What good actually looks like on SoundCloud

The numbers artists are measured against, published. Two ways: across the whole platform banded by follower size, and by scene — because a tech-house producer and a hard-techno producer of the same size are not liked, reposted or commented on at the same rate, and a single average tells one of them something false.

Everything here is measured, not asserted. The per-artist median is taken before the cross-artist percentile, so one viral track cannot drag a band, and the sample is snowballed through the repost and follow graph, so it carries a follower floor and describes working artists rather than SoundCloud’s long tail. We publish the provenance because a table of percentiles with no stated source is indistinguishable from the guesses it is correcting.

Measured across 2,000 public profiles, roughly 223 per follower tier, each artist’s whole catalogue read on 2026-08-28. Nine tiers rather than four, because most working artists live inside the old 1k–10k jump and were being judged against accounts ten times their size.

Repost rate

The number that matters most, and the one almost nobody measures. SoundCloud has no recommendation feed to optimise for — a track reaches people who are not already following you when somebody else reposts it, and in no other way. This is the one band that does not simply track size: it rises from a median of 0.09% under 500 followers to 0.35% around 1,000–2,500, then falls to 0.16% above 100,000. Only 17% of artists clear 1%.

Like rate

The cheapest signal a real listener gives, which is why bought plays show almost none of it — a like rate far below the band while plays look healthy is the clearest bot signature on the platform. It is close to a constant: 4.3% under 500 followers, 3.1% above 100,000. About 23% clear 5%.

Comment rate, and the figure everyone repeats

The common advice is that a healthy track draws comments on 0.5% to 2% of plays. Measured across 2,000 profiles, 28% of artists reach 0.5% and only 8% reach 2% — and 17% get no comments at all. The band describes the top quarter of the platform, not the norm.

Plays per follower

Not banded by follower count, and the reason is measured rather than assumed: the correlation between catalogue size and plays-per-follower is −0.61, against −0.38 for follower count. It tracks how much somebody uploads far more strongly than how good their audience is. The spread is 35x from one end to the other.

Followers per account followed

A credibility check rather than a performance one. A ratio near 1:1 on an account with real numbers is the signature of follow-for-follow, and those followers do not listen. Note how hard this moves with size: a median of 0.3 in the smallest tier against 3,922 in the largest.

The same engagement, measured within a scene rather than across the whole platform: 11,525 public artists, each placed in a scene from their own track tags rather than guessed from a label. At the same audience size the scenes genuinely diverge — under 1,000 followers, techno’s median like rate is 5.1% against 4.3% for hard dance, and techno’s comment rate (0.32%) runs close to double hard dance’s (0.18%). An artist held to the wrong scene’s bar concludes their music is the problem when the benchmark is.

Tech House

Repost, like and comment rate as a share of plays, by audience size. The median is highlighted; the range beside it is the 25th to 75th percentile — where most artists in the scene actually sit.

Techno & Hard Techno

The largest of the three samples, and the scene that engages hardest: like and comment rates run above the other two at almost every size. Techno reaches people through the repost graph, so the repost band is the one worth reading first.

Hard Dance & Hardcore

Hardstyle, rawstyle, gabber and the harder edges. A smaller scene than the other two — the thinnest group here rests on 158 artists — but every band still clears the floor we hold before a number is published.

How this was measured

Both halves read each artist’s catalogue through the SoundCloud API and take the per-artist median first, before pooling. The platform-wide bands rest on 2,000 profiles stratified so every follower tier fills to the same depth. The scene bands rest on 11,525 artists discovered by walking the repost graph outward from known scene artists and kept only when their own tags placed them in the scene — the graph says where to look, the tags decide.

A single scene cannot fill nine follower tiers the way the whole platform can, so the scene bands use four wider audience groups; a group below the floor is left out rather than shown as a thin figure. Where a signal cannot be read it is shown as absent, never as 0% — a gap and a zero mean different things.