People Offer Better Ideas When They Can’t See What Others Suggest
27 July, 2015 / Articles
Companies from BMW to Kraft have invested a good deal in soliciting “open innovation” ideas from consumers, but the results have been underwhelming: Of the more than 23,000 ideas gathered by Dell’s Idea Storm site, only 2% have been put to use, and Starbucks has implemented an even smaller fraction of the 200,000 suggestions submitted to My Starbucks Idea (including 6,000 for new varieties of Frappuccino).
We think companies can do better. The key, we’ve discovered, is preventing participants in these forums from seeing the same body of suggestions as their neighbors. That’s because of the perverse dynamics of brainstorming.
If you gather consumers together to generate ideas, you’ll usually get more participation and more creativity than if you asked those same people individually to put ideas in a suggestion box. But these gatherings quickly find their creativity limited because of social convergence: The participants all see the same ideas, with the result that people pick up on each other’s suggestions rather than offer fresh insights. Other participants keep quiet, either because they worry that their ideas might be ridiculed or because a few talkative people dominate the conversations.
Fortunately for open innovation, the online environment makes it easy to overcome this problem.
The trick is to avoid clustering, where the same people have the same experiences. We want Mike and Beth to see and discuss each other’s ideas, and Beth to interact the same way with Dan, but we don’t want Dan to loop back into Mike’s ideas. No two people see the same batch of ideas, so each person retains an independence that helps to combat social convergence. These partial connections within the overall group of participants provide some diversity while diminishing the pressure for conformity. You still have plenty of interaction, but it’s distributed across the full group, which can be of any size.
To do this, you need anti-clustering software, which is not difficult to create; we were able to develop it ourselves as part of our research. The software makes sure that no participants share the same body of ideas and the same neighbors as anyone else. We applied it to an idea forum for a fictional company trying to develop a better mobile-banking application. We recruited participants and randomly divided them into clustered or nonclustered groups of 10 to 15 people each. To get the discussions going, all the participants had to offer an initial idea, which only their neighbors were allowed to see. We then asked everyone to offer additional ideas, which could be new or could be a more developed version of the first. The nonclustered groups’ suggestions were more creative and had higher market potential than the clustered groups’, according to a panel of independent judges. In other words, less interaction meant more innovation. (We describe our findings in detail in a forthcoming paper in the Journal of Marketing Research.)
In practice, this means that Dell would be better off imposing some structure on its Idea Storm sessions: Instead of making all ideas and comments available to everyone, Dell should limit the exposure. The company would get a boost to R&D that could eventually pay off handsomely.
Of course, Dell or any other company would need to be upfront about what it was doing or it would risk controversy (think of what happened to Facebook when it manipulated the newsfeeds of some participants). The company would do well to explain its thinking, emphasize the randomness of the assignments, and allow participants to see all submitted ideas at the end of each session.
There’s also no reason Dell couldn’t randomly reshuffle participants’ connections over time, as long as it continued to limit clustering. Friends who missed out on each other’s postings in early sessions could eventually expect to be connected.
We looked only at clustering, but software allows other kinds of adjustments. If a participant isn’t contributing, or is contributing too much, he or she could be reassigned to a different place in the forum. People could be reassigned over time by theme or product category — or jumbled together in a diverse mix if that yielded better results. Once participants were comfortable with some structure, companies could try out a variety of approaches in real time, just as Amazon relentlessly experiments with its user experience.
Experimentation is easy, so these structuring techniques could work in any kind of online forum, starting with general chat groups. All that’s needed is some kind of metric for assessing results, so that approaches can be compared. Who knows, maybe even technical-help forums could benefit from some structuring. And with companies relying increasingly on online forums for customer engagement, these software-imposed structures could eventually become a key tool in a social media strategy.
There are multiple reasons for setting up open innovation initiatives with consumers. Both the suggestions and consumers’ votes on them serve as feedback on existing offerings, and participation can bind people to the brand. Even the act of setting up a forum can boost the brand. We suspect that certain companies perceived as stodgy have sought online suggestions mainly for the cachet.
But if you’re seriously trying to use open innovation to boost R&D, imposing limits can make for better results, not just in product ideas but in better customer experiences overall.
