Facebook Rolls Out A/B Split Testing for Content

Facebook Rolls Out A/B Split Testing for Content
  • A/B split tests for content are now available for Facebook, a step up after A/B split testing for Facebook ads.
  • Testing shapes your editorial calendar and content to make better posts that converts.
  • You can stop arguing about what content to create, let the audience tells you what they like.

Creating social media content is a lot of work. You don’t know how your audience is reacting to content until you have published it. Which picture or call-to-action is performing can only be tested when you post similar content twice and then compare results. Sounds inefficient? It is! That’s why Facebook is rolling out A/B Split testing for Facebook posts.

Why A/B Split Testing for content is amazing

A/B Split testing is used everywhere in digital marketing, from landing page performance to Facebook creatives. These days a marketer cannot rely on aesthetics or best practice when it comes to planning a digital campaign. A good marketer proofs his content, subject lines and call-to-action buttons by running two campaigns alongside to each other and then switches the one with the poorer performance off. That way, only the best asset will be carried forward. Up until now, this was not possible for Facebook content but things are going to change. With the rollout of A/B content split testing Facebook enables marketers to test best performing pictures as well as copy with a small audience before pushing it out to a wider audience.

A/B Split testing on Facebook

What Does that Mean for Your Content Strategy?

It means that nothing is set in concrete anymore. Your editorial calendar will be made of different components rather than static posts. You might start structuring your post similar to a blog post with the headline to suck the audience in, body text with the message you want to get across and a call-to-action to drive engagement. Each component can then be split tested with the new Facebook tool before you put all your money on the best performing post.

Same goes for the imagery used in your Facebook content strategy. You will be able to test the imagery for specific posts and potentially also the amount of images used in your post. This way you can build up an image library with the best performing posts which you then can utilise for future promotions to the same audience, knowing that this image attracts or converts.

With A/B split testing you can then test different versions of your content and your ads to see what works best and improve for future campaigns.

Sounds complicated? We predict that it’s only a question of time till marketers have trained the Facebook AI to do exactly the same without our input. By feeding the Facebook catalogue with imagery and posts, we actually train the Facebook machine in language and visual psychology. As this is done on a global scale, the machine can then learn which text or imagery is used to achieve a certain outcome.

Of course, this is only speculation as the tool is not actually launched yet. But considering that the exact same principle is used for training the Facebook advertising machine, we can only assume that in a couple of years from now, the social media platform will be able to suggest the best performing content for the best possible outcome.

A/B Split Testing in Facebook Ads

Split testing is already available in Facebook ads. For example, marketers can split test the same ad for two different audiences and see which one would be more receptive to the ad. Different ad delivery types or ad placements can be compared to each other, too. Other variables that can be tested are product sets and creatives. Audience split testing is probably the most common split test option when creating ads. It allows a marketer to divide the audience into two or more random and not overlapping groups, e.g. married couples vs. singles. Each ad set will be given an equal chance in the auction which enables the marketer to audience produces the best results.

Facebook ads split testing

To get the best results, it is recommended to only split test one variable at a time, e.g. either text or call-to-action. This means if you are testing two different audiences then you should just the same text and imagery so you can distinguish exactly what works and what not. The winning ad is chosen by comparing the cost per result of each ad set, e.g. cost-per-click or cost-per-view. The ad set with the lowest cost per result wins.  When the split test has finished, the marketer will receive a notification with the of the test. This gives the marketer the chance to understand market sentiments and to optimise the ads for future campaigns.

Applying the same logic and process to content seems to be an effective way to take the guess work out of the content production process. It gives way to confirming content directly with an audience and adjusting copy, CTA or imagery before any further efforts are potentially wasted. The longer the split test runs the more data can be gathered. The more precise the audience and or components for the test are narrowed down the more accurate the test.     

Will this new Facebook feature change the face of content marketing?

According to a recent eConsultancy roundtable discussion, marketers struggle to keep up with the high demand of content necessary to run campaigns. Marketing professionals find it difficult to keep producing high-quality, relevant content on a consistent basis. From concept creation to design to copywriting, there is a lot going on behind the scenes before a single post is actually created. The new A/B split testing feature will not take the work away but can streamline concept idea creation. Instead of routing multiple ideas internally to then come up with what marketers think would be the best to roll-out a campaign, mini split-tests can be conducted. Based on consumer insights a campaign can be developed which basically puts content marketing on its head.

Normally, a content marketer would be creating content for each persona and then tailor the content to wherever the persona is in the buying cycle. But as personas are semi-fictional customers it is hard to tell what really resonates with them. Therefore whole marketing campaigns are built on assumptions instead of real-world results. This is where the new Facebook feature will bring real value to the table.


Content production is the most time and cost intensive activity for a business. Potentially automating the content creation process and optimising content for business objectives is therefore every marketers dream. The roll out of this new Facebook tool is one step into this direction.

Sonja Ceri

CMO at

Australian based CMO Sonja Ceri is a regular commentator on the fast-paced media landscape. With a Masters in economics, 10 years of journalism experience and 7 years of agency Directorship, Sonja's interests go beyond text book marketing theory. Mum of one little girl. Agency Director at www.4dp.com.au