Rikiya Takahashi

I appreciate the readers reading this column, which describes not academic research-level studies but rough ideas. I currently hope this thinking becomes the seed of my future research.

Though a bit apart from numerical algorithms, I am also interested in how the principal mental processes of humans affect the massive economic datasets such as point-of-sales (POS) data. For massive datasets, statisticians mainly focus on the profit maximizations using the statistically highly predictive models. Yet, we should also care why such profit maximizations are possible. Why do certain types of customers often and generously buy our high-margin products whose selling prices are high but whose cost prices are low? Why don't they choose the similar-spec products of competitors? It is risky for the firms simplifying the reasons of choices as the powers of brands. For example, if the reason is merely a lack of knowledge about competitors, the firm's current businesses are quite vulnerable, because customers will switch the brands as soon as they know the existences of cheap but similar-spec products of competitors. Or, there might be quite different reasons of choices that the firms have never been able to imagine. For specific individual customers, a minor function of the selected product can become a strong differentiator depending on the specific usages that marketers do not know. For the enterprise customers on B2B businesses, operational constraints will also affect their decisions of purchases. Anyway, customers choose products depending on their own values, and explaining what are their values is still a challenging task in every business.

Prior work shows that learning the human's heuristics to evaluate risks is beneficial for understanding the perceived values of individual customers. As firms can never completely understand each customer's value, the benefit of a purchase decision is fundamentally uncertain for each customer. It is impossible for any customer to preliminarily determine her/his own best-fit solution completely, and the customer satisfactions are created by ex-post assessments of the specific products. Hence the individual's risk evaluation criteria strongly affects their perceived values. In addition, the enterprise customers will probably be more sensitive to the risks of their purchase decisions than the household customers. Interestingly, psychological studies have been clarified that such risk evaluation criteria are affected by perceptional biases (e.g. prospect theory (Kahneman & Tversky, 1979)) and the behavioral economics based on such psychological modeling has been a recent hot topic both in businesses and academics. Since humans adopt the different risk evaluation criteria from those using the strict simulation results given by the econometricians, marketers and product developers can design high-margin products by both improving the qualities of the top-priority functions while giving up the qualities of less-important ones. Even if the firms concern the poor quality of a specific function in their products, some consumers might ignore such poor quality, if the prices are sufficiently economic or the qualities of the other functions are sufficiently high.

The risk-attitudes of the customers are also related to the attitudes to the trade-offs between losses and gains. For example, requiring high qualities in some functions of products will correspond to a risk-hedge behavior for avoiding the troubles caused by the breakdowns of the focused functions. In an opposite manner, compromising for the low qualities in other functions, which are compensated by the economic prices, is a risk-seeking behavior of customers. Some business persons, who are tired for lots of requests from customers, might claim that customers do not consider the trade-offs and they always require high-quality for every function. Yet, in such situations, since the extremely high-grade products become inevitably expensive, the customers are presented trade-offs between quality and price. If customers always bought the expensive but high-quality products, such business would be a priceless happiness for the company. Yet actually, customers who have requested the improvements of many functions often buy moderate-price and middle-quality products, or taking a simplest behavior "no purchase". Hence we can assume that customers implicitly evaluated the benefits for each alternative including high-quality, middle-quality, low-quality products and no-purchase decision, based on their trade-off evaluation criteria. Smart entrepreneurs well know the importance of understanding trade-offs and proactively present multiple alternatives with price-quality trade-offs. Under the product portfolios based on the price-quality trade-offs, customers are naturally segmented by their own preferences and every customer can be satisfied. About the trade-off evaluations, there are innovative prior work for clarifying how the perceptional biases affect the choice functions under the trade-offs (e.g. similarity effect (Tversky, 1972), attraction effect (Huber, Payne & Puto, 1982), and compromise effect(Simonson, 1989)).

In my thought, the key to truly reflecting the inhomogeneous individual values for the real marketing activities is incorporating the human's perceptual models for the statistical algorithms to handle massive data. Though the basic psychological studies can imply many qualitative marketing approaches (e.g. In prospect theory, the concave and the convex utility functions for gains and losses, respectively, suggest that firms should many times give the small amount of rewards such as mileage points, while stimulate the large amount of expenditures in one time by cross-selling. ), unfortunately these research outcomes cannot directly tell us the concrete decisions such as how firms numerically set each price of many products under the given time horizons. For such lots of product portfolios, the standard statistical models such as price-demand regression analysis are more practical than the psychologically theoretical approaches. Currently, there is no standardized approach to incorporate the perceptual models for the large-scale data analytics. In future, many firms will be able to yield long-term large profits by searching for the high gaps between consumers' perceived values and the actual costs in development and production by using massive economic data, as picking diamonds from a large amount of pebbles. How to practically realize such data mining of perceptual gaps is both my and many researchers' mission.

References

• Daniel Kahneman, Amos Tversky, "Prospect Theory: An Analysis of Decision under Risk" Econometrica, 47(2), pp. 263-292, 1979
• Amos Tversky, "Elimination by aspects: A theory of Choice", Psychological Review, 79, pp. 281-299, 1972
• Joel Huber, John. W. Payne, Christopher Puto, "Adding asymmetrically dominated alternatives: Violations of regularity and the similarity hypothesis", Journal of Consumer Research, 9, pp. 90-98, 1982
• Itamar Simonson, "Choice based on reasons: The case of attraction and compromise effects", Journal of Consumer Research, 16, pp. 158-174, 1989