Understanding Recommendation Algorithms
Algorithms have made their way into our lives in so many ways.
What shows we watch on Netflix or videos we click on Youtube, and of course which song to listen to next. These are all decisions that are affected by algorithms, which are responsible for recommending the right content to the right people.
As a music industry pro, you know the importance of getting your music in front of the right audiences.
To that end, understanding how recommendation algorithms work, and how to use tools to maximize your efforts isn’t only useful: it’s crucial. Many would-be independent musicians know the sinking feeling that comes with seeing a release not get the traction it deserves. It’s not because the music is bad, it’s because of a lack of technical understanding, and tools.
So, let’s take a closer look at how recommendations are made, and how you can optimize your releases to show up in front of the right people- Your fans!
Overcoming the Cold Start Effect: Metadata Management for Releases
Recommendation algorithms use 2 different methods to determine which content to display.
- Content Similarities
- Behavioral Similarities
Content similarities are easy to explain on a basic level. If the listener likes listening to reggae, then the algorithm will recommend more reggae.
So, how does the algorithm know where to put a new release? The answer is in the metadata, but we’ll come back to this. To understand the importance of getting the metadata correct, let’s look at the other way in which algorithms work - by tracking our behavior and making comparisons.
Behavioral similarities take into account more factors and create profiles that they base recommendations off of. This algorithm style is known as the Netflix approach, in the data world.
For example, many people who listen to Biggie Smalls would presumably be happy to listen to more 90s era rap from New York, like the Wu-Tang clan. That is more of a content-based assumption, however - some people who like Biggie Smalls, also like Justin Bieber. There aren’t many similarities in the content, so how do algorithms account for that?
By tracking user data and then comparing behaviors, the algorithm can approximate these profiles to make better-informed recommendations. If several other people listen to Biggie and Bieber and you exhibit similar behavior on the platform, it might recommend some Biebs.
To make these recommendations on your tune, you need listens and to get listens, you need recommendations. One of those “it takes credit to build credit” situations.
That brings us back to the metadata.
Getting the metadata right on a new release is critical, and if you aren’t a data analyst, you might not know where to begin. For those who don’t know, the metadata is information embedded into the content, that identifies it to search engines. For music this often is presented in genre and mood tags: “Dark, Ambient Techno”, for example, would be an example of metadata.
Researching tags, genres, subgenres, and knowing what there is to know about metadata is almost a field all on its own. How do you as a music industry professional feel about picking up yet another professional skill set? No? Didn’t think so! That’s alright, we have you covered.
Using a tool like Reprtoir's Content Management System allows you to optimize your metadata perfectly. That means your release will be put in front of the right audiences from day one, allowing it to gain actual listens and get the traction you need to succeed.
From here, you’re going to want to think about optimizing for the behavioral similarities- and that is another blog post for another day. Targeting the right markets and pitching playlists goes in-depth and we’ll be sure to give that topic the attention it deserves.
But we aren’t done with discoverability yet! Social Media platforms present another way to be discovered, and of course, more algorithms to understand.
Social Media and TikTok’s Revolutionary Algorithm
TikTok’s ‘For You’ recommendation algorithm was ranked among mRNA vaccines for its innovations and is proving to be a powerful tool for musicians looking to be discovered.
The team at TikTok made their recommendation algorithm one of their top priorities and it makes sense that they would adopt such a strategy; Youtube’s recommendations are responsible for 70% of the usage on the site. Doubling down on the algorithm was a good plan. It drives the platform, which has seen explosive growth in its short existence.
For creators, there are a couple of factors to success that are consistent with other platforms - content and consistency. This poses a challenge for musicians when multiplied across different social media channels: it’s a lot of time and energy.
Reprtoir offers a content tool that allows music pros to integrate their strategy across multiple channels. The Content Management System (CMS) from Reprtoir, allows music pros to streamline their content output across multiple channels. This maximizes effort and saves valuable time.
It takes a multi-faceted approach to be discovered in 2022. Optimizing releases to be recommended to the right audiences is extremely important.
An all-in-one system like Reprtoir allows music pros to, stay organized, and spend less time gathering and organizing data and more time on what really matters - the music!
