Skip to content
← Song by songWhen a song doesn't work

By ·

When a song doesn't work

A song can fail to work in two ways, and for a long time I mixed them up. One is when it doesn't work while I'm making it. The other is when it doesn't work after it's released. They feel alike in the disappointment, but they call for almost opposite reactions, and confusing them led me to some pretty bad decisions.

When it doesn't work in the workshop

The first is the most common and the least dramatic. I have a lyric I like, I start working on the music and something doesn't fit. The production heads one way and the lyric another. Or the lyric, which looked funny on paper, falls flat once it finally plays.

Since I produce the music with artificial intelligence tools, lots of people think this is fixed by pressing a button until something good comes out. Sometimes it is, true: I generate another version, change the approach, try a different style. But very often the problem isn't the music. It's the lyric. And no tool is going to fix a lyric that doesn't know what it wants to say.

I've learnt to ask myself one question before trying more productions: does this lyric have one idea, or several fighting each other? Almost every time a song resists me, it's because there are two songs crammed into the lyric. One verse tells one thing, the chorus another, and the music doesn't know which to follow. When I see that, I stop touching the production. I go back to the lyric and pull them apart.

Sometimes a song that didn't work turns into two that do. Other times one comes out and the other stays in the notebook forever. That's fine.

There are also songs that, quite simply, won't. I look at them from every angle and can't find the fault. It isn't the lyric, it isn't the music, it isn't anything specific. They just have no life in them. With those I've learnt not to insist too much. I give them a chance or two and, if they're still dead, I let them go. It's a cut that costs me less than it used to. Discarding is part of the job, and in my case a big part: I decide, I discard and I choose, and the verb in the middle is the one I use most.

When it doesn't work out there

The second kind is harder. The song is finished, I like it, I release it with its music video, and nothing happens. Nobody listens, or very few people do, and it sinks straight away under the ones that follow.

At first this hit me hard. I thought the audience was right and the song was worse than I believed. I'd listen again hunting for the flaw, and since when you look for flaws you always find them, I'd end up convinced it was bad.

Now I think that conclusion is almost always unfair. A song not working out there doesn't mean it's bad. It means it hasn't found its people, or it came out at a bad moment, or the title didn't invite anyone to stop, or a thousand things that have nothing to do with quality. I've seen very simple songs of mine travel a long way and others I think are much better stay put. Lots of people discovered me through «Tan Cateto» on TikTok, not through the song I'd have chosen to introduce myself. That taught me humility about my own judgement. I know a lot about my songs, but I don't know which one will find whom, or when.

What I do with them

What I don't do is delete them. A song that hasn't worked is still there, with its page at mikesaintlouis.com, its full lyrics and its music video. Someone might find it a year from now while looking for something else. The website has changed how I see this: songs no longer depend only on the week they came out. They have a fixed address and they can wait.

What I do is give them another life when it makes sense. The GAYFOLK trilogy, for instance, let me arrange songs inside a story in three acts. A song that went unnoticed on its own can work differently when it's in its place within a narrative, surrounded by the ones before and after. I'm not saying that always saves it. I'm saying it gives it context, and sometimes context is all that was missing.

I learn from them too, though less than I'd like. I try to see whether the ones that don't take off have something in common. Sometimes they do: a type of title that's too cryptic, a music video that takes too long to show what it's about. Other times there's nothing in common, and I have to accept that luck gets a vote too.

The difference that matters

What has helped me most is this. When a song doesn't work in the workshop, the problem is mine and I have to fix it. When it doesn't work out there, the problem often belongs to nobody, and all I can do is keep it available and carry on.

Mixing the two up led me to rewrite released songs in my head, to punish myself for decisions that were well made, or the other way round, to blame the audience for songs I knew perfectly well weren't finished. Keeping them apart has taken a lot of weight off me.

I still feel a twinge when a song stays put. I suppose that's inevitable. But I no longer take it as a verdict. I take it as a data point, the kind you note down and then half forget.

All of them, the ones that took off and the ones that didn't, are still at mikesaintlouis.com.