that campaign performs better, you won’t be able to tell what variable was the cause! By testing one variation difference at a time, you can identify the cause of the success or failure from campaign to campaign. 

If you enjoy this Split Test tutorial, we suggest you join our Facebook Marketing Mastery course included in our modMBA program. Click the button below to learn more!

How Split Testing Works

Think back to your high school science classes and hypothesis testing. You would have your control group, then you would change ONE variable in the test group in order to document the differences/results!

EXAMPLE: A/B split testing allows advertisers to easily test the same ad on two different audiences, to see which audience the ad performs better with! Advertisers can use split testing to interpret how changes in different variables might impact the success of their ads!

To get the best results from your Facebook Ads, you have to test different versions of ad creatives (at the AD level) and targeting/audience (at the AD SET level) in our campaigns. To test different Ads and Ad Sets we can do “A/B Split Testing” in our campaigns. 

After doing A/B Split Testing, we can easily find which variables are best for our campaigns or which ones are not!

Now let’s first cover how we do manual split testing!

Facebook Split Test Ads + Split Testing 101: How to Create a Coursenvy Super Split Test Campaign

How Long Do I Split Test?

Optimization is your ongoing job as a Facebook ad manager. You need to be split testing campaigns, ONE variable at a time, at all three levels (objective, ad set, and ad) to optimize for your key metrics (which is typically seeking the lowest cost per conversion).

You should shut off non-performing ads (poor results) every 3-7 days or once your campaign reaches 1,000 impressions (whichever comes first). From my experience, 1,000 impressions is enough volume to know which ads are winners and which ads are losers. 

Combine The Winning Variables & Keep Testing

Once you start to identify your best audience and best ad creatives, combine them into new campaigns to confirm they indeed are the best variables. If not, keep split testing!

Once I find a few winning campaign combinations, I will slowly scale up their budget 20-30% every 2-3 days until I am at a daily budget I am comfortable spending. 

I am constantly testing ads to find more audiences and to obtain cheaper results. Once you get a few of these winning ads, they truly can become EVERGREEN or lifelong ads fo

Facebook Split Test Variables

The variables you test can be ANYTHING you want to test! If you aren’t certain who your target market is, split test to find out! Split test locations, age, gender, languages, interests, placements, connections, custom audiences… the list goes on and on! Just keep creating duplicate campaigns and edit each version ONE variable at a time!

You can split test as many variables as you want in order to identify which ones perform the best with your target audience. Such as which audience drives the lowest cost per click, which images have the highest relevance scores, or what variables increase conversions, while lowering your cost per acquisition. 

Facebook Dynamic Creative

Dynamic Creative takes multiple AD variables (such as images, videos, titles, descriptions, CTAs) and then creates creative variations for each person viewing your ad that you can later track in Ads Manager to see what performed the best (i.e. drove clicks, engagement, conversions!)

Dynamic Creative is a great optimization tool when you’re not sure which AD creative resonates the best with each audience you target.

With the Dynamic Creative option enabled, you can upload multiple creative assets such as images, videos, titles, descriptions, and CTAs (Call-To-Action buttons).

Dynamic Creative will then find optimized ad creative combinations by taking multiple ad components/assets you provided (i.e. images, headlines, etc.) and automatically generating combinations of them.

Dynamic Creative will then optimize the creative components that deliver the best results for each impression served. Then your Dynamic Creative ads will show your targeted audience the highest performing creative combinations

Who is Dynamic Creative for?

The Dynamic Creative option is a great tool for Facebook advertisers with MANY assets who want to learn what creatively resonates with their target audience. 

I like to compare Dynamic Creative to “Machine Learning”… it is a Facebook tool that helps me identify my best AD level creative assets, then I use these winning creatives in my new AD SET level split test campaigns. Once I have my Ad Sets optimized, then I scale up my budget on the winners and shut off the losers!

PRO TIP:  Dynamic Creative is NOT a substitute for split testing. I will only use Dynamic Creative to find my top performing CREATIVES from hundreds of options (i.e. titles, images, etc.). Then I will take these winning ad creative assets (derived from the Dynamic Creative ad) into my new split test campaigns where I am testing targeting at the AD SET level (again, ONE variable at a time to see what audience drives the best results).

BEST PRACTICES: Split Testing

Best Practices for Split Testing: https://www.facebook.com/business/help/290009911394576 

  • Test only ONE variable at a time for more conclusive results.
  • Focus on a measurable hypothesis (i.e. optimizing for cost per action).
  • Use a big enough audience for the test.
  • Use an ideal time frame (4 days average).
  • Set an ideal budget for your test (80%+ estimated Test Power).

    Leave a Reply

    Your email address will not be published. Required fields are marked *