Skip to Content

How Facebook Ad Split Testing Works

In Facebook, you can now compare your ad set strategies with Facebook Ad Split Testing. You can now test 2-3 ad sets with different delivery strategies to see which gives you the best Lead generation results.

How to do a Facebook Ad Split Test

  1. create an ad
  2. Choose your Marketing Objective. You’ll be able to choose the split test after this screen. Split tests can be set up for the following marketing objections: Traffic, App Installs, Lead Generation (Lead Ads), and Conversions.

    Screen Shot 2017 04 24 at 7.08.47 AM
  3. Check the box to make this a split test AdScreen Shot 2017 04 24 at 7.11.24 AM
  4. Name your campaign and set a budget. You will see how it’s split on the following screen and will have choices
  5. Choose your VariableFacebook Ad Split Testing Variables with list of Variables
    These are: Delivery Optimization, Audience, and Placement (see more details on these below the steps)
  6. Set up how you would like your budget to be set. Tell Facebook the exact amounts you’d like to be allowed per variable and when you’d like your ads to show. Your spend and impressions will be divided between  2 ad sets. You’ll decide between an even split and a weighted split between your ad sets.
  7. Choose your variable, A. Delivery Optimization: Optimize for Link ClicksYou’ll be able to split test between link clicks, impressions, and daily unique reach. This has always been a challenge to decide between and now the split test will tell you exactly which one performs the best.
    So Ad Set 1 would be set to Link Clicks and then Ad Set 2 would be set to Impressions. If you have the ability to test 3 ads, you can add in Daily Unique Reach. If not, run the first A/B and test that winner against daily unique reach.Screen Shot 2017 04 24 at 8.47.14 AMB. AudienceScreen Shot 2017 04 24 at 8.54.04 AMYou can only choose from saved audiences to compare, yet you can set up a new saved audience from this dashboard. An example might be for a real estate agent testing ads to build a custom audience with the exact same settings except ad set 2 has a “most likely to move” inclusion. Or each audience is the same except ad set 2 has a income selection or income exclusion. Another example would be to have two saved audiences narrowed down by geography. So I’ll test how 89134 zip code compares to 89135 and so on. There are so many incredible options.

C. Placements

With split testing placements, we can compare ads based on where they appear. There are so many combinations from one selection ad set 1 vs. ad set 2 or multiple selections of each.

See also  Hootsuite vs. Hubspot

We can see which performs better with these examples

  • Facebook Feed vs. Facebook Right Column
  • Facebook Feed vs. Instant Articles
  • Facebook Feed vs. In Stream Videos
  • Facebook Feed vs. Instagram Ad
  • Facebook right column vs. Instagram Ad
  • Facebook Feed vs. Audience Network
  • Instagram vs. Audience Network
  • Instagram feed vs. Instagram stories
  • Android vs. IOS Mobile
  • Automatic Placement vs. Choices
  • Any combination of above.
  • You can test excluding categories such as
    • Debatable Social Issues
    • Gambling
    • Mature
    • Tragedy & Conflict
    • Dating

Rember, all other ad settings will apply to BOTH Ad set 1 and Ad set 2 as well as Ad set 3 if you have that option. This means, if you set each a geographic location, it will apply to both and THEN your variables will be added. So everything else in your ad sets will be identical.

How does Split Testing work?

First, the audience you select is divided into individual groups. This is a random division placing members of the group into groups that do not overlap. Each ad set has an identical ad. There is however, one difference. This is called a variable. The types of variables you can select are: Delivery, Audiences, and Placements.You can then divide your budget into even spend between your ad sets or you can more heavily weight your spend on one ad vs another. Let’s say you are a real estate agent. You are marketing a new listing and want to A/B test audiences. You will first ad in your usual criteria: interests, location, income and so forth. Now, you want to add in only people that are “most likely to move”.

See also  How To Change a Facebook Video Thumbnail

You’ve tested this before and know that this  audience has a higher spend generally as many people are bidding on that audience. So you give 60% of your budget to the “likely to move” and 40% to the ad set two not containing that inclusion and weight the ad spend this way.Once your split test has been succesfully completed, a report will be displayed comparing the ad sets. Of course, the best performing ad set will be declared the winner. You will receive an email with these results once complete. You can then use that winner and split test it against another ad set or scale it because you know it works.

Split Testing helps businesses that are running Facebook ads to comprehend how different variables of their ads affect campaign results. This kind of A/B split testing or A/B/C Split Testing allows you test various versions of your ads so you can see what is creating the best results so you can improve your campaigns.

Split Testing will allow an advertiser to simply test the same ad using 2 different audiences to measure which audience performed best. The advertiser might choose to A/B test two ways of delivering their ads,  such as conversions or link clicks. This can help to determine which generates the best results. An Advertiser would implement Split Testing to measure how changes in different variables such as these might change the success of the ad.

[tweetthis display_mode=”box” remove_twitter_handles=”true”]WOW! Check out this great #blog on #Facebook Ad Split Testing from Lori Ballen! @loriballen_[/tweetthis]

How Facebook Split Testing works

Facebook’s Split Testing feature permits advertisers to create more than one ad set and test them against each other to see what methods produce the best results. Here’s how it works:

  • Split Testing splits your audience into various, non-overlapping groups then  shown ad sets with identical settings
  • This random application ensures the test is implemented fairly because other factors won’t incorrectly change the results of the group measure. It also promises that  each ad set is given the exact same opportunity in the auction.
  • Each ad set that is tested has one major difference. It’s called a variable. These variables can be various saved audience types, placements such as the Facebook feed vs. Facebook right column, or delivery optimizations such as link clicks vs. impressions. Facebook will create a duplicated ad copy only changing that one selected variable.
  • In order to receive the best results from Split Testing, the advertiser only has the chance to test one variable at thta time. For example, when testing 2 saved audiences against each other, you would not also test two delivery optimizations at the same time because you wouldn’t know for certain which change affected the performance of the ad.
  • Split Testing is a strong way to measure ad performance for people. There are no cookies involved.
  • Data is compiled from various devices.
  • Once the ad sets are compared based on your split test goal, the results generate a report. The best performing ad is declared the winner.
  • You will receive an email notification once your split test is finished. You can then use this data to improve your Facebook advertising efforts. You can also then do another split test using comparing that winner against a new ad set with different variations.
See also  Avoiding Facebook Jail - How to Promote Organically Without Getting Into Trouble


You need a business manager account to test a split test inside your Ads Manager.Can I use power editor for my split test?

Both Ads Manager and Power Editor will work for split testing.  Not all objectives are available yet, although more are on the way.

Can anyone create a split test?

What objectives are OFFERED for split testing?

Traffic, App installs, Lead generation, Conversions

Can I test 3 strategies in one split test?

If you can see the option in your Ads Manager, you can test up to 3 strategies. A/B test or A/B/C Test.

Which variables can I test?

Target audience, Delivery optimization, Placements

How long should I run my Split Test?

According to Facebook, a 3 day minimum and 14-day maximum is ideal for a split test. If you run it shorter than 3 days, you’ll run the risk of not having enough data. If you run your split test longer, you may not use enough budget as it would be allocated over various weeks or months. 14 Days is ample time to get what you need so spend the budget and set it appropriately.

Can I choose different images?

Not yet. You can test delivery, audiences, and placements

Sharing is caring!

This site uses Akismet to reduce spam. Learn how your comment data is processed.

You cannot copy content of this page