You played the song for your friends and they said it's fire. You played it for your mom and she cried. You posted a snippet and your group chat went crazy. And none of that told you anything, because everything sounds good when you're in the room.
That's not cynicism about your friends. It's just how being in the room works. People react to you, your excitement, the social cost of shrugging at something you clearly love. A manager who built acts out of a basement said something that stuck with me: he never wanted his artists' fans' feedback. He wanted their patterns. Opinions lie. Patterns don't.
Patterns means behavior. What people do when you're not watching, when there's no social cost, when nobody knows the artist is counting. Here are three ways to get patterns instead of opinions, in order of effort.
1. The demo vote (free, works with any size audience)
Next time you finish a session with three or four unfinished ideas, post five to ten seconds of each and ask one question: which one should I finish?
Three things happen at once. Your audience feels ownership, because they picked it, and people show up for things they picked. You get engagement that would cost real money to buy any other way. And your release decision stops being a coin flip, because thirty strangers with no reason to be polite just told you which hook grabs.
The comments will surprise you. The one you were sure about will lose sometimes. Let it lose. That's the entire point.
2. The private listening group (for when you have even 20 real fans)
Take your most engaged people, the ones who reply to stories and show up in the comments every time, and put your rough demos somewhere private where they can listen. A private SoundCloud link works. A group chat works.
Then ignore what they say and watch what they replay.
Here's the pattern you're looking for: most demos will get played some baseline amount, roughly the same as each other. And occasionally one gets played three or four times as much. Nobody organized that. Nobody voted. They just kept coming back to it, individually, in private. That song is your single, whether or not it's your favorite, whether or not anyone said so out loud.
If no song separates from the pack, that's also data. It means keep writing, and it saved you a release cycle spent pushing a song that was going to plateau.
3. Streams per listener (for songs already out)
For released music, Spotify hands you the pattern for free, and most artists have never looked at it.
Take any song's streams and divide by its listeners for the same period. A normal song sits somewhere around 2 to 2.8 streams per listener. People heard it, some played it twice, life moved on.
Now check every song in your catalog. If one is sitting at 3.5, 4, 4.5 streams per listener, stop scrolling. That number means individual humans are choosing to play that song over and over, on their own, with nobody watching. It's the released-music version of the demo that got replayed four times as much, and it's the strongest promotion signal you own. That's the song that deserves the content push, the ad budget if you run ads, the pitch. Not the newest song. Not your favorite. The outlier.
I've watched artists spend months promoting a new single to lukewarm results while an 18-month-old track quietly sat at 4.2 streams per listener, begging to be pushed. The audience already voted. Nobody counted the votes.
Why this is emotionally hard
Every one of these methods risks the same painful outcome: the data picks a song you didn't. The throwaway beats the labor of love. Something in you wants to overrule it.
Don't, and here's the reframe that makes it easier. The data isn't judging your art. It's telling you which door the audience wants to walk through to reach you. They come in through the catchy one, and then they find everything else. The deep cut gets its audience because the single built one. Every catalog that people binge works this way.
And there's a second, quieter benefit: this protects your nerves. Releasing on a guess and watching it flop grinds artists down worse than almost anything. Releasing the song that already showed a pattern doesn't guarantee a hit, nothing does, but it means you're never spending your one monthly push on a coin flip.
The Monday version
If you want this as a habit instead of an article: once a week, look at streams per listener across your catalog and note the outlier if there is one. Before any release, run a demo vote. When a release is looming and you're torn between songs, private listening group, watch the replays. This is exactly the kind of pattern-reading a manager would do for you weekly, and it's a core part of what I built Musuni to do for artists who don't have one. But you can start with a calculator and ten honest minutes.
Ask for plays, not opinions. Your friends love you. Your listeners' repeat behavior loves you accurately.