MusicWatch surveyed 5,000 US-based Apple Music users and produced some scary results for Apple. Once the survey results hit the headlines Apple was forced to respond, pointing out that MusicWatch's numbers were wildly inaccurate and rather than losing 48% of its trial users as the survey has suggested the churn was actually 22%.
I don't think that's an unreasonable number, in fact given the bad publicity and bugs that have dogged the service since launch, I think retaining 78% of its users is a pretty positive thing for Apple.
As I detailed here, Apple is in this for the long run and an expected conversion rate of 30% is about where Apple needs to be.
Any judgement on how the service will fare in the long run can't really be made after just a few weeks in which no-one has been asked to pay for using it.