The one constant of social media is change. Take a look at the standard Facebook or Twitter feed five years ago and compare it to yours today: the type of content that connects with users is always changing. Once you develop your social media marketing strategy, the easiest way to ensure your brand becomes irrelevant is to stick to the plan. To grow, you must constantly alter your content delivery and stay in touch with the people who value your brand.

Have you ever contemplated how a brand like Buzzfeed or Upworthy seemingly took over the internet in a couple of years? The answer is simple: they embrace zealous use of social media testing. Upworthy, in particular, writes 25 headlines for each article and tests them all before settling on one. You may not ultimately adjust your workflow to write 25 headlines for each piece of content you share, but the blueprint for success is there. When you conduct a social media experiment you will learn more about the type of content that increases engagement. Even with a small staff you can run social media tests regularly to hone in on the key elements of a winning social media post for your brand.

Are you ready to jump in? Before you craft your first A/B test, get some context:

Don’t Confound Your Variables

Before we detail actual social media tests that significantly grow your audience, you’ll need a quick statistics overview. Running a successful social media experiment necessitates the dilution of your data: if you run an experiment and don’t accurately interpret the results, the test is a waste of time and resources.

First, you’ll have to determine the one variable that you’ll be testing in your social media test. Many first-time experimenters who are unfamiliar with introductory statistics may accidentally test multiple variables at a time without intending to.

For example, let’s say you want to change the image in your Facebook post but leave everything else identical. Great! That’s a wonderful way to test the power of the images you’re using. However, if you post these two variations at different times of the day, you’re unintentionally testing the timing of the post as well, which will reduce the reliability of your results. Double check that there is only one differentiating factor in your social media experiment. This variable can be the time of day, but if so the posts should be identical in every other way.

Determine How to Measure Your Success

So you’ve settled on what variable to test. How do you measure how successful your two different social posts are? If you’re conducting a test on Twitter by modifying your profile picture or information, you’ll be looking for an increase in followers. But when conducting an A/B test with two different post, you should choose what metric you believe most accurately reflects success for you.

In some cases, this will be the amount of shares the post garners. Maybe it’s the most retweets, or the most overall conversions from the post itself. This will vary slightly from test to test – use your best judgment and choose the metric for your own means.

Outside factors like Facebook’s algorithm can determine how many impressions, or views, your ad makes in a given amount of time. Don’t rely on simple impressions as a determining factor for success – action is a more reliable metric that you can measure and increase.

Calculate If Your Result Is Significant

So, you’ve run your experiment and you’ve determined that, with post A, 88 people shared it out of 500 that saw it. For social post B, 190 people shared it for the 1000 that viewed it. It seems that, even when you account for the extra views on post B, percentage-wise it increased engagement. So, should you adopt the changes you made to post B going forward in your social media strategy?

The answer lies in crunching the numbers to learn if your results are statistically significant. The actual formula for calculating significance is fairly complex, utilizing standard deviations and other statistics measurements to determine a result. Luckily, for us marketers, there are multiple online tools that can quickly tell us what our results mean.

Two popular options for determining significance are tests from KissMetrics and VWO. We’ve inputted the hypothetical results for the above test into the Get Data Driven test via KissMetrics to yield the following results:

Screenshot from [http://getdatadriven.com/ab-significance-test]

In this case, the results are not statistically significant. The extra shares on post B are not enough to warrant a change in strategy. You would need to run additional tests to determine if that variable can increase your engagement or conversion.

Now that you have a handle on how a social media test is run, let’s cover a few tests that marketers love.

1. Alter Headlines to Increase Clicks

Choosing the best headline, as David Ogilvy famously noted, should be how you spend 80 cents of every dollar.

No matter how engaging and well-written your promoted content is, it will get lost in the shuffle if the headline doesn’t generate excitement. Trying out different headlines to get more clicks and views is perhaps the bedrock of social media testing, and is why, for better or for worse, writing the best headline has become an exact science.

Once you determine if you’d like to increase clicks or shares, write two compelling titles and determine what’s “different” about post B, the one you’re testing. Twitter is an ideal platform for testing headlines – with no images needed, you can focus on what text connects to users scrolling through their feed.

Split testing your headlines and tweeting both options at roughly the same time will provide you with enough data to determine which headline is best. Choosing the best headline, as David Ogilvy famously noted, should be how you spend 80 cents of every dollar. Make your headlines value-driven and include more actionable verbs to draw more readers in. By testing your headlines regularly you will learn what phrasing connects to social media users currently.

2. Tweak Design to Appeal to the Senses

By adjusting the design (mainly, the color or typography) of your post, you can measure users’ engagement with that color in the context of your brand.

There is plenty of science and research dedicated to determining the best colors to utilize in your social media marketing. This lengthy post from Conversion XL delves into multiple case studies for call-to-action buttons, examining what colors work best.

Often, on Facebook, Instagram, and other platforms, you won’t have a standard button in your advertising, but you’ll still need to determine what color connects most with your audience. By adjusting the design (mainly, the color or typography) of your post, you can measure users’ engagement with that color in the context of your brand.

When you set up your design-based social media experiment, take one of two routes: either, test colors that are diametrically opposed (like green versus red), or colors that are very close (seafoam versus teal). Going for two tests is generally best: test colors that aren’t similar whatsoever, then once you have a template color try tweaking it slightly to further optimize your success metric.

3. Shift Image Focus to the Customer or the Product

For brands that offer visually iconic products (food, cars, clothing) seeing the product is often more important for the customer.

This classic social media experiment is largely informed by your past marketing experience for your brand. While keeping the copy of your ad on Facebook (or whatever platform) identical, choose two separate images to split test. Answer this fundamental question: for your brand, do users respond positively when they see the advantages of the product, or a satisfied customer?

Cloud-based technologies or conceptual services sometimes see more success when their product is humanized. Featuring a smiling associate or giddy customer in your ad may prove more effective than a generic image of a cloud with an arrow. However, for brands that offer visually iconic products (food, cars, clothing) seeing the product is often more important for the customer. And of course, context is everything. As Hootsuite determined in their own social media tests, images of body parts other than faces saw more engagement overall for their brand.

Running this test is a starting point for many successful social media ads. Note that, with the amount of posts and related content your brands generate, these tests should serve as a guide, not an all-encompassing mandate. Don’t eliminate product-based imagery altogether based on one social media marketing test.

4. Reword Your Call-to-Action to Raise Conversion

By testing multiple calls-to-action you can find the verbs and the tactics that connect well across the board.

For social media marketing, your call-to-action is arguably the most important aspect of your post. When you directly appeal to viewers to engage with your brand, the phrasing is paramount. What tactic actually drives clicks, shares, and purchases? As with all industries, the answer is brand-specific. By testing multiple calls-to-action you can find the verbs and the tactics that connect well across the board.

Many brands today are thinking outside of the box, using absurd humor or an extremely casual tone to sound authentic. Breaking down the barrier between your brand and the customers is a great method of gaining clicks. But it’s only one of many methods. When testing modified calls-to-action, test two of the following strategies against each other:

  • Use humor to break down the barrier. Here’s a great example from Canva.
  • Demonstrate how your customer’s life will improve with your product. Citing studies or satisfied customers always helps. Such as: “90% of our customers stay asleep throughout the night. Are you ready for the best sleep of your life?”
  • Appeal to the fear of missing out on something beneficial. Without your product, something negative could occur. An example: “Don’t wait until it’s too late. Contact us about cloud storage services today.”

5. Consider the Time of Day to Expand Your Audience

To nail down the most beneficial social media marketing strategy, you should always determine what times of day and what days in the week see the greatest traction for your posts.

With the above social media marketing experiments, the two posts should generally be posted at the same time of day and be visible to different users. However, to nail down the most beneficial social media marketing strategy, you should always determine what times of day and what days in the week see the greatest traction for your posts.

If you use a product like Audiense Community Manager, you can determine the exact times of the week that your audience is online, and schedule posts accordingly to reach the maximum pairs of eyes per post and grow your audience as fast as possible. If you’d prefer to run tests the old-fashioned way, you can use the following guidelines from KissMetrics:

  • If you’d like the maximum amount of retweets, try tweeting at 5 pm on Wednesdays or on weekends. If you want to test morning engagement, use these times as your control and proceed accordingly.
  • To boost your overall click-through-rate (CTR), try posting either at noon or at 6 pm.

Perfecting your social media posting calendar should be your top priority before testing small differences in content.

With this road map you can finetune your marketing efforts online and reach more users by crafting engaging content. As you test new images and budget accordingly, The Shutterstock Editor will prove incredibly helpful: choose from our vast collection of stock images and create a compelling visual social media post in minutes.