The big knock on massive open online courses, or MOOCs, is that while thousands of people may register for them, only a sliver of students ever successfully complete them.

MOOC retention rates are appallingly bad. In many cases, just five percent of people who sign up stick with a course and pass it.

Why does this matter? Because of the democratizing, up-by-the-bootstraps, we-are-the-world potential of the MOOC, where anyone who wants to improve themselves can sign up to quaff the cup of elite education from their own home. The closely watched trend is really only two years old, but it was coronated before it hardly began, with the New York Times calling 2012 “The Year of the MOOC.”

The big MOOC companies — CourseraUdacity and edX – are well aware of the criticism. And each has figured out a way to put a good spin on the numbers. They all insist, and fairly so, that people who simply press a few buttons to sign up for a free online course aren’t equivalent to students who show up in person, or pay money for the course they attend.

At the same time, the three are working to develop their business models, so it’s crucial that they demonstrate that what they do is worthwhile. (edX is actually a non-profit, but it’s run just like a startup with 80 people out of Cambridge, Mass.)

Retention Report Cards

To bolster the argument in favor of MOOCs, each of these companies has been studying retention and completion rates for students. And they’ve reached some pretty obvious conclusions: People who sign up for a course certificate are much more likely to finish. People who pay for that certificate are also more likely to finish. And people who receive tutoring in pursuit of that certificate are much more likely to finish. So Coursera, Udacity and edX are differentiating a bit as they choose one or more of those areas on which to focus.

The people behind edX — which came out of Harvard and MIT and now has many more partnerships, including a recent one for leadership courses announced with the World Economic Forum at Davospublished a big paper this week about the demographics of their first year of courses.

What they found: a 5 percent completion rate across nearly 850,000 registrations. The authors argue, however, that focusing on the completion rate (which they don’t even bother to compute in the paper) ignores the benefits of low-touch education and the fact that many students accessed educational content they never would have.

edX president Anant Agarwal

edX president Anant Agarwal

Still, edX does pay attention to those numbers. And it’s starting to do a better job of tracking whether students are likely to finish, president Anant Agarwal said in an interview.

For a recent computer science class from Berkeley, edX split out users’ intent from the beginning. People on an audit track had completion rates of 5 percent, people who were seeking a certificate on the honor code had completion rates of 3 percent, and people who had their ID verified to receive a certificate had completion rates of 52 percent.

Meanwhile, Coursera also published some data on “the variability in student intent” last year.

For a Stanford science writing class, 24 percent of students who said in a pre-course survey that they planned to finish actually did so. Less than 2 percent of all students completed it successfully.

So Coursera began offering students the option to have their identity verified for a fee of $30-100 per course. It found students who registered for this “Signature Track” completed the course 74 percent of the time.

Udacity is taking a somewhat different approach, having honed in on the idea that mentoring and tutoring are key to retention. In tests of classes that provided hands-on help to each student, retention rates grew from 5 percent to 60 to 90 percent, according to CEO Sebastian Thrun.

Udacity has gotten a ton of press for the low pass rates for students who participated in remedial classes through a program it developed with San Jose State. But in Thrun’s view, the fact that 83 percent of students stayed till the end of the course was a marker of success, despite the fact that 50 percent of the class failed.

Thrun’s take is that tutoring encourages students to make a personal investment in a course, and stick it out. Obviously, it also promotes success. But at least initially, retention rates might have higher value than completion rates.

Staying Power

Which brings the conversation to business models.

Udacity CEO Sebastian Thrun

Udacity CEO Sebastian Thrun

“My general take is we, like others, made the initial promise to democratize education as much as we can,” Thrun said. “So that’s why we are sticking with free content. We draw the line at free services.”

Thrun said he sees a free MOOC course as the equivalent of a textbook. For many students, the real learning comes out of personalized coaching, so that’s the premium service. “The context of having people talk to you and mentors and people who hold you accountable makes a big difference,” he said.

So Udacity, which this week overhauled its website to bring coaching front and center, is now offering courses with a premium coaching option that costs $150 per month. The company is also experimenting with a more formal masters program for about $7,000 at Georgia Tech, where the first class of students was just accepted. Its goal: help students get technical jobs.

Paying in some form for courses appears to be the MOOC business model of choice, so far. “What’s really interesting is when we first started we expected an exponential explosion. We had 160,000 registrations without a name,” Thrun said. “But the number of students signing up went down, not up. Now we get 50,000 students in first round of a class. I don’t think an advertisement-type model would have saved the day for a free version.”

Coursera is also headed that way, but it charges less and doesn’t focus on tutoring. The company also has the biggest scale so far: 6 million students, compared to 1.6 million for Udacity and 1.8 million for edX.

Coursera also has most funding: $85 million, compared to $20 million for Udacity and $60 million for edX.

Coursera co-CEO Andrew Ng

Coursera co-CEO Andrew Ng

Diverging Goals

Asked about business models in an interview this week, Coursera co-CEO Andrew Ng said, “It’s not what we’re focused on right now. We think if we can change the world there will be many ways to make a profitable company. So far, the certificates financials look promising and growth is accelerating.”

Coursera is now focused on formalizing the credentials it provides. This week it launched specialization certificates, which are essentially series of courses like college majors or minors.

“One of the most fascinating trends has been the rise of credential value, and the fact that employers are increasingly taking MOOCs very seriously,” Ng said. “Companies have many unfilled positions, and want to hire someone who already has the skills for them. So the technology to deliver education at scale is increasingly appreciated for individuals to upscale themselves.”

But where the venture-backed companies are basically contractually obligated to dream big and vague when they talk about business models, edX’s Agarwal offers the least optimistic vision for financial success for MOOCs.

“I can see a path to modest revenue — $50 to 60 million in revenue would be perfectly OK,” Agarwal said. “I cannot see a path to revenue in the billions without doing things to upset our partners or selling user data.”