My Analysis of USAFA Memes on Instagram

Last week I spent like four hours going through all of the Instagram posts I had put together with one thing in mind:

I need to figure out what is funny to grads and cadets

It was an iterative process and I used my own methodology. You or some other expert on YouTube might do it completely different than me, but I figured I would write out my process in case it can be helpful.

START WITH A FILTER

I had a hunch that looking at every post would be both too time consuming and introduce a lot of noise so I wanted to come up with a filter for the posts I would be looking at.

300 likes seemed to be a cutoff point between “meh” and good, so I decided to only record data related to posts that received at least 300 likes. That number also represents roughly 10% – 15% of the accounts audience, so double-score.

ESTABLISH YOUR CONTENT DOMAINS

Up to this point in the Instagram account’s life, I have been coming up with ideas and posting them without any rhyme or reason.

The backstory there is that this project is a fun outlet for me and I want to keep it that way. I know myself and one easy way to get me to stop caring is to make it feel like a job. 

I have chosen not to do that, and it is working for me. 

At some point, you’re going to need to come up with some content domains and I ended up coming up with these domain categories, so that I could capture different aspects of the post to get a more detailed understanding of what is funny and what is not based on Likes and Shares:

• Date of Post – self explanatory
• Meme Description – short phrase to describe the post
• Subject – arbitrary breakdown of meme description
• Style – video, static image, or carousel
• Theme – additional way to group like content around Style and Description 

MORE INFO BEHIND MY CONTENT DOMAINS

I came up with a few extra “data domains” to be able to slice up the memes and better understand what is funny.

It could that there is actually a standard way of doing this… something for me to look into for sure. But this is what I came up with and some additional detail behind it all

Subject

I just looked at the posts and then over the course of my four hour review it became clear that there were underlying themes in each post that could be grouped together. Here is a table of the subjects so you can understand what I mean:

If you’re a cadet or a grad, this passes the “sniff test” right? There are things that were just universally funny about the United States Air Force Academy that 10% -15% of us can all Like and agree on enough to click the Like button.

Style

I thought that post style is interesting to capture because I have spent A TON of time editing video posts that I think are really funny and they either come up shorter than I think they should OR end up flopping altogether.

Sidenote: I know that I am supposed to be making Reels because Mark Zuckerberg wants me to – but I never liked Reels with the types of memes I wanted to create. Now I am thinking that there may be more of an opportunity to use Reels, so I am kind of excited to get into them.

That all being said – the four Styles of posts that I have typically made include:

• Static Image Posts
• Static Image Slideshow Posts
• Video Posts
• Video Slideshow Posts

Pretty straightforward.

Theme

This is interesting to me because I had the Subject domain but it was causing some problems and I really wanted to tease out additional information about the posts – that should make sense when you see the list below.

The main takeaway is that I wanted to get at an additional aspect tied to the core of the post, in hopes that it might shed some insight into what is funny enough to like.

Record your Metrics

The next step for me was to go through each post with 300+ likes and record all of the metrics that Instagram captures. 

Note that I decided to record these metrics one to two years after some of the content had been pushed out on Instagram and I think that meant that a lot of the older data turned to zeros… either that or they changed what they were tracking and reporting in 2023. One of those.

• Likes
• Comments
• Shares
• Saves
• Reach
• Engaged
• Profile Activity
• % Non-Followers*
• Follows

*The % Non-Followers metric is really interesting to me – and I am thinking that it is a way to increase engagement – which has probably been written about a ton and I just haven’t looked it up at all because I don’t know that i need a 19 year old girl telling me how to “game the algorithm” right now. It’s not really what I’m trying to do, but I am interested if I can use it for content ideas. 

Pivot Your Data Set & Analyze the Results

The whole point of creating the Subject, Style, and Theme domains was so that I could pivot the metrics data around them. The whole point of doing this is to understand what my audience likes so I can give them more of it.

The extra categorization makes analysis a bit easier.

I took the three data domains above and pivoted the data around Post Count. I then added Likes and Shares, which are two metrics that I think are important indicators for growing the page’s audience:

The point of this exercise is to understand what is funny, and the methodology is to compare Likes and Shares among different content types.

My thoughts are that by looking at the percentage of posts in a certain domain, compared to the total percentage of Likes and Shares – I can start to see:

1. What content is working and I need to keep doing
2. What content is punching above its weight class and I need to do more of

My Analysis

Based on this work – Here is how I am interpreting the pivoted data set:

Subject

When you look at the Subject data pivot a couple things jump right out at you related to Likes and Shares. Let’s break them down:

Posts that make fun of Air Force’s rivalry with Army and Navy account for slightly more likes than their post share.

Posts that make fun of the following topics account for WAY more shares than their percentage post share:

• Mitchell Hall
• USAFA being a prison
• Air Force Football
• The Honor Code

Style

The only thing that jumps out when looking at Like and Share data related to the post’s style is that Static content gets shared. 

Almost three quarters of the account’s shared posts were still images, which could explain why most mil-meme accounts prefer creating still image content and “hate Reels.”

Theme

As a reminder, this was a content domain I created to tease out additional characteristics of each post and group them together.

There was not much going on with the Likes, however there are a couple insights around posts that were shared that emerged.

• Current events get shared at twice their rate of being posted.
• Contemporary meme designs and comparison posts get shared slightly more than their post rate 

Bringing it All Together

When I started this exercise, I thought that I might find some definitive answers on what cadets and grads think is funny.

That wasn’t exactly the case, but what I did find was helpful.

There are definitely post subjects, styles, and themes that do better than others. And I can keep that information in mind as I continue to try to make cadets and grads smile, moving forward.