Managers who wish to improve customers’ image of their companies can run communication campaigns, improve products, add personnel, and so forth. But how can managers decide which improvements in perceptions would be most beneficial and cost-effective? The authors present a method to (1) determine the company attributes that are relevant to customers; (2) rank the importance of those attributes; (3) estimate the costs of making improvements (or correcting customer perceptions); and (4) prioritize image goals so that the improvements in perceptions obtain the maximum benefit, in terms of customer value, for the resources spent.
1. J.C. Bevis, “Corporate Image Studies,” in Handbook of Marketing Research, ed. R. Ferber (New York: McGraw-Hill, 1974), pp. 206–218.
2. H. Barich and P. Kotler, “A Framework for Marketing Image Management,” Sloan Management Review, Winter 1991, pp. 94–104.
3. E.R. Gray and L.R. Smeltzer, “Corporate Image — An Integral Part of Strategy,” Sloan Management Review, Summer 1985, pp. 73–78.
4. P.E. Green and V. Srinivasan, “Conjoint Analysis in Consumer Research: Issues and Outlook,” Journal of Consumer Research 5 (1978): 103–123; and
P.E. Green and V. Srinivasan, “Conjoint Analysis in Marketing: New Developments with Implications for Research and Practice,” Journal of Marketing 54 (1990): 3–19.
7. D.R. Wittink and P. Cattin, “Commercial Use of Conjoint Analysis: An Update,” Journal of Marketing 53 (1989): 91–96.
8. Green and Srinivasan (1978).
9. The ideal-point model suggested in the conjoint analysis literature, which uses linear and quadratic terms, sometimes runs into the technical difficulty that, even though the observed part-worth values may be increasing over the levels, the fitted ideal-point model reaches a maximum value within the relevant range so that there is a decrease in utility for values beyond the maximum. This is illogical because, in our context, the attributes are such that higher values of the attribute imply greater preference. See:
Green and Srinivasan (1978) and
D. Pekelman and S.K. Sen, “Improving Prediction in Conjoint Measurement,” Journal of Marketing Research 16 (1979): 211–220.
10. In Figure 1, the shape parameter S is given by I (0.5)S = Q, so that S = [log(Q/I)]/[log 0.5].
11. For discussion of the relative merits of the conjoint and self-explicated methods, see:
Green and Srinivasan (1990).
12. Suppose the unit used in Step 5 above to obtain the resource requirement parameter R is 0.25. Then by the proportionality assumption made in Step 5, the resource required for a 0.05 improvement on attribute j is (0.05/0.25) Rj = 0.2 Rj. The constant term 0.2 does not affect the comparison across attributes and hence is not considered.
13. If the part-worth functions are concave (i.e., exhibit diminishing returns) as displayed in Figure 1, this step-by-step marginal analysis has the desirable mathematical property of producing undominated solutions. That is, at every step of the procedure, the customer value of the image is maximized subject to not exceeding the corresponding resource level. See:
B. Fox, “Discrete Optimization via Marginal Analysis,” Management Science 13 (1966): 210–216.
If the part-worth functions are not concave, the step-by-step marginal analysis needs to be replaced by dynamic programming. See:
R.E. Bellman and S.E. Dreyfus, Applied Dynamic Programming (Princeton, New Jersey: Princeton University Press, 1962).
14. P.E. Green, “On the Design of Choice Experiments Involving Multifactor Alternatives,” Journal of Consumer Research 1 (1974): 61–68.
15. V. Srinivasan and A.D. Shocker, “Estimating the Weights for Multiple Attributes in a Composite Criterion Using Pairwise Judgments,” Psychometrika 38 (1973): 473–493.
16. The resource requirement part-worth function would be scaled such that there are no incremental resources required to maintain the image at the current level on that attribute.
17. The step-by-step marginal analysis would produce undominated solutions as long as the customer attribute utility functions are concave and the resource requirement functions are convex. If the concavity-convexity assumptions are violated, then the step-by-step approach needs to be replaced by dynamic programming so as to maximize benefit for different resource levels. See:
Bellman and Dreyfus (1962).