The method

Music marketing,
measured.

Most marketing tools sell vibes. We sell math. Elevate is built on a public regression model with an R² of 0.91, which means we can predict your Popularity Index from a single number with extremely high confidence.

The model

Popularity Index from listeners.

Spotify scores every track on a Popularity Index from 0 to 100. In July 2025, three independent researchers analyzed thousands of tracks and showed that PI is almost entirely a function of one variable: listeners in the last 28 days.

PI ≈ -21.71 + 5.90 × ln(listeners + 1)

R² = 0.91. Mean absolute deviation = ±2.5 PI points.

What it lets us do

Inverse the equation.

Because the relationship is monotonic, we can solve in the other direction too. Given a target PI, we know exactly how many monthly listeners you need. Combine that with a real cost per conversion from Meta ads, and you have a budget. No guesswork.

listeners ≈ exp((PI + 21.71) / 5.90) - 1
Why it matters

The thresholds.

Spotify's recommendation systems gate themselves on PI. Reach the right thresholds and the algorithm starts doing the work for you.

  • 22
    Spotify Release Radar
  • 26
    Discover Weekly
  • 30
    Spotify Radio
Honest about uncertainty

No model is perfect. Ours has a mean absolute deviation of about 2.5 PI points. That is why every prediction we surface in the product is presented as a range, never as a single number. We'd rather be honestly approximate than confidently wrong.

Credits

The PI prediction model used by Elevate is based on public research by Marc Schmieder, Xixue Lou, and Andrew Southworth, Spotify Algorithm Analysis — Part 2, July 2025. The methodology and dataset are open-source on GitHub.

Read the research