Lots of talk about redesigning websites lately. Maybe it’s because summer is ending, and the Holidays are right around the corner (for e-Tailers, that is)?

First, there was Jeff Sexton’s post about asking the right Persuasion Architecture questions before redesigning, which was inspired by a Seth Godin post. Then, Jakob Nielson had some good thoughts…

The recently awarded Netflix Prize reminds us that if you have a data-driven problem, it's possible to tap into a vast market of talent to deliver improved results. The internet makes it very easy for groups from around the world to self-organize, and then to self-sychronize to solve your problem. But as Greg McAlpin — the leader of the group that lost the prize by just twenty minutes — notes, most crowd sourcing efforts fail. In the case of the Netflix Prize, there were over 51,000 contestants organized into over 41,00 teams in 186 countries. The brilliance of Netflix management was in crafting a contest structure that made it very easy for them to "grade" the results. They also made the goal clear — at least a 10% improvement in film recommendations.

This approach is contrary to how corporations are programmed to operate. No organizational system could afford to screen 51,000 participants and 41,000 teams! Only a market can do that. Furthermore, the teams decided among themselves when they would collaborate. The better groups waited to see which teams were performing well, and then merged with them to improve performance — again, not a behavior that a companies usually foster. Lastly, the folks who came in second did not win a consolation prize. They lost by 20 minutes — and they lost. Again, a company would have a hard time enforcing such a strict standard after their team's hard work and operable product.

As my friend Tom Malone, over at the MIT Center for Collaboration, has noted, there is no magic to crowdsourcing — but you need to design what you want from it. I believe that there are many problems in companies that look a lot like the problem Netflix faced — such as insurance underwriting, oil exploration, or evaluating an employee's probable success. As long as the problem has lots of structured data and clear outcome variables (and the company can successfully design a way to keep customer and proprietary data safe) it makes sense to design a competition that uses the market in ideas. Then you just have to decide how much it is worth to you to have a 10% advantage over the competition on a key issue.

If it is just too hard for your organization to create such a prize, there are ways to unlock internal talent on important opportunities. Internal competition is the lifeblood of improvement. When I was an MBA student, I remember reading about succession planning at GE. At the time, Reg Jones was the outgoing CEO; to find a successor, he split the company into three "sectors" — which had a hodge-podge of businesses in each of them — and Reg put a different management team in charge of each one. As a student I could not see any coherence in the products and services in each group until it finally dawned on me that Jones was creating three mini-GEs, each with a broad range of businesses, and he was having a horse race among the three management teams. The winner? Jack Welch and his team. And of course, Jack used a similar strategy upon his own exit.

Is there a vital part of your business that you can create an internal competition around? Do you have ten call centers, say, which could be portioned out to different management teams to find new ways of driving satisfaction and productivity? Could you modify your website for segments of incoming customers and let different management teams experiment with the customer experience? Do you have different territories in which different senior executives can experiment with new sales and service models? Set the goals and ground rules — remember that clarity was a key component of the Netflix Prize's success — and see who wins.

Often, the biggest challenge in mobilizing the talent inside your organization is learning to live with different solutions to the same customer problem. You need to ask yourself whether your current approach is designed to deliver what customers really want from you, or designed to keep employees comfortable. There's a lot of improvement to be had — if you can get over the fear of ruffling feathers.