You have an album out, or five finished songs, and one of them has to be the next single. Everyone around you has an opinion, and so do you. Data won't make the decision for you, but it can show you which songs people already respond to and save you from choosing by mood alone.
Test snippets before you decide
The cheapest test is a short clip. Post 15 to 30 seconds of two or three candidate songs on TikTok, Instagram Reels or YouTube Shorts and watch what happens. To keep the test fair:
- use the same kind of video for each song, for example the same camera setup, the same length and a similar caption
- post on different days at similar times, instead of all three in one afternoon
- pick the part of the song that works as a hook, usually the chorus or the first line
- give each clip a few days before you judge it
Don't stop at views. A view count mostly shows how far the platform pushed a clip. Compare the things that need an effort from a viewer: shares, saves, comments, profile visits, and especially comments such as "what is this song?" or "when is it out?". Those tell you someone wants the full track.
Keep the sample size in mind. Three clips with small differences prove nothing. A clear gap, repeated across more than one platform, is worth taking seriously. If the result is a tie, that is information too: the choice comes down to the other factors below.
Read what your album tracks are doing
If the album or EP is already out, you have free research. Look at each track, but not only at total streams, because the first songs in a tracklist and the ones on playlists collect more plays for reasons that have nothing to do with quality.
Look at two other things where the platform shows them:
- Saves. A listener who saves a song wants to come back to it. A track with few streams but many saves compared with its listeners is a possible hidden single.
- Repeat listens. Divide streams by listeners to see how often the same person plays the song. A high number means the song holds people.
Also check countries and platforms. In the CULT dashboard you'll see streams by platform and country for each reporting period, once the partner reports are checked, with the source and period shown on every report. A song that is quietly strong in one country can be a reason to choose it for that audience. Saves and playlist adds you'll find in the platforms' own artist apps, where they are shown.
Remember that reports arrive with a delay, so give a track a few weeks before you decide it has no future.
Listen to your audience in words
Numbers don't contain sentences. Read your comments and messages for repeated phrases: "this line", "play this at the wedding", "I can't stop singing the chorus". At shows, watch which song gets the phones in the air and which one the audience sings back. After a concert, ask a few people one simple question: which song would you send to a friend? It works better than "which one do you like".
Write the answers down. A note file with three columns, song, what people said and where, takes ten minutes and gives you something to compare.
Add your own instinct
Data tells you what people respond to today. It doesn't tell you who you want to be next year. A single will represent you for months: you'll play it live, make content about it and explain it in interviews. Ask yourself:
- Is this a good first impression for someone who has never heard me?
- Does it fit the direction of my next release?
- Can I make several different pieces of content around it?
- Will I still want to play it in six months?
A simple scorecard helps. Give each candidate a mark from 1 to 5 for audience response, first-listen clarity, fit with your direction, content potential and how much you still love it. Add the marks and look at the result, but treat it as a conversation starter, not a verdict.
When data and instinct disagree, you don't need to choose one forever. Release the song the audience responded to as the single and let the one you love be the focus track on the album, or the next single.
Mistakes to avoid
- Choosing a song because one clip went unexpectedly viral. Ask whether the song itself deserves the attention.
- Ignoring context. A ballad will lose a test against an upbeat track on short video, even if it is the stronger song for your core listeners.
- Waiting for perfect data. Set a deadline, for example two weeks of testing, then decide.
- Changing your plan every time a number moves. Pick a method and trust it for this release.
A short checklist
- Post snippets of two or three candidates in the same format.
- Compare shares, saves and comments, not only views.
- Check saves and repeat listens on tracks that are already out.
- Collect what people say at shows and in messages.
- Score the candidates against your own direction, then decide by a set date.
Pick one, announce it and move on. A good single released on time beats a perfect one that never leaves the folder.