There was a time when music recommendations were fairly simple. You found out about new music by either reading about it in a physical newspaper, listening to it on the radio, viewing it on a TV, visiting a concert/festival, or following your friends suggestions. That was basically it.
Today, recommendations are almost a whole new business within the music industry. When the economic model changes from being based on copies and units, to listening and usage, the name of the game is to make people consume your music as much as possible
, rather than buy it as products.
Recommendation technology is a crucial part of the FuturePulse
project, specifically recommendation services for three kinds of stakeholders within the music industry: Labels (master owners), Live actors and Background music providers.
Playlists on streaming services have become extremely important as exposure vehicles for the music industry towards end users. Basically there are five different forms of playlists:
1) Curated content-based playlists,
2) Autogenerated content-based playlists,
3) Curated context-based playlists,
4) Autogenerated context-based playlists,
5) Private playlists.
are playlists that focuses on the content, f.e. which genre it is, which country it comes from, language etc (examples on Spotify are RapCaviar and Viva Latino, as well as top charts for each country).
are playlists that focuses on the situation surrounding the music, f.e. which mood you are in, if you are working out, eating dinner, relaxing, or any other context in which the music is being played.
One analysis showed that context-based playlists attracts more followers in general, than content-based, although content-based playlists constitutes for the major part of the playlists on Spotify. The median of followers for content-based playlists was 103,000, while the median for context-based playlists was 160,000. [i]
Still though, recommendations are mainly based on your own listening history in relation to what everyone else is listening to, rather than a more nuanced picture based on a larger ensemble of data. Weather and temperature, which time of year it is, time of day, if you are on work or at home, if you are alone or together with others, your feelings and thoughts at the moment, can have a huge influence on what kind of music you will be attracted to, when listening to it for the first time.
In one context, Foo Fighters "The Pretender" might be the perfect track, while in another context, Sigur Ros "Njosnavelin" might be the much better one, for the same person, regardless of the listening history. The more contextual data we can base our analysis on, the better our recommendations will be.
One of the use cases in the FuturePulse
project is focused on recommendations in regards to background music for businesses. Studies have shown that music adapted to a brand´s image and values can increase revenues substantially. [ii]
Depending on different contexts and variables, we want to recommend the best music available to establishments. But how do we choose that?
Context is really everything here. How crowded a public space is, what time of day it is, if there is a holiday, what the main age group is among the end customers to a retail store, recognition levels of the music among that specific target group, what kind of moods and feelings the music gives in the particular context, are such variables.
Recommendations are extremely important for businesses, since the "right" music can lead to increased revenues. Background music providers that can scientifically prove that their recommendation engines, in effective and automated ways, can increase reveneus as well as customer and personel satisfaction, have a huge advantage on the market.
conducted a survey earlier in 2018 among almost 800 small, medium sized, and large companies in USA, Greece, Spain, France, Germany, UK, Sweden and Italy. It showed that the most important attribute, according to the respondents, was moods
. How the music influences customers moods was seen as the most crucial. After that, energy levels
and levels of recognition
came as the second and third most important attributes. Therefore in FuturePulse
, we are focusing on finding solutions for how to identify these attributes for music and connect the right attributes with the right brands.
One other result of the survey was quite surprising, how many of the companies that were using consumer services rather than licensed background music provider platforms (see below). As you most certainly know, using YouTube or a private Spotify account in a business is a clear violation of the Terms and Conditions provided for by streaming services.
One could argue that with a much more fine-tuned and intelligent solution for music recommendations to businesses, companies would start to use valid services to a larger extent. The offer would not just be "background music", in the old traditional radio channel fashion, but rather "dynamic music recommendation that changes depending on context and increases revenues as well as customer and personel satisfaction".
That is the goal of FuturePulse. Stay tuned.
Researcher, Soundtrack Your Brand, Linnaeus University