An algorithm is traditionally known for their placement in math. The reference is found more often in regard to Artificial Intelligence, or A.I. It is everywhere across social media. Why? Well, it’s nothing short of them being the tastemakers of the 21st century, is all.
Let’s Begin This Chat Regarding Algorithms With Spotify
Spotify And The B.a.R.T. Algorithm
Spotify has their own proprietary algorithm. It’s shortened to the name B.a.R.T. from Bandits for Recommendations as Treatments.
There are three core ways in which B.a.R.T. processes data to better understand audience preferences. One is called N.L.P., or Natural Language Processing. N.L.P. analyzes words, intonations, languages and song content in order to offer the listener suggestions. It is using familiarity within national trending interests to make recommendations to be confident the listener will enjoy them.
After N.L.P. Comes Raw Audio Analysis
Second, there is Raw Audio Analysis. This is where the ‘mood’ of the song selected is determined by B.a.R.T. The decision is made based on similar characteristics to others and molds them into its own determination of ‘genre’.
Can’t Forget The Algorithms Collaborative Filtering Functionality
Also, Collaborative Filtering is in use by B.a.R.T. To me, Collaborative Filtering is the big brother-iest of all 3 analyses B.a.R.T. performs. This is because it compares your off-platform habits and interests in relation to new releases to determine future programming.
That’s just freaky. It takes all the fun and danger out of finding your own musical preference. If everything is plastic and pasteurized with regard to what’s being curated for you, what proof are you aware of to even back up this ‘fact’? Have you tried listening to wildly different music from your usual? And what’s the worst that can happen? A song comes on and it is not to your liking – Skip it! It shouldn’t ruin your day because a pop hit enmeshed itself within your hip-hop playlist.
Exploit And Explore
There’s 2 concepts with B.a.R.T. ‘Exploit’ and ‘Explore’ lead to how a recommendation lands in someone’s account. ‘Exploit’ uses all of a users data / activity to tailor playlists suitable to them. ‘Explore’ causes Spotify to hunt the Globe in pursuit of audio it analyzes and deems close enough to your taste to make a recommendation.
Also, the number 30 is important. It is B.a.R.T.’s cutoff time. If a song is played for 29 seconds that’s a ‘Nope!’ It’s not a fit. Hit that thirty second mark, though, and B.a.R.T. checks its box to make sure to provide you similar recommendations going forward.
You will see B.a.R.T. present when ending play of the list you have on now. Since it will automatically jump into another list its certain you will like based on data collected while analyzing your taste.
Releases And What The Algorithm Means For Your Music Marketing
The first 24-48 hours following a release are the most important to B.a.R.T. Forget your back catalog, buddy! No time for that old stuff; Move, move, move (I’m speaking as B.a.R.T. and just realized I never quite qualified that)! In that 2 days after release, B.a.R.T. pays extra close attention to listeners, a low frequency of skips of playbacks, and through play of tracks. The good stuff! More approval gained by your music in B.a.R.T.’s eyes, the wider it opens the floodgates to increase exposure whereas it was not absolutely certain this was a good move prior.
Tricking The Algorithm
A familiar term by now with a low success rate. I mean, otherwise, B.a.R.T. would be unemployed. I’ve tried to ‘trick’ the algorithm before. With enough people performing the same action, you can sway it here and there. However, do remember Spotify has 240 million subscribers. To think you and 30 friends will ‘over-run’ B.a.R.T. with your requests alone is nonsense.
Again, don’t listen to only me. I’ve been there, done that and developed a schema to go by. Never stop experimenting to further showcase your art. Even the most humble of amounts of algorithm trickery deserve exploration by every musician to determine validity. I even continue to dabble with my ongoing attempt to trick B.a.R.T.
If you think back to maybe January 2020, Daniel Ek stated that artists were not going to be able to survive soon with little output. He pondered on acts who release every few years and, essentially, said, “Well, try harder.” As a way of saying, “You can hear my album once I hear yours, Sir,” I decided to follow Ek’s recommendation as a challenge.
I began my ‘Song In Every Six Series’ where a new single is released every 45 days.
I output with this frequency for a number of reasons, but mostly algorithmic.
Why So Often?
Non-algo reasons would be ’I like to,’ ‘I do it as a big middle finger to Ek to say, “Ok, I did it – now what?”’ – It keeps me on my toes and always brings focus back around to music no matter how lost I may find myself.
Ek said to do every six weeks, so I figure the guy must know something about his own business. I could stand corrected, but we’ll pretend a moment. The reason why this was suggested is actually laid out above. New releases carry more favor from the algorithm, as do tracks which hit Discover Weekly.
I’m just waving my arms in the air as often as is reasonable to try to snag B.a.R.T.’s frequent attention. And it works mildly, I must admit. I find more listeners trickling in and on a more consistent basis following on of these releases. By spreading out our efforts to the recommended time frame, we’re cozying up to B.a.R.T. already!
Politics aside a moment, putting together an entire single, getting it mastered and running a campaign every 45 days is a fun challenge. It’s not as taxing as it may sound. If anything, you find a nice groove eventually. If able, you should do it to push and test yourself. It may make B.a.R.T. very happy without compromising any ideals you hold.
Instagram uses a more multi-faceted algorithm. Spotify’s is super, duper complex. Instagram needs to address more features than Spotify, though. How do they pull it off? And how does the Instagram algorithm work?
To understand this algorithm we would first want to understand what features it applies to. We are talking about feed posts, stories, explore page, IGTV videos and Reels.
6 key factors are considered for feed posts by the Instagram algorithm. Interest, relationship, timeliness, frequency, following and usage.
6 Key Factors On The Algorithm
Interest – Instagram is tracking you, no d’uh. Don’t be naïve. You could have read the terms and conditions to understand that. Through this tracking they ascertain a good idea of what you will like, watch and comment on/about. They aim to figure out where you will engage. For engagement, you are rewarded, since Instagram makes more money for the time users are on their platform for. There will be a coveted position atop people’s feeds for being interesting and chatting with people.
Relationship – Interactions come heavily into play here. The algorithm wants to figure out those closest to you and creates an association. From there, posts from these users appear atop your feed and vice versa.
Timeliness – How old your post is will also be a factor in algorithmic favoritism. Instagram favors the latest and greatest. WHEN is also a factor into timeliness. The time of day when you typically receive the most hits to your page and posting in accordance with that is favored as well.
Frequency – The less you use the app, the more you make the algorithm mad. Without regular posting it gives up trying to match you with the latest and greatest. Instead it swings for more general interests.
Following – You’re more favored for how many people to can draw to the platform. This is as gauged by likes/comments/shares/etc.
Usage – If you’re a ‘power user’ you will end up running out of suggested contents as provided by the algorithm.
When one reviews their bar of stories toward the top of the app, the other individuals profiles listed are the ones you interact with most frequently. They are most effective when used on a consistent basis.
Instagram Explore works nearly the same as your feed. The difference being Explore’s displaying of accounts you don’t interact with, but might enjoy doing so.
IGtv and reels prioritize towards familiar accounts you engage with as well.
The more social you actually are on this social media platform, the more the platform rewards.
Goals of YouTube’s algorithm – there’s two main goals. Find the right video for each user and get the viewers to keep watching. The presence of YouTube’s algorithm is most often seen in the form of search results and recommendations. However, we can’t discount the effect of YouTube’s homepage, trending videos and subscription notifications.
YouTube values engagement the same as the other platforms we have discussed. The uniqueness here is that YouTube also places equal weight on the metadata applied to each video.
Video performance drives analytics that determine content rank. Then, the YouTube algorithm matches content to viewer via their watch history and what other users in similar demographics to them have also watched.
The aim is not to turn this into a scoring system. More so, YouTube wants to put content into the hands of users who will appreciate it.
Twitter’s algorithm uses four indicators to rank users along its algorithm’s guidelines.
Those four indicators are titled
Recency is rather self-explanatory. Depending on how recent the Tweet is, it may or may not gain algorithmic favor against a Tweet similar in nature. The newer of these two example Tweet would be the one posted most recently.
Engagement, as we’ve seen, is a fundamental feature for each and every social media platform. Each platform decides how exactly they want to rank their users form the algorithm’s determinations. For Twitter, these determinations are based off of how retweets yours receives, likes on your Tweet, Mentions of your Tweet and Comments made on your Tweet.
The third criteria to making the Twitter Gods happy is to the use of ‘rich media’. Meaning, high quality video, photo and even written content.
Finally, the fourth factor, relies on your usage of Twitter. The more frequent a user you are, the more it likes you.
The Twitter Timeline Is Composed Of Three Aspects
There’s what is obviously sitting atop the feed thanks to algorithm favoritism, assumption.
Twitter also has a feature, ‘In case you missed it,’ where the algorithm determines a set of older posts which may interest you.
Last, you are presented Tweets in reverse chronological order, if you’d like to view it that way.
What’s interesting is Twitter permits you to turn off your algorithmic curation of posts. Highly unusual to read when reviewing the others, I KNOW!
Consistency is encouraged to boost ranking, posting at intelligent times, and posting high quality content. Also, be sure to engage in others’ posts. What the Twitter algorithm really likes to see is a rapid response time from yourself if engagement is sent your way.
Much like the rest of the algorithms we’re reading about, engagement on Facebook is also determined through things like likes, shares, and comments. What makes Facebook a bit more unique is that it will actually disregard posts it deems have no favor for the user/intended recipient.
The next part is so Science-Fiction I don’t know how to explain, except to use their exact language – Their algorithm Runs a ‘neural network’ to uncover behavior and it’s tie to user interaction. … Yes, what they said.
This is also something I find rather creepy about Facebook. Facebook presents a cross – section of media types knowing which one you’ll choose, for no other reason to provide you an illusion of choice and subsequent trust IN facebook.
Algorithm favoritism on Facebook again comes down to 4 key indicators.
There’s relationship. Relationship is your actual relation to the other party, be it family, in-laws, or really close friends. The algorithm shows them up-top your feed, expecting this is what you wish to see.
Content type is another factor. Depending on the preference of the user as gathered by the algorithm, this will be shown atop the feed. The least preferential sink to the bottom.
Popularity plays a solid factor as well. The more followers you have the further your message will spread, keeping users on Facebook’s platform which is their end goal anyway. For this, they again reward you.
Recency is one we also just discussed with regard to Twitter. Newer postings are given algorithmic favor over those which happen to older.