A/B testing as a marketing strategy goes way back to the days before the internet even existed, when direct mail marketers used it to conduct small tests on a tiny fraction of their contact lists before committing to the massive cost of printing and mailing a campaign.
It was a time-consuming process, since they had to start their tests many weeks—or even months—in advance of a campaign launch.
Fortunately for modern marketers, social media A/B testing is possible in real time, allowing you to refine your content marketing and advertising strategies on the fly, building a deep understanding of what works—and what doesn’t—that will help you make the most of your social media spend, and your content team’s time.
Bonus: Download a free guide that reveals how to increase social media engagement with better audience research, sharper customer targeting, and Hootsuite’s easy-to-use social media software.
What is A/B testing?
A/B testing is a form of real-world research that allows you to test small variations in marketing materials to determine which is most effective with your particular audience.
To perform A/B testing—also known as split testing—you separate your audience into two random groups and show a different variation to each one. You then compare the responses to each variation to determine which is most effective based on your chosen metrics.
It’s important to understand that classic split testing should be used only to test one variation at a time. So, if you want to test a couple of different headlines, a couple of different images, and maybe even a couple of different styles of article or ad format, you will need to conduct multiple tests.
After all, if you test two wildly different variations, you won’t know which elements impacted the results. And you may sell yourself short by making incorrect assumptions about how the different elements of your content or ad work together to affect response.
How can A/B testing help your social media strategy?
When you first started social media marketing, you had to base your strategy on some assumptions. For example, you might have read that tweets with photos get 35 percent more retweets. But does this hold true for your brand, and your audience?
A/B testing social media can help you find out, for example, how a tweet with a video performs compared to a tweet with a photo for your specific followers. Then, if the tweet with the video performs best, you can test variations in the video itself—short versus long, sound versus silent, animated versus live action, and so on.
This same logic applies across all elements of your social strategy. You can use A/B testing to test the most minute details of your social media marketing plan—in fact, this kind of testing is most effective when used to test the smallest of changes.
You can apply this strategy across all social networks, performing different tests on each individual platform so that you don’t mistakenly assume that what works best for your brand on Twitter will also work best on LinkedIn.
Over time, you will gain insights about what works best for you on each social network, helping to refine your social media strategy. But you should continue to test small variations, even when you think you have a winning formula. The more you test, the more granular your understanding will be, which will allow you to stay on top of changes in audience preferences over time.
A/B testing your organic content can also provide valuable insights about what performs best, so you can make the best decisions about what content is worth paying to promote and what elements to include in your social media ads.
What can you A/B test?
You can A/B test any component of your social media posts and ads, but let’s look at some of the most common elements to test.
The type and style of language you use in your social media post is worthy of extensive testing. For example, you could test:
- Post length (number of characters)
- Post style: a quote versus a key statistic, for example, or a question versus a statement
- Use of emoji
- Use of a digit for posts linking to a numbered list
- Use of punctuation
- Tone of voice: casual versus formal, passive versus active, and so on
Check out the slightly different tones (“a breeze” versus “convenient”) in these two tweets from Ikea.
— IKEA USA (@IKEAUSA) June 8, 2017
— IKEA USA (@IKEAUSA) May 24, 2017
Headline and description in link preview
The headline and description in a linked article preview are highly visible and important to test. You can apply all the same tests mentioned for post content to your article headline. Keep in mind that you can edit the headline in the link preview, so it does not have to be the same as the headline on your website.
The posts below link to the same content but use a slight variation in the link preview headline. One incorporates a digit, while the other does not.
The content headline on the website itself is also slightly different: Success in Six: 4 Best Practices for Building Impactful YouTube Bumper Ads.
Call to action
Your call to action (CTA) is another incredibly important part of your marketing—it’s where you ask readers to engage. Getting this right is critical, so be sure to use social media A/B testing to try various approaches through multiple tests that get into microscopic detail.
Note the two different CTAs—”Install now” and “Use app”—in these Facebook ads for World Surf League.
Use of image or video
As mentioned above, while the research suggests that posts with images and videos perform best overall, it’s important to test this theory with your audience, and on each social network. For example, you could test:
- Text only versus posts with an image or video
- Regular image versus animated GIF
- Photos of people or products versus graphs or infographics
- Length of video
Hootsuite testing has shown, for example, that our Twitter audience responds very well to GIFs—so, not surprisingly, we incorporate them often in our tweets. But since audience preferences can change over time, we continue to test GIFs versus other image types—as in the tweets below.
— Hootsuite (@hootsuite) February 8, 2016
— Hootsuite (@hootsuite) July 25, 2016
If you’re using social media ads, test different formats to see which are most effective for your different kinds of content. For example, in your Facebook advertising, maybe carousel ads work best for product announcements, but a local ad with a “Get Directions” button works best when you’re launching a new store.
A/B testing Facebook ads formats against one another can help you determine which to use for each type of promotion.
Hashtags can extend your reach, but do they annoy your audience or drive down engagement? You can find out with social media A/B testing. Don’t just test using a hashtag versus using no hashtag. You should also test:
- Multiple hashtags versus a single hashtag
- Which industry hashtags result in the best engagement
- Hashtag placement within the messaging (at the end, the start, or in the middle)
If you use a branded hashtag, be sure to test it against other industry hashtags, too.
Flytographer used two different tweets—one packed with hashtags and one with no hashtags at all—to link to the same content.
— Flytographer (@Flytographer) June 5, 2017
Best Friends, Now Sisters, Celebrate 45 in Miami! https://t.co/63MuY2Gc0N
— Flytographer (@Flytographer) June 5, 2017
This one is a little different, since rather than showing variations of your post or ad to similar groups, you show the same ad to different audiences to see which gets a better response.
For example, maybe A/B testing Facebook ads will show you that some target groups respond well to retargeting ads, but others find them creepy. Only testing theories like this can tell you for sure how your particular audience segments will respond.
Your targeting options vary by social network, but you can generally segment by gender, language, device, platform, and even specific user characteristics like interests and online behaviors.
Your results can help you develop specialized campaigns and a strategy for each audience.
For example, these two ads for the same product target two different audiences: agencies and startups.
This also works a bit differently, since you’re not creating two different versions and sending them out to distinct groups. Instead, you should monitor your profile on a particular social network to establish a baseline number of new followers per week. Then, try changing one element, like your profile image or your bio, and monitor how your new follower rate changes.
Try to post the same kind of content and the same number of posts during the weeks of your testing to minimize the influence of your posts and maximize the impact of the profile change you’re testing.
Airbnb, for example, often updates their Facebook profile image to coordinate with seasonal events or campaigns. You can bet they’ve tested to ensure this strategy helps, rather than hurts, their Facebook engagement.
You can also use social media A/B testing to help you make decisions about the content on your website.
For example, A/B testing social media images can provide a sense of what works best with a particular value proposition. You can use that information to influence which image to place on the landing page for the relevant campaign—but continue to test to make sure the image performs as well in the website context as it did on the social network you used for testing.
How to run an A/B test on social media
The basic process of A/B testing has remained essentially the same for decades: test small variations one at a time to discover what works best right now for your current audience.
The great news is that social media has made it much easier and more efficient, so you can run tests on the fly rather than waiting months for results to come in by mail.
Remember: the idea is to test one variation against another, then compare the responses and choose a winner.
Here’s the basic structure of an A/B test on social media:
- Choose an element to test.
- Dig into existing knowledge for ideas about what will work best—but never be afraid to challenge assumptions.
- Create two variations based what your research (or your gut) tells you, remembering to have only one element that differs between the variations.
- Show each variation to a segment of your followers.
- Track and analyze your results.
- Choose the winning variation.
- Share the winning variation with your entire list, or test it against another small variation to see if you can improve your results further.
- Share what you learn throughout your organization to build a library of best practices for your brand.
- Start the process over again.
Hootsuite can help you manage your next social media A/B test. Schedule your posts, track the success of your efforts, and use your results to adjust your strategy.