Characterizing Opportunity Spaces in Innovation: Evidence from Large Samples of Ideas in Five Domains

Abstract In a typical innovation process, the innovator identifies or generates a set of raw opportunities or “ideas” before selecting one or more for further investment. This mode of innovation can be thought of as search within the opportunity space. In this paper we characterize some properties of innovation opportunity spaces as reflected by the set of ideas collected, and address three main questions: (1) When a large number of independent efforts to generate ideas are conducted in parallel, how likely are the resulting ideas to be redundant? (2) Do independent exploration efforts tend to result in clusters of similar opportunities or more uniformly distributed opportunities? (3) Are isolated opportunities more valuable than those that are located within clusters of similar opportunities? Our approach is to analyze the similarity of raw opportunities within five data sets comprised of between 232 and 311 ideas each. These data were gathered incidentally to our teaching university electives on innovation over a three-year period. We estimate similarity using a novel coding method performed by 230 independent judges. We estimate the quality of the opportunities based on 10-20 peer evaluations for each idea. We find that there is remarkably little redundancy in opportunities. For one broadly defined domain, 90 percent of ideas were unique, and for a very narrow domain, 68 percent were unique. We find that there is more redundancy as the innovation problem is defined more narrowly. We find that opportunities are located in clusters, apparently aligned with areas of high perceived user need. Finally, we find that opportunities that are similar to other opportunities are not necessarily of lower quality; for two of the five domains, we find that quality is increasing and then decreasing in the number of similar opportunities.