How the SpotCheck bot-risk score works

What the score is

The bot-risk score is a 0–100 estimate of how likely a playlist’s activity is artificial, where higher means riskier. It’s a decision aid, not a verdict: it tells you where to look and what to weigh before spending money, drawing only on public metrics.

The signals it weighs

The score blends several independent signals so no single metric dominates:

• Followers against listening. Spotify gives every track a popularity score based on recent plays. When a playlist has over 10,000 followers and its typical track scores 5 or less out of 100, its followers aren’t playing what they follow.
• Followers per track. More than 10,000 followers for every track on the playlist. We set this aside when the tracks are clearly being played, so large brand and editorial playlists are not marked down for their size.
• Growth-pattern anomalies. Spikes and sustained drops in follower history, once we hold snapshots of the playlist to compare.
• Name and description. A generic name such as "chill vibes" on a playlist over 50,000 followers, a missing description, or a follower count in exact thousands. Each is a weak signal on its own.

Why history matters

A single snapshot can be gamed; a trend is much harder to fake. SpotCheck records follower snapshots over time, so the score gets sharper the longer a playlist is monitored — a flat-then-vertical-then-flat line is a strong artificial-growth signal that a one-off number can’t show.

This is why the score you see on a first check is a starting point, and why monitoring (Premium) tightens confidence over weeks.

How accurate is the score?

We don’t have an accuracy figure for real playlists yet, so we don’t quote one. What we have is a test set of 40 constructed cases. We wrote 25 of them to match known bot patterns, such as a bought follower package or an overnight follower spike, and 15 to match real curators, including small personal playlists. None of them is a real Spotify playlist.

Every change to the scorer runs against that set, and the build fails if a constructed curator reaches the medium band or if the scorer stops catching a bot pattern it caught before. That shows the scorer does what we designed it to do. It can’t show how often it is right about the playlists you check, which is why we are building a labelled set of real playlists and will publish accuracy from that.

The design choice stays the same: when SpotCheck tells you a playlist is risky, we would rather miss a subtle bot than accuse a real curator. A bot whose only tell is one weak signal, such as a suspiciously round follower count, can score below the medium band.

How to read it

Treat low scores as "fine to pitch," medium as "verify the specifics before paying," and high as "likely artificial — avoid." Always glance at the contributing factors rather than the headline alone: two playlists can share a score for very different reasons.

FAQ

Validate before you pay

SpotCheck scores bot risk on any Spotify playlist, right on the page. Free to check — see plans.