How to Split Test Your Facebook Content

Posted on March 25, 2013


Split testing is everywhere in the digital world – you name it and it’s being split tested right now.  However, the concept of split testing Facebook posts doesn’t seem to have taken off, and we think it’s about time it did.facebook-pano_19027

Like with some of my previous articles, a lot of Community Managers out there are going to be grimacing as yes, this will mean multiple posts!

Why Split Test?

Split testing is an absolute fundamental in digital marketing.  The ability to test, learn and optimise to improve results in short cycles is one of its’ great advantages.  Split testing allows us to get better at just about anything, and needs to continue to be embraced.

How Is This Relevant For Facebook Content?

Facebook content is no different to any other medium in terms of the need to test.  If you’re launching a big campaign for instance, it’s important to know which part of the content is working for which segment of your audience.

Say you’re posting with product content in a focused campaign.  It’s absolutely crucial to know the following for this specific content campaign:

  • What is the best time to post?
  • How long should the posts be?
  • What call to action drives conversions?
  • Which images work best?
  • Which audience segment responds best overall or to certain posts?

Knowing the above will allow you to increase your conversions in this instance by optimising content length, image, time of posting, call to action message and targeting criteria.  The same applies for engagement focused and other campaign types in the exact same way.

The questions really should be, “How is this not relevant for Facebook content?”, as why would you not want to improve your conversions and engagement?

Sounds Great, How Do I Do It?

There are a few different options as to how to split test your content on Facebook, and it’ll largely depend on personal preference and who your audience is as to which will be best for you.  Here’s how we tend to work at Zazzle Media:

  1. The first step is to decide what you want to test and what you will measure as success for each variant.  Decide whether you want to test by age brackets, or by gender for instance (we’ll show you how in a sec!)
  2. Once this framework is in place, set yourself some sensible boundaries and parameters – for example with testing images in posts you only want to test a certain amount, not hundreds. It will depend upon your time capacity and deadlines as to how many variables you want to test at the start of a campaign, but we’d recommend at least three of each variable should be tested; an expected winner, a similar alternative, and something slightly more left-field.
  3. Get your variables designed, written and ready to go based on your testing criteria.  For instance you may pick three images, or you may write three different call to actions to test. Alternatively you might instead be focused on audience segmentation, in which case define the parts of your audience you want to test against (e.g. the conversion rate for 18-25 year old females vs 26-34 year old females).
  4. Post your content & measure the results
  5. Analyse the data and see what works and what doesn’t and push your learnings into the next phase of the cycle

Aside from the process, there’s the physical act of posting on to Facebook in this way.  There’s a couple of ways to do it, first up is by using Facebook directly and using the targeting features (normally available once you have 5,000 fans on the page):

Split Test Targeting

Split Test Targeting 2

Doing it this way will require multiple posts to be created aimed at different segments.  One way to create a fair test is to break the ages down into say 3 year age groups so you’d do posts for:

  • Females Aged 18-20
  • Females Aged 21-23
  • Females Aged 24-26
  • Females Aged 27-29
  • Males Aged 18-20
  • Males Aged 21-23
  • Males Ages 24-26
  • Males Aged 27-29

Whilst that means doing 8 posts potentially, this will allow you to test with a segmented audience.  This can be used to see how a post performs amongst certain demographics.

Alternatively, to test how different versions of a post perform fairly, the best way is to layer the ages as best you can, such as in the below example:

Age Layering

This would result in 10 posts, but would give as fair a result in this context as possible.  Alternatively, you can layer your years by two or three year gaps, however the wider it goes the more diluted the results will become.

Don’t forget that you can test on Facebook by locations, ages, genders, and of course by anything in your content (copy, call to action, link, image).

The other ways to physically do this include using any of the standard posting and scheduling tools (Sprout Social, Hootsuite, PostPlanner etc) and using the targeting functions built into those apps, which use the options on Facebook.NetLinkedSolutions Logo w website

Key Takeaways

  • Leave no stone unturned: The more you can test and the more data you collect then the more efficient you will be able to make your campaigns over the long-term.  It may seem a pain right now, and your Community Manager will hate you for asking, but the benefits will pay off
  • Don’t be afraid to experiment: As long as you are working within the parameters you set, you shouldn’t worry too much about ‘getting it wrong’.  Failing is the best way to learn that something doesn’t work, and optimising those variations out.
  • Do this on your next campaign: Now you know it can be done, there’s no reason to not try this in preparation for your next campaign.  Ultimately, why would you not invest a little time in improving your engagement and/or conversion rate from social?

Authored by Ben Harper: Ben is a former data analyst turned social media expert having honed his skills within both major corporations and cutting edge social start ups and it is that unique mix of skills that makes him a key member of the Zazzle Media team.