Tutor profile: Raj D.
Point of parity (POP) or point of difference (POD)--which is more critical for success in positioning a brand?
The above question often triggers a debate. Those who think POD (attributes that are unique to the product) is more critical tend to outnumber those who think POP (attributes required for category membership) is more so. However, I also note that people often underestimate the value of POP. Consider the simple case where POP and POD can have only two levels: absent (0) or present (1). The two levels give rise to four possible scenarios: (a) POP present and POD absent (b) POP absent and POD present (c) POP present and POD present and (d) POP absent and POD absent. Among these, (c) is the best and (d) is the worst. But the interesting question is whether (a) is better or worse than (b). Scenario (a) is better than scenario (b) because the presence of POP ensures category membership and thereby a position in the market. In the absence of POD, such a position is likely to be weaker than that of competitors with POD. However, scenario (b) disqualifies the product for category membership, and thereby, denies it a position in the market. Thus, POP is more critical than POD. POD is a prerequisite for successful differentiation, and differentiation is crucial for above-average earning. However, POD is worthless without the support of POP. Many new products (e.g., Iridium phone, Google glass) die because, though they score high on the difference (POD), they score low on the relevance (POP). On the other hand, POP can be worthwhile without POD. Me-too products survive. Moreover, we must appreciate the challenge involved in maintaining POP which, like POD, keeps evolving in the market. Sometimes established brands (e.g., Nokia, BlackBerry) die despite their POD because they cannot keep pace with the evolving POP in the market.
The owner of a new shopping mall wants to know if playing a certain kind of music there can increase sales. The idea is that pleasant music will make people stay and look around longer. And, over time, that will help raise sales. To test the notion or hypothesis, the manager, in an initial experiment, varied the number of minutes of music in the mall in a day and recorded the average minutes of customer visit on the same day. For 100 days, he varied the minutes of music between 0 to 720 (throughout the day) and recorded the data. Contrary to expectations, however, the correlation coefficient (Pearson's) for minutes of music and average minutes of customer visit was low at .32 and lacked significance at .05 alpha. Power of the sample was .81. Is the evidence against the manager's notion?
Probably the evidence is against the notion. Lack of correlation does tend to rule out causation. However, before concluding, we must examine a range of issues about the statistics and the design/conduct of the experiment/trial. A frequent issue in such puzzling situations is the level of data used in the correlation analysis. The test appears to have used aggregate level data, and so it might have failed to detect correlation at the subset level of data. For example, maybe in evening hours, music increases stay duration, but in morning hours, it does the opposite. Or it could be that increasing music duration from 0 to 360 minutes increases customer stay duration but increasing it from 360 to 720 minutes, does the opposite. The above scenarios are easy to detect in a scatter plot, and they call for testing at the subset level of data (e.g., morning and evening, 0-360 minutes and 360-720 minutes). Notably, they invalidate the reported correlation test result on the ground that the aggregate level data violates the linearity assumption in Pearson's correlation test. Sometimes, removing outliers from the data, if any, can reverse the test result. Also, we need to make sure that the distribution of the data is close enough to normal and homoscedastic. Pertinently, issues of internal validity in the design and conduct of the experiment from which the manager collected the data could also have confounded the results. We can say that the reported evidence is against the tested notion insofar as we can eliminate the presence of the above issues.
The golden rule of management is that there is no golden rule--how does this rule or guideline help us manage better?
I provide here two answers that differ in standpoint. First, the above golden rule suggests that the effectiveness of any rule in managing, no matter how useful in the past, depends on the situation or context in which we apply it. It is the situation which finally determines what is the best rule, if any, for successful resolution of the management problem. Of relevance, we must be ready to adapt prevailing rules to fit the situation better. Second, the science of management suggests that rule-based or heuristic approaches are often inferior to analytical or systematic approaches. Rule-based approaches use only part of the available information and therefore become increasingly sub-optimal as the situation increases in complexity. Thus the above guideline advises us to do away with rules for managing when the situation is complex, and the stake is high.
needs and Raj will reply soon.