Dear Creators, The Algorithm Is The Roulette Table In Zuckerberg’s Digital Casino
Overview
In today’s world it is undeniable that “Content is KING’. However its imperative to include that DATA is also apart of the royal ruling c-suite of our digital lives.
As a music artist in 2022, which by default means I’m also a content creator, I am constantly trying to figure out the best ways to make sure my content reaches as many people as possible.
‘Expert’ Barrage
If you are an artist (of any medium), athlete, chef, food blogger, educator, photographer, fashion model, or any other creator who takes their passion seriously then at some point while curating your digital footprint I know you have encountered those Content “Coaches” or Content “Experts”.
You know the ones. The ones who have the secret sauce and can just skip the line at “Club Algorithm” and uniquely speak straight to the boss in VIP. “Post at 8am”, “Post at 11am”, “Post and then close the app”, “Post 3 times a day”, “Post 4 times a day on days that start with the letter ‘T’”, “Post hanging upside down”, “Use 10 hashtags”, “Use 4 hashtags and tag your geo location”, “Post once a day and take it down and post it again in 2 hours”. The list goes on and on.
Outside of being outrageously tired of the advice barrage from the so called experts, my curiosity led me to wondering if an application of Data Science and Data Analytics could be applied to uncover if the social media algorithm operates with any rhyme or reason or is it just a casino where OF COURSE the more times you gamble the higher probability you have of winning.
Luckily for me I am also a software developer, who’s homogenizing the tech and music industries. Think what Kanye did for fashion as a rapper, I’m doing for tech. I realized this was an excellent opportunity to use software development to evolve, improve, and exponentially maximize my super sonic artistic endeavors and potentially help all creators.
Setting The Scene
For 3 months straight, from June 1st — August 31st, I posted once a day on TikTok and Instagram. The reports and data for this article are based on the findings from the Instagram data. I will post the reports and findings from the TikTok data in a follow up article.
I kept the data I recorded housed in a google sheet with the following categories:
Date — Calendar Date in M/D/YYYY Format. (i.e. 6/1/2022)
Day — Name of Day. (i.e. Thursday)
Day Of Week — Noted By either Weekend or Weekday (i.e. Weekday)
Time — Time Content Was Posted in Military Time (i.e. 4:47:00 PM is 1647)
Post Type — How Was Content Posted (i.e. Reel)
Content Type — What Type Of Content (i.e. Face_Card)
Accounts Reach — The Number of Unique Users That Saw My Instagram Post : #
Plays — If Reel, Number Of Times It Was Played : #
Accounts Engaged — Total Number of Accounts That Interacted With My Content In A Given Time Period: #
Profile Activity — Number Of Accounts That Took Action On The Post : #
Likes — Number Of Times Post Was Liked : #
Comments — Number Of Comments : #
Shares — Number Of Times Post Was Shared : #
Saves — Number Of Times Post Was Saved : #
The ‘Content Type’ category had the following types:
Face_Card — Post Showing Off Face
Music_Duet — Music Duets
Inspirational — Inspiring Post
Music_Funny — Funny Content Centered Around My Music
Relatable — Relatable Life Content
Web3 — Blockchain Based Content
Funny — Laughable Content With No Focus
Music_FrntPrv — Previewing Music (From Home Studio: Frontal Shot)
Music_Snippet — Previewing Music (Not From Home Studio)
Response — Video Responding To Comments
Hair — Post Centered Around My Hair
Day_In_Life — Vlog Type Video Content
Music_DIL — Vlog Type Video Centered Around Music
Music_Tech — Music Post Centered Around Tech
Music — Non Specific Music Related Content
Music_Performance — Music Content Centered Around Myself Performing
BTS — Behind The Scenes Content of Shoots and Other Creations and Processes
Face_Card_Skin_Check — Showing Off Face Apart Of My Skin Check Series
Gratitude — Post Showing Gratitude
Music_Filler — Lazy Music Related Content.
World — Authentic Aesthetics Welcoming People Into My World
The data points were recorded 24 hours after each piece of content was posted. (i.e. a post on Wednesday at 4pm would have its data recorded on Thursday at 4pm).
The main question I know my fellow creators, entertainers, and myself want clarity on is “WHAT IS THE BEST TYPE OF CONTENT TO POST FOR OUR ‘NICHE’?” and “WHEN IS THE BEST TIME TO POST THAT CONTENT?”
Here, BEST, is defined by the number of Accounts Reached and Accounts Engaged.
BEFORE WE GET TO THE FINDINGS, PREDICT WHAT TIME YOU THINK IS THE BEST TIME TO POST AND WHAT CONTENT DOES THE BEST. WE’LL SEE IF YOU’RE ACCURATE BY THE END OF THIS ARTICLE.
TikToks Influence
Ok, now to what you’ve been waiting for. Is trying to reach a larger audience on instagram a game of Blackjack where Zuckerberg is the dealer or not?
Over this time frame of 3 months there were 92 post in total. 60 Reels, 31 Non Reels (Photos, Videos, or a Collection of Both), and 1 Collaborator post which I ended up dropping from the data set.
It is no secret that since the rise of TikTok every social media platform such as Youtube and Instagram has made the shift to prioritize vertical videos. It made the most sense to post a majority of reels over this time because Instagram pushes Reels out to more people and I want everybody to experience the wave I’m creating.
Machine Learning
First I attempted to use python to do a linear regression on the data to see if there was a strong enough correlation between the variables of Time, Content Type, or Day on Accounts Reached and Accounts Engaged.
The only model that yielded a decent accuracy was Non Reels for Accounts Reached. This means if we input data into our machine learning model such as Day, Time, and Content Type it will actually be able to give us a ballpark number for how many accounts this post will reach.
However I personally felt there wasn’t enough data in the test set to really know how good these models were performing, so it was time to look under the hood, take a deep dive, and make our own deductions.
Besides, if LA has taught me anything this past year its to, “Never Trust The Models” ; )
Diving Deeper Into The Data
Time
First let us look at how the variable of Time affects Accounts Reached for Reels and Non Reels.
During the Weekend there is a significant spike in accounts reached at the hours of 8, 11, and approximately 15:30. During the Weekday there is a significant spike in Accounts Reached at the hours of 8:45 and 15.
During the Weekday there was a spike in Accounts Reached between the hours of 14 and 17 and then at the hour of 22. During the Weekend there was a spike in Accounts Reached between the hours of 10 and 11 as well as 16 and 18.
THE DEADZONE. Looking at both of the charts for Reels and Non Reels we can clearly see that there is a huge drop-off of Accounts Reached after 17 which is 5:00pm PST.
Content Type
Next let us see how Content Type affects our variables of Accounts Reached and Accounts Engaged.
Originally when I did this section of the analysis I had grouped all Music post together and all Face_Card post together. Pictured above are the results of that methodology of thinking. This initially had me believe the only well performing content categories are Responses and Relatable. I knew that was inaccurate based on how some of my Music and Face_Card post have gone ‘mini viral’. However my colleague Tyrone O’Garro re-performed the analysis while keeping the sections containing Music and Face_Card separated and his results provided excellent and more accurate insights.
Reels with the Response Content Type had the highest average Accounts Reached. Followed by Relatable, Music Tech, Music Performance, & Face Card Skin Check. Reels with the Response Content Type had the highest average Accounts Engaged. Followed by Relatable, Music Tech, Music Performance, & Face Card Skin Check.
The posts with the highest Accounts Reached also have the highest engagement. We ‘know’ that the Instagram algorithm will show your content to more users the more people engage with you. So yes, please ‘Smash That Like Button’ and ‘Comment Below’ your way to a bigger reach.
I wondered if the top categories would remain unchanged if the Content Type was the same, but the Post Type differed. Heres what the data says.
The post featuring the Content Type Face Card has the highest average Account Engaged and Accounts Reached. (This is excellent for my ego I am ashamed to say).
Post Type
What type of post should we be posting? Reels would be the obvious choice, but does that mean pictures, videos, or a combination or both should be rendered obsolete?
It shouldn’t be a surprise that Reels have the highest Accounts Reached, as stated earlier how TikTok has forced other visual platforms such as Instagram and Youtube to implement and maximize vertical video content. However what is interesting is the data on Accounts Engaged for Post Types.
The Content Type Video Slides had the highest average Accounts Engaged for any Post Type. Followed by Photo and in a close third is Reels.
Day Of Week
Lastly we get to the findings of how content performs based on if it is a weekend or weekday.
This variable yielded the largest disparity. On the Weekend Accounts Reached increased by 37.8% and Accounts Engaged increased by over 50%.
Conclusion
In conclusion we can clearly see that there is no magic potion comprised of Content Type, Time, or Day that when mixed perfectly together will produce a ‘viral’ moment or generate a surge in the amount of people who see or engage with your post.
Which in my opinion makes perfect sense because if there were any rhyme or reason to the Instagram Algorithm there would be less of an emphasis on the need for Ads. Which is quite literally its revenue model. It would make no sense for Instagram to have a defined method for its users to have their content shown to a larger number of people …. FOR FREE.
However we did find that there is a clear surge, around 40%, in Accounts Reached and Accounts Engaged for all Post Types posted on the Weekend vs the Weekday. Is this when people are more glued to their phone? Do they have the mental and emotional capacity to actually interact because the weight of the workday no longer exists? I cannot say, but regardless, to all my weekend warriors here’s a toast to you all for your support!
So to all my fellow artists, athletes, chefs, food bloggers, educators, designers, models, and other creators we are most definitely locked inside the Zuckerberg Casino.
I will be at the blackjack table making sure the entire world hears my sound.
Best of Luck To You All.
