As per instructions, I have administered a survey to a sample population that for several different "jobs-to-be-done" asks the survey participants to rate the importance of the "job-tobe-done" and the degree to which they are satisfied with current products/services they use today to help them get the job done. Both on a scale from 1 to 5.
Thus, the raw data looks like:
Name Profession O1_importance O1_satisfaction O2_importance O2_satisfaction etc
John Programmer 5 2 3 3
Mary Art Director 3 2 4 4
Sean Professor 2 3 2 2
etc
According to the article Turn Customer Input Into Innovation, HBR January 2002, there is a variable called the opportunity score, which is calculated as [importance + MAX(importance—satisfaction,0)]. The author defines "importance" and "satisfaction" as ten times the percentage of customers that rate their level for a particular "job-to-be-done" on the high end (4 or 5) of a 1-to-5 scale. Thus, if 75% of the customers deem it important and 40% state they are satisfied, the importance becomes 0.75*10 = 7.5, the satisfaction 0.4*10 = 4.0 and the opportunity is 7.5+(7.5-4.0) = 11.0
So far all good, opportunity scores can be given to the "jobs-to-be-done". However, the instructions on how to segment these survey respondents into opportunity-based segments seem nonsensical to me:
Use nonhierarchical clustering algorithms to execute the clustering process. The clustering algorithm should focus on the opportunity scores given to the "jobs-to-be-done" and place the respondents surveyed into a predetermined number of segments based on their responses.
To identify those "jobs-to-be-done" that would make the best segmentation segmentation variables, use factor analysis (a common statistical technique) to group like "jobs-to-be-done" together into a number of distinct opportunity-based factors
There are multiple ambiguities in the literature which makes it difficult to understand exactly how to perform this in practice. Several hours of Googling, reading up on statistical methods and re-reading the instructions trying out tons of stuff has yet to help make this clear. I find no existing examples of how survey respondent segmentation based on market opportunity score is done in practice.
