After six years worth of idea submissions by EMC employees, the EMC Innovation Showcase has yielded a fairly robust set of data over which we can run analyses. In recent posts I have shared analytic discoveries about clusters of innovators, geographic boundary spanners, verbosity of idea submissions, and topic trends and analysis.
During this year, EMC Distinguished Engineer and Data Scientist John Cardente has taken an exhaustive look at a variety of parameters related to the submission of employee ideas. I have looked through his results and considered how to operationalize some of the insights that I've gained.
Here are some of the parameters that are important to analyze in regards to idea contests:
- Team Size: how many people are submitting an idea together?
- Length of problem description: how many words does it take to describe the problem?
- Length of idea: how many words does it take to describe the idea?
- How many people commented on the idea?
- What was the time span (in days) during which people commented?
- How many "sponsor tags" were on the idea (sponsors at EMC create specific challenges for employees to solve)?
- How long before the deadline was the idea submitted?
All of these parameters can be analyzed and mapped against "the results" of the idea. Was it promoted as a "good idea" (i.e. selected as a finalist)? Was it ultimately selected as a winner and resources applied to advancement?
John produced the kernel density plots depicted below:
Here are some of John's own conclusions (word for word):
- Good contest ideas are characterized by detailed problem descriptions.
- Good contest ideas attract a greater number of comments.
- Large teams aren't necessary to generate winning ideas.
- Early contest submissions appear to have a disadvantage. Perhaps this is because early submissions inspire others to submit higher quality ideas.
How can I operationalize these findings?
Firstly, I can market these results, especially on our idea portal (for example as part of the idea submission workflow).
Secondly, I can actually run analytics against individual submissions and make recommendations on the fly, e.g.: "Good ideas, on average, contain 500 words or more. Your submission of 200 words falls well below this length. You have 30 days until this contest closes, do you wish to add additional detail to your idea"?
My team is currently closing out the 2012 Showcase and ramping up the creation of new tools and processes for 2013. John's innovation analytics data has been (and will continue to be) tremendously helpful in this regard.
Steve
Twitter: @SteveTodd
Director, EMC Innovation Network
