Data can answer your big social media questions

You’ve likely seen a myriad of articles, suggestions and theories about the optimal frequency to be posting to your brand’s social media pages and wondered, “but what is right for my page and my audience?” Maybe you even read our blog post on that exact topic. Well, the good news is that you can stop worrying about how often everyone else is posting on social media because all that really matters is what your social media data shows.

In my previous post about posting frequency, I suggested that you determine how many times of day you should be posting to your pages by posting at the same times every day but varying the number of posts per day. Admittedly the content is likely to be different so you’ll see some natural variations, but if you have enough examples, let’s say 20-ish for each group, it will likely give you a general sense of what works and what doesn’t.

Collect your social media data

We’ll start with an example. Let’s say you weren’t sure whether you should be posting once or twice a day on Facebook. Your audience is full of moms who are up obscenely early with their little ones and at the end of the day those moms generally put their kids to bed and sink into the couch to zone out on Facebook around 8 or a little after. So you schedule your posts for the next couple of months for 6:30 a.m. and 8:15 p.m., mixing up whether you’re posting once or twice a day. Some Mondays you post at both times, and some Mondays you just post at 8:15 p.m.

Sort your social media data

Once you have your data in hand, start with Impressions because there will be impressions to track, and we’re testing whether or not the algorithm is showing your audience the content. Download the social post data from your test period and divide the posts into two groups. Group one will contain posts that were the only post that day. Group two posts were posted on days when there was more than one post. So now you’ll have two columns with post impressions.

Perform the calculations

Here’s where the math happens – we’re going to do a t test. A t test tells you the probability that something would have happened by chance. In other words, it tells you whether a result is unexpected. Fun fact: the t test was introduced in 1908 by a chemist who worked for the Guinness brewery and wanted to monitor the quality of their beer. To do a t test you can use your online calculator of choice. I happened to find this one with a quick Google search. For this calculator, I chose the data entry format “Enter up to 50 rows” since I didn’t have more than 50 entries. I copied and pasted my two groups of data into the calculator. Then, I chose the unpaired t test and clicked “calculate now.” This will take you to a page that gives you results that say something like the following:

P value and statistical significance:  The two-tailed P value equals 0.0184 By conventional criteria, this difference is considered to be statistically significant.

Interpret the calculations

The key piece of information here is whether it is or is not statistically significant. In this example, the calculator tells me there is a difference between my two groups. This means either it is likely better to post just once a day or likely better to post twice a day. We can figure out which by seeing which group has higher average impressions (the total impressions in your group divided by the number of posts). In this case, let’s imagine that posts on days with two posts received fewer impressions. Why does this matter? Let’s say I had someone pushing me to post twice a day to Facebook because they had to get their very important news out. I could say, if your goal is to get more people to see your very important information, let’s prioritize our posts. I ran a t test on our page’s data and there was a statistically significant difference between post impressions when there was only one post a day and when there were two a day. Ideally, they’re dazzled by your math skills and they realize that their very important content will best be suited for the next day.

If the difference was NOT statistically significant, in our example, that would mean we are not hurting ourselves by posting twice a day because neither option had significantly more or fewer impressions. If there was no statistical significance and I had someone pushing me to post twice a day to Facebook because they had to get their very important news out, I could say, “OK, that sounds great.”

Don’t get hung up on the details, unless you can, then do

This method of testing is not perfect. Not by a long shot. There are all kinds of variables we haven’t controlled for, but it can begin to give you a sense of what makes sense for your audience. If you start to see positive or negative results based on the actions you took, then certainly adjust your tactics. If you have enough time, resources and knowledge to conduct a thoroughly scientific and sound experiment for yourself we recommend doing so. This is meant as a quick way to check your own data and get a sense of what works for your audience.