CD Baby has never had an annual fee. See http://members.cdbaby.com/cd-baby-cost.aspx It was always just a one-time up-front cost because for every incoming album, I (and later, others) would do about 45 minutes of work: scanning album art, digitizing the CD, spell-checking the bio and song titles, listening to some of the music to include in future recommendations, and finally putting the CD on the physical shelves in preparation for sale. Even with digital distribution, this up-front cost was still needed because a new album arriving meant many gigs of uploading out to 50+ different digital retailers. So, CD Baby's main profit model was a cut per sale. 9% of digital income, or a flat $4 per physical CD sold.

Tunecore's model has always been to take 0% or almost no fee per-sale. Instead that annual fee is their main income. I think you might be thinking about Tunecore, in your comment here.

Funny thing is (and I feel like I'm saying this confidentially, but fuggit, HN comments, here we go), I always thought Tunecore's model was kinda brilliant because it tapped into the ambitious musician psychology better.

At CD Baby, we'd often get emails/calls from musicians thinking of signing up, saying, "Let's say, conservatively, that we sell 100,000 copies. You'd be making $400,000 just off of our one album!"

I'd wince and say yes, that's right. Then I'd tell them that the average artist - https://sivers.org/lines - sells under 20 copies, not over 100,000. But everyone thinks they're the exception.

So that's the kind of person that would see Tunecore's model and think, "Ha! Only an annual fee and then I get to keep 100% of my 100,000 sales? Hell yeah! I'm going to save $400,000 going with Tunecore over CD Baby."

Ideally, a company could offer both pricing models, and let the client choose their optimistic or pessimistic sales prediction.