Our Insights

LM Lab: Walking Through Facebook Ads Split-Testing

November 7, 2018

Not sure if the image of your logo or that cute kitten will drive more engagement for your ad? Going back and forth about whether to send folks to a product page or Contact Us? Stop the guessing with Facebook ads — and employ a quick and easy creative test.

When we think about creative testing, it’s important not to test just to test, but to test when there is a specific question that needs to be answered and a clear strategy driving the implementation of the finding. Facebook provides a great tool for split testing, also known as A/B testing, which can help determine the answers to the following questions:

  • Which call to action should I have present to my audience for the best results?
  • What image will drive the best engagement?
  • What landing page should I be sending my audience to in order to drive the desired action?

Facebook’s split testing capability divides your audience into random, non-overlapping groups to ensure the test is conducted fairly, without skew from other factors outside of the one testing variable. In other words, it’s pretty darn smart.

Interested in running a creative split test on Facebook? Here are some next steps to follow:

  1. Decide what will be tested — and what questions you want to answer with this test.
  2. Define a measurable hypothesis.
  3. Define an ideal audience for the test that’s large enough to support a test and not being used for any other FB campaigns. (i.e., retargeting audience)
  4. Pick a timeframe. Facebook recommends a 4-14 day depending on your objective.
  5. Determine the budget to support the test.
  6. Put together creative with only one variable: the one your testing. (You can conduct other tests later.)

For split testing you’ll need to keep a few things in mind:

  • The test may impact overall media campaign performance during the testing timeframe since budget needs to be allocated for a separate testing campaign that may not yield as high of performance. The testing audience must also be separated from the general campaign so there is no audience overlap to skew test results.
  • Only test one variable at a time — a single CTA, headline, visual element, etc.
  • Testing needs to be done during non-peak seasonal timeframes.
  • Since separate media budgets are needed to support these tests, it’s important to focus on key variables that there are questions around that can be scaled out to the full campaign.

That’s it! Now it’s time to test your hunch and optimize your ads.

Suzanne Shaw