Spotify’s “Discovery Mode” campaign system has been pretty controversial since its launch. For added promotion across the platform, Spotify lowers an artist’s royalty payout—something that critics decry as streaming payola. Now Spotify is giving artists an AI analytics tool to “understand your campaign’s impact” and better prove its worth.

“Campaign Lift” uses machine learning to estimate stream growth by comparing actual numbers in Discovery Mode contexts against predicted baseline streams (without a Discovery Mode campaign) during the same period. The estimation is based on Spotify’s historical data, the artist’s monthly listeners and genre, and previous appearances in personalized playlists.

“This provides an up-to-date estimate of Discovery Mode’s impact during this campaign period by reflecting the rate at which streams in Discovery Mode contexts have changed compared to the streams we estimate your songs would have had without Discovery Mode during the most recent campaign period,” Spotify explains.

Artists enrolled in Discovery Mode campaigns with a song that has at least 500 streams “within Discovery Mode contexts” in the previous month can find the new metric alongside the existing “Historical Stream and Listener Lift” data. That tool compares a song in Discovery Mode’s streams and listeners to the figures from about a month (28 days) before its first campaign. Metrics are provided from the 11th day of the campaign until its end date.

Spotify also notes that all metrics update daily, and long-term engagement metrics continue to update for 14 days after the campaign ends. The company’s announcement further points out that while your songs are heard in Discovery Mode, “it’s important to zoom out to understand how they’re performing across all of Spotify,” and that “organic trends on and off platform can also have a big impact on your performance on Spotify,” such as viral social media posts.