The Making of a West Coast Swing Set

29 Mar 2023

A few weeks ago, I finally did something I’d been meaning to do for ages: I organized some music in a spreadsheet and used it to DJ for West Coast Swing. I’ve been DJing at Westie dances for several years, but I’ve never really organized my songs much – I think I was waiting to feel like I had a better sense of my overall DJing strategy. Roughly speaking, my process has been:

1. Maintain two WCS playlists: one for more typical West Coast Swing songs, and one for music that’s a bit weirder, or that I just heard somewhere not-at-WCS but I think might work.
2. Prior to each dance, construct a shortlist of songs ~1.5-2x as long as I need for the night.
3. Choose approximately the first hour of music and put it in a sensible order.
4. At the dance, pick songs from the shortlist depending on what seems to be resonating, what makes sense as a next song, etc. I’ll occasionally draw other songs out of my 2 main playlists, but I rarely do this.

That last step is pretty qualitative, but there are definitely patterns I tend to follow as I’m selecting “what would make sense next”. My goal in making a spreadsheet was to be able to lay those out a bit more programmatically, in order to make the process of choosing a next song easier and faster. I also thought this would be a good opportunity to open a discussion about DJing strategy for WCS, one of my favorite topics! I love seeing the differences and similarities in how different DJs think. I just saw a post from Andrew Roth the other day that also discusses some of these items, and here’s an oldie but goodie from Koichi that references BPM and energy too.

Starting at the end

This exercise, for me, was mostly about quantifying the qualitative traits I think about when I DJ. I’ll go into each of those traits in detail, but first I wanted to share the finished playlist: here’s the spreadsheet view, and the Spotify playlist which I put together for Portland’s new third Saturday dance at Dance with Joy. Most of these traits are totally subjective, so they reflect only how I hear each song and may not match with how you hear the same song. This includes the “genres” column, so don’t come at me for my genre labels!

The following sections will break down these columns from left to right, and then I’ll reflect on my experience using this spreadsheet in the wild.

Individual traits

Energy

The first two columns reflect the energy of the song – Andrew’s post has much more detail about why a song might feel lower or higher energy, but I am not a musician, so I just go by feel. I’ve rated each song’s starting energy and ending energy on a scale of 1-5, where 1 is lowest energy and 5 is highest energy. I noticed while I was assigning these numbers that I tend to conflate the actual ending energy of the music with how I anticipate the dancer will feel after dancing a song. This is most noticeable in a low energy fast song – even if the song itself ends on a low-energy note, my heart rate is up from dancing quickly for 3 minutes, so I feel ready for a higher energy song. I decided not to try to separate those two things out, so this is more of a holistic energy rating.

I think there are two key reasons that energy is important:

1. Sets where the energy remains the same for a long time feel stale and de-energizing. Even if the songs themselves are super high energy, if they’re all high energy, I find the dance overall to be ironically low energy. You’ll see this theme a lot in this post: variety is the most important thing to me when I DJ. I absolutely hate it when a night feels flat and “samey”; I find that super uninspiring, and just a recipe for a bad dance night! For variety in energy levels, I aim for a few energy peaks per dance, with the biggest peak somewhere in the middle of the dance. I like to ease people into a dance with medium energy, and taper out the end of the night with some lower energy songs.
2. It is jarring to go between two songs of very different energy levels. Imagine dancing to a sleepy cozy song, and then the next one starts with like a loud trumpet blast or something – unpleasant. This is why I chose to include two separate columns for the starting and ending energy. Many songs are basically the same energy throughout, but some come in very softly and ramp up, or start with a BANG but then settle down. When I’m joining songs together, I think about joining the ending energy of one with the starting energy of the next, aiming for a smooth transition.

I also translate these starting and ending energy levels into a “transition type” field (“rising”, “falling”, or “none”), which is a useful filter field for finding songs to quickly shift the energy level.

Here’s a visualization of the starting song energies of my set (darker yellow = higher energy):

I made the mistake here of transitioning way too slowly into my first energy “valley”, holding at a 2-3 energy level for too long early in the dance. I think I actually have this issue quite frequently: early in the dance, I’m mostly focusing on playing “easy to dance to” songs for those coming out of the beginner classes, and my personal comfort zone in WCS is around a 2-3 energy. I definitely noticed that energy lull during the dance and went “uh oh” – you’ll notice I get out of that energy valley much more quickly than I got into it hahah.

BPM

This is the only field that I mostly left alone after the export tool auto-generated it (aside from fixing some songs where the robot detected the wrong time signature or read a doubled BPM).

In my opinion, drastic swings in BPM are not jarring in WCS the way that drastic swings in energy level are. In fact, I think dancing two songs in a row that are the same BPM feels actively bad! If I do play two songs in a row with very similar BPMs, I choose different genres to increase the contrast between the two songs.

Another thing that I am careful to do is following up a very fast song (~115bpm or higher) with a song below 100 BPM, because I am not in excellent aerobic shape and I figure I can’t be the only one. On a personal level, I prefer slower songs anyway. This worked great for me at Mission City Swing, which is full of other people who largely dislike dancing fast, but I think Portland tends a bit faster than San Francisco on average. Or, possibly this is just reflective of an overall trend in the dance right now. When I first started (in 2013 or so), the typical Westie song was faster than it is now, then it slowed, so maybe now it’s speeding up again (much to my dismay).

Here’s a BPM visualization (darker purple = faster):

That white spot in the middle is a 65bpm song that I really enjoy, but in retrospect, was too slow. My bad – but it’s important to try new stuff! I’m also not sure why I decided to play those faster songs kind of close together towards the end. I’m guessing they were just well-received so I went with it? It’s unusual for me; in the past, I’ve found that people don’t want to dance fast at the end of the dance.

Genres

I don’t know if I’ll catch flak for this, but I think it’s important to label genres in whatever way makes sense for you individually. In this context, who cares if the song isn’t a real blues song, or if “baddie pop” and “angry cali late night” (shoutout to James Atwill for that one) aren’t real genres? I have a mental image of what those genres sound like, so they’re useful labels for me.

I think genre is what most people consider first when they think about “variety”. My personal genre biases definitely influence my sets, but I am also thinking about trying to play a wide range of music to cater to different tastes, and so the night doesn’t get stale. A lot of my set is pop because those songs tend to be crowdpleasers, but I don’t play more than two of the same type of pop in a row. Actually I try to avoid playing even two in a row, and when I do, they tend to be different in some other way. Just glancing through, I see two instances where I played two of the same genre in a row:

1. Ring Pop - Stripped -> You Shook Me All Night Long - Acoustic. These are both “singer songwriter”, but they’re very different BPMs and styles.
2. Let Me Love You - Acoustic Version -> Orion’s Belt. I have these both labeled as “R&B”, but the latter is much less acoustic and more late-night-y.

Both of these are also towards the end of the night, when I’ll let myself get a little more samey: “samey” also tends to feel “comfortable”, and that’s nice for winding down a dance night.

I thought it might be fun to chart these genres by frequency, as a kind of personal DJ signature. Songs tagged with more than 1 genre are over-represented here (I just counted them in every genre I had listed):

I’m a little embarrassed, but not surprised, by how well-represented “singer-songwriter” is hahah. Also, I guess I only like blues that is jazzy (whoops).

Acousticness

The export tool I used created a column for this, but the robot and I must have very different opinions of what “acousticness” means, because I ended up changing it for almost every song. I don’t have much to say about this column, I think about it roughly in the same way as “genres”. The interpretation is pretty intuitive too – imagine a dance night where every single song was synth-heavy? Or where the DJ didn’t play a single song released after the year 1995? It would get so boring!

Here’s the pretty heat map (darker blue = more acoustic):

As you can see, I like to keep this metric moving for most of the night. Similar to my genre comment above, I let myself get a bit more “samey” towards the end of the night.

Difficulty

I have to admit that I struggle with estimating the difficulty of a song. I’m trying to average a whole bunch of people’s dance experiences together, so there’s no way to really be accurate here. I assign a difficulty by imagining the song from the perspective of a lead who has only been learning this dance for like 3 months and has done no other styles of dance before. Based on what I’ve experienced with these dancers, here are some things that I think they find challenging:

1. Songs that are too slow (below 85bpm)
2. Songs that are too fast (above 110 bpm)
3. Songs that don’t have a “boom tic boom tic” drum section (like a flowy acoustic cover for example)
4. Songs that have a “boom tic” but also have other percussion on top of it which is distracting
5. Songs that have lots of surprising stops or weirdly timed phrase changes

Familiarity also impacts how difficult a song is to dance to (i.e. it’s easier to dance to a song that you know very well), but since I have a separate column for that, I’ve tried not to let familiarity influence the difficulty rating. I’ve rated the difficulty on a scale of 1-5, where 1 is a song I could imagine playing during a beginner lesson, and 5 is a song that even an experienced dancer might struggle with.

It’s tempting to think that lower difficulty is always better, but this is very untrue. Sometimes I just get SO SICK OF hearing “boom tic boom tic” in every song, and it’s all I can hear, and I feel like I’m going crazy. It’s also hard to push the boundaries of variety in other categories (e.g. genres, bpm) without getting into a higher difficulty range, and it’s important to me to expand the comfort zone of WCS a little :)

Since dances usually start with a beginner lesson, I like to keep the difficulty in the 1-2 range for the first hourish to encourage those newbies to stick around. I will sometimes throw in a 3, but I don’t think I would play a 4 or 5 in that first hour. After that, I start to play more difficult songs, but never too many in a row because I don’t want people’s brains to get too tired. Towards the end of the night, I drop the difficulty level slightly to make the dance feel more restful.

Here’s a map of the difficulty through the night (darker red = more difficult):

You can really see the ramp up and down pattern in this visualization! You can also see where I’ve sprinkled in some “level 1 songs” to give the floor a break between hard songs.

Familiarity

The single most difficult column to guess at, especially if you’re DJing in a new city! Sometimes I fantasize about putting together a big survey to find out which songs people actually know. I mostly just go by what I’ve heard other DJs play at conventions and local dances, with bonus points if the song is by a popular artist, a top-40s type song, or is just super predictable. At a recent convention, I danced to a song that I had never heard before, but that I SWEAR could’ve been generated by an AI that was fed the top 100 most-played West Coast Swing songs. Despite having never heard that song before, it definitely felt familiar.

My toxic trait as a DJ is to play too many unfamiliar songs. I have really dialed that back over the years so I think it’s okay now? But my personal favorite songs to dance to are ones that are new to me, or that I don’t hear played much at dances.

I don’t pay too much attention to the contrast between one song and the next by this metric; I just use this column to make sure there aren’t like 5 overplayed songs in a row. The column is also useful as a filter, since I use high-familiarity songs to revive the floor after an unpopular song, or to “reward” dancers after a difficult or strange song. Finally, I like to end the night on a familiar song, just to make sure everyone gets a nice comfortable last dance.

Here’s a map of familiarity through the night (darker green = more familiar):

Hmm, now that I’m looking at this, is it maybe still not dark green enough? Though honestly I have so little faith in my ability to guess these numbers anyway, maybe it’s meaningless. I’ll leave a link at the bottom for your anonymous feedback ;)

So how did it go?

Overall I was happy, but the system was a bit cumbersome to use. Since I DJ from Spotify, I had all this information as a separate Excel file. My file had 2 tabs: one with my shortlist, and a second with the actual playlist I was building. To add songs to my set, I’d copy/paste the title and artist cells from the first tab into the second, which would automatically pull in the rest of the fields and mark those cells green in the first tab to indicate that the song has already been played. To choose a next song, I found myself mostly filtering by “energy” and “difficulty”, and picking from that view.

The hardest part was having to remember to actually change the Spotify playlist after copy/pasting the song in the spreadsheet. I also had to search for each song I wanted in Spotify (the keyword search and sort by title view in Spotify are frustratingly incompatible with reordering the playlist). Maybe it’s finally time to get a DJ software, which I bet would make this process a lot less annoying.

Manually tagging each song was a little tedious, but it went pretty quickly once I got into the flow of it!

To be honest, I was surprised by how well the spreadsheet worked as a DJ tool. I was expecting to find a song after filtering the sheet, then listen to the actual transition to test it out and go “nah that’s not right”, but the success rate was actually quite high. If you want to give this a try too, feel free to message me and I’ll send you my Excel file and a link to the export tool I used (if you also DJ from Spotify).

Have feedback for me?

I am always happy to accept feedback via DM, but if you’re shy you can use this link to drop anonymous feedback (on either this post or my DJing)! Please try not to be too mean; I’m a delicate flower.