A partnership announced today between Knewton, a New York City startup, and Pearson, the world’s biggest publisher of print and online textbooks, may offer some help with this challenge. Knewton, founded in 2008, has created adaptive learning software that customizes lessons and textbook material based on what a student knows and what she needs to know to learn more quickly. Earlier in October Knewton raised $33 million in venture funding from Pearson and Peter Thiel’s Founders Fund. Knewton had previously raised $18.5 million from the likes of FirstMark Capital, Accel Partners, Bessemer Venture Partners and angel investor Reid Hoffman.
Pearson will be using Knewton’s software to power its online reading, writing and math courses for incoming freshmen in remedial classes. Some 750,000 students logged in daily to Pearson's MyLab or Mastering online series in 2010. “If technology can improve students’ chances of success, institutions can save money by graduating more students on time and doing less remediation,” says Greg Tobin, Pearson’s president of higher education mathematics group. He cites a “slew of data” that show significant improvement in student performance with an adaptive learning system such as Knewton: “I’ve seen one to two grade levels in improvement and 5% to 20% better student performance.” Arizona State University, one of Knewton’s test beds, is already considering cutting its remediation classes to half-semesters because so many students are moving through the material faster.
Knewton, which started in online test prep, created a way to “guess” the material students need to see or work on next, similar to the way online music service Pandora can guess which folk-rock songs you’ll like based on your preference for Neil Young. With a Knewton-powered textbook no two students see the same material. Paragraphs grow or shrink based on how students perform on quizzes and games. If your performance on cell division consistently drops after 15 minutes of work, Knewton will cut you off after the fourteenth minute. If you learn better with videos, you'll get more videos. Since math and science courses are easily broken down into related sub-categories, Knewton can make fine-grained correlations among students. Someone in the 95th percentile in equilateral triangles, for example, is almost certainly in the 94th percentile in isosceles triangles and probably doesn’t need a ton of help there, so it’ll guide them toward another topic.
The Pearson deal (and the big slug of cash raised) provides Knewton with a huge advantage in the adaptive learning category: steep switching costs. Every Pearson textbook will have to be tagged to work with the Knewton prediction engine. Pearson has close to half of the higher-education textbook market and an even bigger share in digital books. Pearson will provide Knewton with millions of its students’ performance data to improve the system’s accuracy and sensitivity. As the two companies expand their partnership into higher-ed courses and K-12 education, the plan is to eventually create next-generation textbooks that write themselves on the fly. Jose Ferreira, Knewton’s founder and CEO, acknowledges the switching costs to Pearson but adds, “Pearson is smart to be disrupting itself by investing in us.” (Knewton gets paid per student that uses the system.)
As textbooks go more digital, prices will drop just as they did in music and best-selling novels. One of the best ways Pearson has to maintain its pricing in textbooks is to increase its added-value by integrating its material with intelligent systems like that of Knewton.
