MOOC antagonists have long used the dismal completion numbers as a sign that MOOCs don’t work. MOOC providers have countered that retention is an outdated concept that shouldn’t be applied to online courses. Neither are right.
Why we should not care about retention: Retention is a very limited predictor of learning. It most reliably describes the students’ ability to stick with something they may or may not enjoy, or get value from. As such it might be a good indicator for grit, but not much more.
Why the MOOC providers have to care about retention: Because you can’t have it both ways. If you believe in the format of courses, which you presumably do, if that’s what you offer (e.g. Coursera, edX, etc.), then retention has to be a key indicator for your success. Maybe the wrong people sign-up, or signing-up is just like highlighting something in the course catalog, but in both cases it is on you to improve your product. Make it harder to sign-up, or offer a highlight function that does’t register you for a course, or offer two levels of sign-up (a few years ago, P2PU experimented with the options to “follow” or “participate” and, as expected, more people chose to follow). Or move away from courses…
Judging from MIT’s recent report on the Future of Education at least some of the key people are correctly shifting the focus of discussion from retention to the much more interesting question of how to design good learning experiences. The traditional course model will be only one option in the portfolio of answers.