How to Split Test Facebook Posts and Facebook Ads

social media how toDo you use split testing?

Did you know you can split test both your Facebook Page posts and your Facebook ads?

Split testing (also known as A/B testing) is the best way to make sure every status update gets maximum engagement and every ad is optimized for conversions.

In this article I’ll share how to split test Facebook posts and ads.

Why Split Test Facebook Posts and Ads?

Split testing compares two similar but different versions of something (e.g., a Facebook post or ad, landing page design, etc.). When each version is published, you track its performance and determine which one provides the best results related to your goals. You keep the one that’s working.

ab testing facebook

Find out what you need to know about AB Testing on Facebook.

Split testing is simple enough, but it’s an important component of shaping your marketing tactics. You can’t overlook it. It’s a continuous process—just because one version works best now doesn’t mean it will work best next month or even next week.

Split Test Facebook Page Posts

Here are the steps you need to follow to split test your Facebook Page posts.

1. Post With Purpose

What do you want your post to do? What results are you looking for? The answers to those questions determine what and how you share content with your current and potential Facebook audiences.

Most, if not all, of your goals are going to rely on getting people to click your links and share your posts. Audiences respond best when you inspire, unite, amaze, advise or give (e.g., discounts or coupons). Use those tactics to get more clicks and shares.

washoe casa facebook post

Washoe Casa educates and inspires with a post about their foundation.

A word of advice: Any status update that links back to your website or blog must reinforce the promise you made on Facebook.

While Upworthy-style headlines and status updates are all the rage these days, one of the criticisms of that style is that the content linked to the eye-catching and click-worthy headlines doesn’t always deliver on the promise.

2. Create Two Versions of Posts

To split test your Facebook posts, write a pair of updates you can test against one another. The key is to change only one or two elements so you have a good idea of what’s motivating any increases in engagement.

In the example below, Post Planner used the same article link but changed the post comment. The first version included a question.

postplanner test a update

The A test as a question.

The second version used a statement with a strong word (beware) to evoke emotion and curiosity. A small change in your text may make a big difference in engagement.

postplanner test b update

A small change to your posts helps you see what your audience likes.

A few ideas for your split test are posting an update with a photo vs. text only; picture A vs. picture B, both with the same text; a picture vs. video post; call to action A vs. call to action B, both with the same photo.

3. Use Trackable Links

In order to track your two posts you’ll have to add something to the end of each URL so you can tell them apart in your report. If you’re fairly savvy with Google Analytics, you can use UTM tracking parameters to track engagement differences within your Analytics dashboard.

An easier way is to create a customized Bit.ly URL for each post you want to compare. For example, if the URL Bit.ly creates for you is bit.ly/123456, you’d add “ab1″ to the end so it becomes bit.ly/123456/ab1. Then yound so “ab2″ to another version so it becomes bit.ly/123456/ab2.

Below, ShortStack tested whether long or short posts resulted in more interaction with our audience. We tracked each update with a customized URL.

shortstack posting test a with custom url

ShortStack shared a customized Bit.ly URL for easy tracking.

In the second version of the update, we used the same image, but posed a short question and used a different custom link.

shortstack posting test b with a change to custom url

For this post we added SociallyStacked_2.

With your trackable links in hand, decide which post you think will do well and which one won’t. Post the weakest one first, then hide it and post the other version (the one you think will get more clicks). Make sure each post is up for the same amount of time.

Just a quick note: Hiding a post is better than deleting it if you want to preserve the interactions (e.g., comments, likes and shares).

4. Compare Post Engagement Results

To find out how each post did, paste the corresponding Bit.ly URL in your browser’s address bar and add a + sign to it. This will tell you how many times the link was clicked on.

If the first post did better, make it visible again and hide the second one. If the second one did better, leave it up!

Our Bit.ly report showed that after 15 minutes, our posts (question vs. statement) performed similarly.

ctr results for custom link a and b

Track the clicks each link generates.

You don’t have to split test every status update obviously, but if you’re trying to accomplish specific goals via your Facebook posts—sales or otherwise—it’s a relatively easy way to tell which kinds of updates your fans respond to best.

Now that you have A/B post testing under your belt, let’s get started with split testing your Facebook ads.

Split Test Facebook Page Ads

Here are the steps you need to follow to split test your Facebook ads.

1. Create a New Facebook Ad

Before you start testing your Facebook ads, familiarize yourself with creating Facebook ads and using Facebook’s Ads Manager.

When you’re ready, go to your Ads Manager, click the green Create Ad button and choose one of the eight ad objectives.

facebook ad goal options

Determine what kind of results you want to get out of your ads.

You’ll see that Facebook gives you a few options for bidding: CPC (cost per click), CPM (cost per 1,000 impressions) or Optimized CPM.

If you’re just getting started with Facebook ads, choosing the best option can be tricky. I suggest running a few tests with each one.

facebook ad bid options

You can choose CPC, CPM or Optimized CPM as your bidding type.

What you’ll most likely find is that your ad budget and advertising goals determine whether CPC, CPM or Optimized CPM is your best option.

2. Test Different Copy and Creative

Create a pair of Facebook ads to test against one another. Just like your post tests, change at least one element of an ad so you have a good idea of which copy or creative is performing best. You can find and measure the results of your split testing in your Facebook Ads Manager.

Below, Baremetrics changed both the ad title and description, but used the same image in each ad.

baremetrics ad comparison

Use subtle changes in ad pairs to see which one does better.

A few ideas for your Facebook ad split testing are using an ad image vs. the same ad image with text overlay, ad image featuring a human vs. an image not featuring a human or an ad with a call to action button vs. without a call to action button. The options are endless. Base your testing choices on your goals and objectives.

Frequently testing new ad creative and copy is vital if you want your ads to sustain their performance. If you have the same ads running for too long (more than a few weeks), they become stale and less appealing to the users you’re trying to reach.

3. Experiment With Audience Targeting

In your Facebook Ads Manager, take a look at More Demographics and you’ll find advanced targeting options.

As you know, Facebook tracks just about every online move you make. If you’re a Facebook advertiser, that means you have access to hundreds of audience targeting options.

Targeting your Facebook ad audience can be complicated with so many options to mix and match, so it’s important to test several to see which audiences respond best to which ad.

choosing a target audience on facebook

Facebook’s Ads Manager and Power Editor ensure you reach the right audience.

For narrow testing, you could try location + languages + interests, purchase behavior + age + relationship status, household composition + “newlywed” + politics and so much more.

For a more precise audience, create a custom audience, lookalike audience or website custom audience.

A custom audience lets you target users based on an email list, phone number list, Facebook UID list and/or Facebook App UID list.

Lookalike audiences help you reach people who are similar to your current customers for fan acquisition, site registration, off-Facebook purchases, coupon claims and brand awareness.

website custom audience reaches people who have visited your website through retargeted ads on Facebook.

choosing a saved target group on facebook

Saving your target groups saves time when building future ads.

After you discover which type of targeting combination works best for you, create a Saved Target Group (also known as a Saved Audience). This ad feature saves you a ton of time.

Within Audiences in the Power Editor, click Create Audience and select Saved Target Group. When you get to the targeting step in a new ad, click the Use Existing Target Group button. Manually inputting your targeting options for each new ad is a thing of the past!

Conclusion

Facebook offers so many opportunities to fine-tune your social media marketing efforts. Split testing ensures you’re not wasting time with the wrong ones.

You have an amazing amount of data at your fingertips. Use it to maximize your overall Facebook engagement and target the most precise audience for your campaigns.

What do you think? Have you ever tried split testing? What have you tried and what have you learned? Please share your experience, questions and advice in the comments below.

Tags: , , , , , , , , , ,

About the Author, Jim Belosic

Jim Belosic is the CEO of ShortStack, a self-service software that allows businesses to create engaging campaigns for social, web and mobile. Other posts by »




More Info
  • http://www.eBizROI.com Rick Noel, eBiz ROI, Inc.

    Interesting article Jim. Some great tidbits to test.

    In your experience, does split testing posts run the risk of showing your most engaged users the same post twice? Is there a way to prevent that ? Also, testing posts during a period of similar audience engagement as per insights, should be factored in. Thanks for sharing.

  • Allison

    Can you split the audience on the post test as well or do you post them simultaneously?

  • Tania Yuki

    Quick question – when you split test page posts, are you effectively posting the same message to the same audience twice? Confused…

  • Jessica K

    This doesn’t really explain how you split the audience or create two similar audiences without overlap unless I am missing something.

  • Brittany

    For the post split testing if you are comparing variables such as post lengths, how do you know if the difference in performance is based on the length of the post or due to being posted a different time that made it perform better?

  • Alistair Beech

    I can see the point of this when creating ad’s, but not so much posts (unless you can explain how to Split audiences too?)

    This will sound incredibly dumb – But – surely it’s universally known that audiences like short-ish posts with a nice image over reams of text and nothing to look at?

  • Grant Perry

    I agree split testing has it’s place but the biggest issue is in regards to reduction of social proof. If you only promote one post you can build up valuable likes, shares and comments. That gets diluted once you split it into different tests. That can be a particular issue to audiences who don’t yet know your page or your brand.

  • Pingback: Marketing Day: Twitter App Install Ads, Google Cardboard & Viewable Video Ad Impression Metrics()

  • Tina VZ

    I’m also wondering how your audience wouldn’t see both posts at the same time. Hiding a post, as mentioned, only hides it from your page and it does not hide it from the news feed of your audience.

  • Blake Jamieson

    Good stuff Jim! Curious about one thing …in the example with Stripe analytics, it seems like the test would be more conclusive if they isolated the test to one variable at a time (headline OR body text ..instead of both at once). Is there a reason behind testing both at the same time?

  • http://www.paulparsons.me/ Paul Parsons

    The A/B testing of a Facebook post doesn’t really work without posting the same content twice to your fans, which doesn’t make sense right? I think you should have made this article more about the landing pages from the posts and the adverts, but not posts – unless you have some new way of doing this without newsfeed spamming :-)

  • Archie

    correct

  • http://www.shortstack.com/ Jim Belosic

    Yep — that’s what we do. We’re running with the assumption that not all of our fans see every post. In the example above we put up a post, left it for 15 minutes, hid it while we put up the second post and after 15 minutes checked Insights. We then unhid the first post since it had done better, and hid the second one. Wow. it makes my head spin just to write that!

  • http://www.shortstack.com/ Jim Belosic

    Good question Brittany. Obviously you could go very deep with this kind of testing but we’re looking at the verbiage that seems to grab our fans’ attention.

  • http://www.shortstack.com/ Jim Belosic

    Good point, Grant. Thanks for reading and commenting!

  • http://www.shortstack.com/ Jim Belosic

    Hi Rick, yes that’s certainly a risk. And that’s why we don’t do this very often. We use it occasionally to gauge what are fans are responding to. Thanks for your comment.

  • http://www.shortstack.com/ Jim Belosic

    Hi Alistair, we tend to mix up our posts quite a bit: Sometimes we use images and sometimes we don’t. We use split testing on ads frequently but on posts less so, just to see what are fans are responding to. Curiously, our text-only posts often do very, very well.

  • http://www.shortstack.com/ Jim Belosic

    @blakejamieson:disqus There are lots of copy variables a person can test, especially with Facebook ads. But a/b testing copy can go beyond testing headline tweaks — you can test two opposite forms of messaging. With Stripe’s imagery the same, it seems Stripe choose to split test their copy to see which kind of messaging performs best for them: More advanced acronym heavy copy that emphasises their product’s features versus direct copy that emphasises their product’s benefits.

  • http://www.shortstack.com/ Jim Belosic

    @allison you can split test your posts by audiences by adding targeting to your status update before you post them. Also, it’s best to share your posts at different times. Either on the same day at another time or on the following week on the same day and time.

  • http://www.shortstack.com/ Jim Belosic

    @PaulParsons1981:disqus the majority of a page’s Facebook fans do not see all of a brand’s posts. By split testing posts with the same content, it gives more of brand’s fanbase an opportunity to see a message and/or important piece of content. Also, it helps a brand gauge which types of posts and/or posting times get the greatest amount of reach. For example, you could test posting a message at 2:00 pm, then a similar post at 2:00 am and see which post was interacted with the most and/or had the most organic reach. There are two different audiences on Facebook during these times so there’s a small chance the same fans will see the same message twice. And even if they do, there’s no harm done. The beauty brand Benefit Cosmetics is a great example of why posting the same/similar messages can actually be advantageous: http://maximizesocialbusiness.com/pending-title-3-valuable-lessons-you-can-learn-from-2-successful-brands-on-facebook-14356/

  • http://www.shortstack.com/ Jim Belosic

    @jessica this article is not so much about split testing audiences as much as it is about split testing your messaging to find out what your audience responds best to.

  • http://www.paulparsons.me/ Paul Parsons

    I get what you mean, but this would be one of the reasons I would unlike a page, why would I want to see the same content posted by a brand, A/B works on the web and through emails because you can segment the audience, via social media it doesn’t, unless Facebook and Twitter introduce a feature like that (which would be bloody awesome).

  • Sebastian

    Hi Jim, thanks for the post.

    On one of the answers below you mentioned you are running this strategy with the assumption that not all the fans see every post.

    My only concern with this strategy would be that, due how the news-feed algorithm works, the most engaged fans are the ones who are more likely to see most of our posts.This could mean that our best fans – the ones that are truly engaged with the page – are the ones who will be receiving/viewing most of the duplicated content. Therefore, the test would be running based on their reactions and not based on a representative sample.

    Other than that, it is a very valid and interesting approach. Thanks for sharing.

  • Pavel Mezihorák

    Regarding “split testing posts”. Isn’t it better to
    create dark/unpublished posts? I guess if you create it within one ad
    set, it will not be shown to the same audience. Also if it makes a
    little bit sense, you do not have to show it to your fans at all and
    try it on a similar audience instead.

  • K Chaitanya Vemula

    Nice concept,

  • Shanda Boyett

    Great post that got me thinking. Has anyone turned on “Post Targeting and Privacy” in Facebook settings so they can limit who sees a post by location, for example, as a way to conduct A/B testing of content that doesn’t require the same people to see similar posts?









Pinterest
Join our Social Media Marketing Networking Club
Check out the Social Media Marketing Podcast!