The levels of market segmentation vary from mass marketing to segment marketing, with an extreme limit of "segments of one" (Kotler &Keller, 2009, p.210).  The trend toward market segmentation is caused by the increases in the variety of advertising media and number of distribution channels.  These are making it harder to realize mass marketing.  So, to adapt, companies resort to "micro-marketing" (2009, p.208) through segments, niches, local areas, and individuals.  This post explores two methods of presenting choice: by-attribute and by-alternative. Which one is used has interesting implications for customer satisfaction and sales encouragement.

Customization of the product itself is the fifth component of one-to-one marketing ((Kotler &Keller, 2009, p.135).  The Journal of Marketing Research academic treatise reviewed in this post (Valenzuela, Dhar, & Zettelmeyer, 2009) deals with comparing two common methods of presenting choice in customer self-customization procedures.  The treatise touches on implications for determining customer preferences, increasing customer satisfaction and loyalty, and maximizing sales revenue.  Their premise, which is supported by four research experiments, is that the two most common methods result in different choices being made by customers.  Because the selections are influenced by presentation method, they also caution against using historical option selections as evidence of market preferences when developing new products or options.

The two methods compared are called by-attribute and by-alternative.  The by-attribute method requires the customer to select an option value for each of several feature attributes in isolation.  The by-alternative method requires the customer to select from a long list of alternative models that result from combining several combinations of feature options.  Where the by-attribute method is typified by Dell Computer, the by-alternative method is typified by Gateway.  Neerja Arora, from the A.C. Nielson Center for Marketing Research refers to the by-attribute method as "product customization" and the by-alternative method as "product proliferation" (Arora et al, 2008, p.313).  I prefer to call the later method "model proliferation" because, in my line of work, that is how the alternatives are presented.

The following hypotheses are asserted, supported by well designed experiments, and well presented discussion and results (characteristic of an effective marketing research process defined in Kotler and Keller (2009, p.91-105)):

H1 - There is a difference in the distribution of preferred options between by-attribute and by-alternative self-customization procedures.  The by-attribute method increases the preference for intermediate (middle position) value options, whereas the by-alternative method encourages extreme value selections.

H2 - The by-attribute method results in a lower degree of decision difficulty, although more questions are answered.

H3 - The by-attribute method results in a greater degree of satisfaction with the decision/choice.

H4 - Customers are more likely to defer the purchase decision when the by-alternative method is used.

H5 - Even if the by-attribute method is used, if there are non-explicit interactions and dependencies between attributes, there is less difference from the by-attribute method.

The first experiment, designed to ascertain H1, involved self-customizing a travel insurance policy.  It was presented as a three-option selection of three features by-attribute, and as a list of 26 different policies selected by-alternative.  A second experiment was designed to confirm the H1 results in an actual purchase situation where Hong Kong students were given HK$10 each and asked to select a self-customized pen that cost anywhere from HK$4 to HK$10.  The students were allowed to keep the money they did not "spend".

The third experiment was designed to confirm H2, H3, and H4.  Students at a US west coast university were asked to self-customize a laptop for their own use.  There were three features, each with three option values presented for the by-attribute method, and 26 different models presented for the by-alternative method.  In addition, the students were asked to gauge their choice difficulty, choice satisfaction, and the likelihood of purchasing the laptop if they were in the market for a laptop.

Finally, a forth experiment was designed to confirm H5.  In this experiment, the price was the same for all resultant products, but compromises had to be made between the option values of the three features (i.e. a DVD player could have high image quality or high audio quality, but not both).  The idea was to create a situation where there were interactions/dependencies between the features in the by-attribute method, to see if that would affect its advantages over the by-alternative method.

There are some major implications of H1 for marketers.  First of all "customer preferences are often constructed and not just revealed in the process of choice" (Valenzuela, Dhar, & Zettelmeyer, 2009, p.755).  This seems to contradict the utility of Kotler's and Keller's various different techniques of past-sales analysis as a part of estimating future demand (2009, p.116) for self-customized products.  Also, it seems to limit the usefulness of  today's prevalent personalized advertising (Arora et al, 2008) for self-customized products, since history of selections may not actually be a perfect indicator of market preference.  Finally, it has implications for pricing, as a company that uses the by-attribute method of self-customization should make sure margins are high in the intermediate options, and a company that uses the by-alternative method should make sure margins of models exhibiting extreme option values are high.

The implications of H2 and H5 are that making choices easier and less complex for customers is better achieved by reducing interactions between features than by reducing the number of features and options.

Valenzuela, Dhar, and Zettelmeyer assert that because by-attribute selections associate an option value with price, the customer doesn't perceive a selection as a loss or compromise of quality, only a compromise of price.  Whereas, when various option values are combined before price is presented, the various configurations/models each have a quality perception based on the compromise of option values selected.

Even though Forrester Research says 53% of people who put items in their online carts leave the site without buying (Prince, 2004), H4 says fewer purchases of by-alternative choices are purchased than by-attribute purchases for self-customized products.

This set of four experiments and the presentation of results was a significant body of work toward understanding why the by-attribute method of self-customization is preferred over the by-alternative method.  In the by-attribute method, the customer focuses on implementing choice selections rather than evaluating choice tradeoffs.  The most significant contribution is that it shows how and why the method of self-customization influences choices. Secondly, it separates the issues of complexity and conflict, the later being more detrimental to the self-customization experience for the customer, and ultimately for the marketer.


Arora, N. et al. (2008). Putting one-to-one marketing to work: Personalization, customization, and choice.  Market Letter, 2008, Vol 19, p.305-321.

Kotler, P. and Keller, K.L. (2009). Marketing management. Upper Saddle River, New Jersey: Pearson Prentice Hall.

Prince, M. (2004). Online retailers try to streamline checkout process. The Wall Street Journal (Eastern edition), November 11, 2004, D2.

Valenzuela, A., Dhar, R., and Zettelmeyer, F. (2009). Contingent response to self-customization procedures: Implications for decision satisfaction and choice. Journal of Marketing Research, Vol. 46, December, 2009, p. 754-763.