A vendor is planning to announce a crowdsource platform specifically for Facebook ads at the upcoming ad:tech conference. The announcement, while not a surprise, given Facebook’s ubiquity, does raise an interesting point: Facebook ad formats are still so new that not much industry knowledge has accumulated as to what works and what doesn't. Crowdsourcing a campaign could address that.

So could multivariate testing, writes Melody Park at Marin Software. 

"As the Facebook Ads platform is still relatively new, a standard method of performing Multivariate Testing is virtually non-existent. This in turn, can make it a daunting task for the advertiser to develop and theorize the best way to conduct a fair test between different ad variations."

Facebook’s Frequency Capping

The biggest pain point for marketers that want to run such a test is Facebook’s Frequency Capping restrictions, she said. Because Facebook stops showing an ad after it means a certain threshold, the advertiser loses the crucial volume needed for the test.

"The best way to work around Facebook’s Frequency Capping is to properly set up the testing," Park writes. To test creatives A, B, and C with a standard set of demographics, create three separate campaigns for creatives A, B, and C. All three campaigns should have the same demographic targeting to ensure a fair test, she says.

"Building your testing this way will prevent any segment overlap between the different ad variations. Facebook will always deliver the intended volume as long as there are no overlapping segments."