Last week, the test prep company Knewton announced that it had raised $33 million, bringing the total amount raised by the company to $54 million. And today, another test prep startup Grockit announces its latest fundraising: $7 million, bringing its total investment to over $24 million.

Clearly there's big money in test prep. It's a multi-billion dollar industry, fueled in no small part by the pressures on students to do well on SATs, GMATs, LSATs, and the like.

It's not a new industry by any stretch. The College Board first administered the SAT back in 1926, and companies aimed at helping students score better have grown as the demand for the exams has increased. And test prep isn't entirely about online opportunities either. Some of the biggest players in the industry -- Kaplan and Princeton Review, for example -- still offer offline courses and tutoring.

So what then do Knewton and Grockit do to warrant such massive investment? In part, it's the promise of adaptive learning.

Building an Algorithm for Learning

Both companies offer online services that provide a personalized learning experience by applying mathematical modeling to students' studying. Their software is "adaptive" as the content it presents changes based on the students' performance -- by their strengths and by their weaknesses. Algorithms dictate the most appropriate next question for students. The right content with the right level of difficulty.

The possibilities for more personalized learning are frequently touted as the benefit of these types of adaptive platforms. And while some of the research behind adaptive learning goes back over 50 years old, the increasing amount of computer-mediated learning nowadays certainly provides more opportunities for more analytics and more student data (and in turn, better models and better adaptive software).

I think it's interesting to note that one of the investors in Knewton's latest round of funding is Pearson. Think control of content plus curriculum plus assessment, now with an adaptive platform. That's a pretty powerful platform in the hands of the largest education company in the world.

Evolving Learning Analytics Platforms

I sometimes hear a lot of skepticism about adaptive learning. After all, how can "learning" be reduced to an algorithm? And while the models might work well for multiple choice tests, how can they be applied beyond that realm? Doesn't the emphasis on a mathematical, computerized solution here leave out the human and the social element?

One of the things I find particularly interesting about Grockit is that its platform very much takes the social into consideration. It describes itself as a "social learning" company, supplementing the individualized practice (the adaptive learning piece) with social learning opportunities. That means it isn't simply one student working alone with the software. Grockit also offers a way for students to work together (through Grockit itself as well as via its new Facebook app).

Indeed, despite the promises of adaptive learning algorithms to "get it right," Grockit recognizes that it's actually that social element that's just as important, if not more so, to student engagement but to their improvement. The startup has found that when students study with others, for example they spend more significantly time solving problems and they're more likely to get the answers correct than when they study alone.

This emphasis on helping match students (algorithmically or not) with others working on the same or similar problem set is something Grockit focuses on in addition to its work with learning analytics. I got a sneak-peak this weekend at a new Grockit feature -- Grockit Answers. The new tool allows you to start a Q&A from any video on YouTube. Rather than having to wade through the comments on YouTube itself (um, that would be the antithesis of social learning right there), the new tool lets you watch the video via a separate interface. There, you can ask questions that are time-stamped, so that your question is tied to a specific spot in the video. When others watch that video, those questions pop up for them at that very spot, and if they can help, they can provide an answer. It's a nifty way to combine both sychronous and asychronous feedback for learners. (And it's a good way too for teachers to get data about what point in a video students are getting "stuck.")

Big Bucks for Big Data

Despite the test prep industry being a massive one, I think that the explanation for investor interest in both Knewton and Grockit isn't simply about their getting a piece of the standardized testing pie (Pearson's investment aside, of course). I think it's also a sign of the increasing importance of data analytics -- something that's true in almost every industry right now, not just in education. While, true, "data-driven education" is already a political buzzword, I think we're just beginning to see the impact of what data and analytics can have in education. It's not surprising that the early movers here are two test prep startups: testing is the only data that some politicians seem to care about. But I think there is some very interesting data science going on here too which will hopefully prove lucrative to learners, not just to investors.

Audrey Watters


Hack Education

The History of the Future of Education Technology

Back to Archives