How to set up content experiments to optimize social media content
When it comes to social media marketing, if content is King then optimization is Queen. Whether you’re an eCommerce company or an individual influencer, the goal should be to fuel your strong content and heal your weak content. And to do this, you need insights that expose what products, messaging, and images resonate with your customers. This post will highlight some best practices and useful tools to improve your social content through content experimentation.
What are content experiments?
Let’s use an analogy to talk about content experiments. Let’s say that you’re a baker (content marketer) and you have a ton of ingredients (your brand) in your kitchen to make the perfect cookie (content marketing). As a baker, you’re trying to make the perfect chocolate chip cookie, so you make a batch with a little more salt, and one with a little more sugar, then test those cookies on your friends to see which cookie they like best. You then take the most popular cookie to the county bake sale, because you know if it’s more popular with your sample group, it will also be more popular with your target group. This is like a simple content experiment.
As a content marketer, you probably have a good idea of who your customers are before you know exactly what kind of cookies they prefer. So running experiments is a great way to optimize your content on a small slice of your audience before taking it to larger channels.
The various parts of a content experiment:
If the first rule of running experiments is completely opposite to the first rule of fight club–you want to talk about it and track every detail. Start with a Google spreadsheet that your team can access and edit. Keep this bookmarked and try to visit it a few times a week, if not every day to keep your content experiments top of mind.
This is my personal favorite part of experimentation. Generating a hypothesis requires both strategic and creative thinking. Try to ascertain relationships that can be measured, and if optimized, improves one of your brand's strategic goals. Before moving forward, ask yourself, “does this hypothesis suggest an actual decision or action?” If not, dig deeper.
Example: I believe that social media posts with images of people perform better than those without people.
Outline the experiment and answer the who, what, where, and when of your test. As a team, determine how often and how long the experiment should take. A simple change to your content design might give you good results within a week, but a substantial shift in brand might take months. Your timeframe is reliant on the data you can produce. In quantitative analysis, 30 is the magic number. You need a sample size of at least 30 before you can reasonably expect an analysis based upon the normal distribution to be valid. So for social content, you’ll need at least 30 impressions. If you’ve ever taken a statistics course, this probably sounds familiar. But it’s really a rule of thumb because the larger sample–the better.
Example: I will run this test on Instagram by posting one image of our product with a person, and one without, one week apart. I’ll give each post one week to generate results. Each post will be posted Monday morning at 9 AM one week apart.
Execute the Test
Launch your experiment and wait for results, but don’t cut your experiment short if you don’t have enough data. Adjust the timeframe if needed. Cancel the test due to any confounding events, which are variables that are out of your control. For example, if you’re testing two designs on a social media platform, and that platform suddenly crashes (like during the great AWS crashes of 2021), then you should cancel your test and restart.
Quantitative analysis is much too large a subject to tackle in this post, so try to keep your analysis to simple observations. At the beginning of your experiment, you outlined a hypothesis. Do your results confirm that hypothesis? For example, if your hypothesis was, “I believe that social media posts with images of people perform better than those without people.” You should be able to certify that hypothesis based on data like engagement and CTR.
With Stagger, you can quickly analyze content embedded within your web platforms and see how they’re performing. Stagger’s data dashboard collects insights on how to optimize your story to your audience and where they’re interacting most, making it the perfect pairing for content experiments.
Rollout your Findings
Did your experiment confirm your hypothesis? Congratulations, now you can roll out new standard work to optimize your social content.
Summary of Tips for Running Social Content Experiments
- Track experiments in a place accessible by your team, like Google docs
- Aim for a sample size of at least 30. Adjust your experiment timeframe to reach the desired sample size.
- Use a tool like Stagger to track engagement on social and embedded site content.
Experimentation is as much of a practice as it is a culture. Getting started with experiments at your company is an exciting way to optimize your content, but it takes a lot of effort and consistency to do it correctly.