netflix-logoEnter a growing legion of middlemen, outfits that hook up companies (anonymously, if that’s what they want) with individuals or groups that might be able to solve specific problems.


Netflix, the Los Gatos, California, movie-rental outfit, did some reaching out in order to improve its Cinematch recommendation system, which uses customers’ rental histories to suggest other films it thinks they will like. The system already scored well with Netflix’s busiest renters, 90% of whom said they liked the recommendations. Netflix aimed to raise that satisfaction level by 10% and in 2006 announced a global online competition to find a better forecasting algorithm. It posted a sample of more than 100 million customer ratings for techie contestants to play with, and offered $1 million to whoever could hit the improvement target first.

“We wanted to make the best better,” says vice president Steve Swasey. “By throwing the challenge open in a worldwide contest, we got thinking from some of the best computer scientists in the world, and it cost us relatively little. We could never have afforded that much talent and work internally.” Three years and 40,000 entries later, Netflix announced in September that it had a winner: a seven-person team of software and electrical engineers, statisticians, and machine-learning researchers from Austria, Canada, Israel, and the U.S. The winners, who had started off as members of different teams and then got together online, came up with a solution that plugged that missing 10%.

The Netflix contest was the brainchild of CEO Reed Hastings, 49, “but it wouldn’t have happened without the board’s support,” Swasey says. The company has since launched another $1 million contest to improve its recommendations for more sporadic movie renters, this one using demographic information.

Not every company wants to advertise its needs or shortcomings quite so publicly, nor are many ready to invest the millions it can take to set up Web portals and train people to operate a direct online solicitation. Enter a growing legion of middlemen, outfits that hook up companies (anonymously, if that’s what they want) with individuals or groups that might be able to solve specific problems. InnoCentive, based in Waltham, Massachusetts, has a network of some 180,000 “solvers,” about 60% of whom have master’s degrees or Ph.D.s, says CEO Dwayne Spradlin, 43, though “anyone in the world is free to join.” They can visit the business’s website to check out the challenges posted on behalf of various companies—“seekers,” in open-innovation parlance—and the rewards being offered. InnoCentive, which takes a 40% cut of the winnings, claims a success rate of about 50%. Colgate-Palmolive used the company’s site to offer $25,000 to anybody who could dream up a better way to inject fluoride powder into a tube of toothpaste.

A few weeks later a Canadian hobbyist with a background in physics suggested imparting an electrical charge to the powder and grounding the tube. He won the reward, and InnoCentive collected its cut.

Other middleman organizations work in pretty much the same way. TopCoder, a Glastonbury, Connecticut, company, helped Lending Tree, an online mortgage originator, build its website, posting the specs of what the financial concern wanted on a site open to a global network of 225,000 software writers and graphic designers. Lending Tree also uses TopCoder to conduct regular “bug races,” in which it pays sums that can run from $10 into the thousands to the first techies who can sniff out and repair recently discovered software bugs. “It’s find it and fix it. Whoever does it first gets paid,” says TopCoder CEO Jack Hughes, 48. “This marks a significant shift in how companies get work done.”

It’s also cheaper. Lending Tree would need to employ a whole team of experts to address the thousands of small glitches TopCoder’s techies have identified and fixed. Relatively low costs have enabled some smaller companies to innovate on a shoestring. Example: SunNight Solar LLC, a tiny Houston manufacturer of solar-powered flashlights for the developing world that wanted to make an affordable solar-powered device capable of lighting a whole room. Mark Bent, 52, SunNight’s founder and CEO, took the problem to the InnoCentive site, offering a $30,000 reward to whoever came up with the best idea. After culling more than 40 viable responses, he picked a design from a New Zealand engineer that combines a polycrystalline solar panel, an environmentally friendly rechargeable battery, and high-efficiency light-emitting diodes. “Getting access to that wealth of talent from around the world is a game-changer for a company like ours,” he says.

There’s no other way we’d be able to do it.” The company is now making solar-powered room lights that sell online for $39; nonprofit organizations pay less. Boards eyeing open innovation need to be aware of potential pitfalls. One big worry: Who actually owns an idea submitted over the Internet? And how can a company make sure that a competitor doesn’t steal an idea floated in an open forum?

Many companies, including P&G, try to protect their intellectual property by breaking down their needs for large projects into smaller components to disguise the targeted, complete-picture outcome. Middlemen can also help on this front. InnoCentive posts specific needs and will protect a client company’s identity if asked to. It also requires its solvers to sign nondisclosure agreements and transfer all their rights in their ideas to the seeker, including the right to develop derivative products, says CEO Dwayne Spradlin. InnoCentive offers various protections to its solvers as well.

Open innovation may ruffle the feathers of a company’s in-house R&D teams, whose members might well feel threatened and could try to block or sabotage it. P&G’s department heads and developers were initially “incredulous that this could work,” recalls CTO Bruce Brown. “We’re a company that had incredible success developing things internally. Getting people to believe it could be done differently and still win was a major challenge.