What’s the best way to brainstorm? It depends on what you want to get out of the process.
According to a new study, brainstorming sessions that generate one great idea vary structurally from sessions that produce several above average concepts. The coauthors of the paper, Karan Girotra, a professor at INSEAD, and Christian Terwiesh and Karl T. Ulrich, both professors at the University of Pennsylvania, note that while managers tend to focus on the latter stages of research and development, they are not as structured or rigorous about the beginning stages. And no matter how well-conducted the latter stages of the innovation process, they cannot elevate a fundamentally low-quality idea.
It’s natural to think that gathering ideas isn’t something that can be codified. After all, some of the most celebrated discoveries and inventions -- from penicillin to 3M’s Post-it Notes -- have sprung from serendipity. But, Girotra observes, in a business environment in which many product development techniques have been commoditized, or even outsourced, idea generation and its selection process may be some of the last sources of competitive advantage.
Studying brainstorming isn’t a new concept. But existing research often focuses on how many ideas groups come up with, as opposed to evaluating the merit of those ideas. Why is existing research fuzzy when it comes to evaluating the caliber of these ideas? Because it turns out that brainstorming groups “are very bad at evaluating ideas,” according to Terwiesch. “Certain members will get hung up on certain ideas, and often there is a strong personality whose opinion will dominate.”
To combat this dynamic, the authors, in their research, split people into two groups: those generating ideas and those assessing them. After a group came up with new product ideas, researchers asked as many as 20 outside experts to subjectively assess the concepts.
Two types of groups generated ideas. One followed a traditional model, assembling a group -- in this case, students studying product design -- and having them come up with appropriate product ideas for dorm rooms. They worked solely in a group. The other group took a hybrid approach: Those students worked on ideas by themselves before coming together to share their thinking.
Which technique yielded the best ideas? Strictly speaking, the traditional brainstorming groups came up with the very best ideas. They also came up with the very worst ones. In other words, their results’ quality varied much more than did the hybrid group’s results. The hybrid group produced more ideas that were, on average, of higher quality. But, as Girotra notes, “when it comes to innovation, the extremes are what matter -- not the norm and not the average.” So, if both groups work for the same amount of time, the traditional brainstorming team “significantly outperforms” the hybrid group when it comes to producing the best ideas, according to the authors.
This finding contradicts most existing literature on the subject, which tends to conclude that while working in teams is more satisfying, working alone generates the most effective ideas. But “what we found makes sense, since the most successful creative firms do mostly use team processes for brainstorming,” Terwiesch says. “We just brought some new thinking to the subject.”
Unfortunately, as noted above, brainstorming teams don’t always recognize the best ideas they’ve come up with. The ability to evaluate ideas well “is compromised in a group, where members are more likely to want to second the boss,” says Terwiesch. And it’s also important to note that the hybrid group came up with many more ideas than did the traditional group. This is logical: The traditional group, in which participants spent all their time together, had to spend more time listening to others as opposed to coming up with its own ideas.
This article is adapted from “Where the Best -- and Worst -- Ideas Come From,” by Josh Hyatt, which appeared in the Summer 2008 issue of MIT Sloan Management Review. The complete article is available at http://sloanreview.mit.edu/smr/.