Why subscribers cancel
Clustered cancel reasons and your save rate, and how to read them without discounting away the ones that are fine.
Every cancellation records its reason and is grouped with the others, so Retention AI shows your top causes ranked by count, alongside your save rate: what share of attempted cancellations you kept.
The clustering is deterministic on every plan, which means it works from day one with no AI involved. On the Growth plan and up, AI Cancel-Reason Insights refines the grouping and suggests what to do about each cluster.
Reading it honestly
The top cluster is usually the one you can do least about, and the second is usually the one worth fixing. "Too expensive" often means the value was never shown; "too much product" is a frequency problem you can solve by offering a longer interval; "no longer needed" is genuinely fine and should not be discounted away.
Related articles