I recently wrote a piece for Billboard in which I analyzed the first week of airplay for Beats 1, Apple Music’s 24/7 worldwide radio station. While Apple provides playlists (through Apple Music’s Connect feature) for the programs curated by hosts like Zane Lowe and specials like St. Vincent’s Mixtape Delivery Service and Q-Tip’s Abstract Radio, there is no native way to see a list of everything on Beats 1’s airwaves.
Enter developer Callum Jones, who created @Beats1Plays, a Twitter account that accesses the metadata of the Beats 1 stream and sends a new tweet every time the station plays a new song. What results is a Twitter feed that gives up-to-the-minute snapshots of what is on the air. However, searching through a Twitter feed manually is cumbersome, and since Twitter only allows you to download an archive of your own tweets, not another account’s, I needed another way to access the tweets’ data in a way that would also let me sort and count the number of appearances for each song.
Having decided not to learn how to quickly use the Twitter API myself, I spent some time searching for solutions that would let me freely download someone else’s tweets and export them to a spreadsheet-compatible format. I eventually stumbled upon GrepTweet, a service built by Kai Hendry that lets you quickly backup and search tweets. (Grep stands for globally search a regular expression and print – more simply, search your data based on an inputted string.) One more obstacle presented by the Twitter API is that they limit access to just the most recent 3200 tweets. Since I only needed the first week of data (at a rate of one song every four minutes, that’s a little more than 2500 tweets), that issue was avoided, though GrepTweet also offers a daily backup if needed in the future.
To initially fetch an account’s tweets on GrepTweet, simply enter the account’s username and click the “Fetch tweets” button. GrepTweet will begin obtaining tweets, and you’ll be directed to a page like this: http://greptweet.com/u/beats1plays/ From there, you can search based on a string, such as an artist’s name, and results will quickly appear below the search bar. In order to get the most recent data, you’ll need to click the Update button first; once the new tweets have been fetched, you’ll be able to search the most recent set of 3200 tweets. You can also click the Download button to get the data as a .txt file.
After downloading the data, I pasted it into a spreadsheet. In order to separate the song and artist data, I split the tweet data based on a delimiter. (Since Microsoft Excel does not allow multiple-character delimiters, I first found all instances of ” – ” and replaced them with a symbol that wasn’t in any of the song titles. I then manually edited the few instances where ” – ” appeared in a song title.) After splitting the artist and song title columns into two, I removed the links and hashtags that followed in many of the tweets, leaving me with just two columns of artists and song titles. From there, I added a header row to my data, then counted the occurrences of each song and sorted the results to get an ordered list of the most-played tracks.
While I won’t be surprised if Apple (or Jones or another third-party developer) eventually builds a more native, natural way to see and sift through everything that Beats 1 plays in the future, the method above provides such a solution in the early days of the station’s history. Feel free to comment below or send me a tweet if you have any questions.