ARTICLE
Usability Implications for 1to1 Marketing
Marc Resnick, resnick@eng.fiu.edu
Usability Solutions

What is 1to1 Marketing

Originally, 1to1 marketing was a term used to convey the idea that potential customers could be segmented so specifically that a different marketing message could be sent to each one based on his or her individual demographic and psychographic characteristics. The more data that was collected about each demographic and psychographic group and their purchasing and behavioral habits, the more specific segmentation strategies could become. Before the advent of the Internet, it was not feasible to achieve true 1to1 customization, but the idea was valid. Marketers segmented potential customers into the smallest segments possible given the costs and benefits associated with data manipulation and the potential value of each segment.

The main objective of 1to1 marketing is that customized ads are much more likely to elicit a response. Not only are inappropriate ads wasteful of resources, but they also can clutter and obscure the marketing message and even alienate potential customers, thus reducing the effectiveness of better ads later on. Customized ads can convey the impression that the company really knows the customer and has his or her needs in mind. Typical customization includes the choice of product (potential customers receive an ad for a product they actually want to buy) or the shopping process (potential customers can receive unique discounts, expedited shipping or other benefits according to their individual preferences). In order to determine what ads are most appropriate for each customer, detailed customer models need to be developed.

Customer models

The marketing literature includes descriptions of many general customer models that can be adapted to a company's individual marketing strategy. Cuthbert (2000) describes six basic models of e-shopper based on Internet experience, price sensitivity, and product type. For example, a "surgical" shopper knows precisely what he/she wants and doesn't want to waste time navigating a site or browsing among many products. This shopper needs an effective search function and any advertising must download quickly. Surgical shoppers are more likely to respond to reputation management systems that provide ratings of competing brands to enhance his or her ability to identify the ideal product for his/her needs.

Steiger (2000) differentiates between expert, educated, and novice customers based on their knowledge of the product domain. Experts are familiar with the characteristics of specific products. Educated customers are familiar only with the domain. Novices are familiar with neither. Novices will be more receptive to advertising of products within the domain and those from companies known from other domains. On the other hand, experts may only be receptive to functionally related and accessory products.

As marketing migrated on-line, companies began to conceive of true 1to1 customization. Data from site registration, clickstream histories, pop-up surveys, checkout forms, and other customer touchpoints could be aggregated across a wide variety of sites (ignoring for the moment customer concerns about privacy) to develop a very extensive database on each customer. Companies such as DoubleClick were created specifically to allow data to be aggregated from multiple vendors and shared among them. Bricks and clicks organizations could even integrate data collected in stores by sales clerks or purchase history into these databases. Data mining techniques could then be used to develop sophisticated customer models that could predict what products and offers each individual is most likely to respond to. Marketing messages based on these predictions could then be delivered through email, digital print mail, or even directly on the web site in real time.

Customized ads provide several benefits in addition to improved appropriateness. Customization decreases the effort to make a purchase. A link to the product or service can be placed directly into the ad (literally for email ads or a URL for print ads). For email ads, the customer's profile information can be prepopulated into any checkout forms by using individual URLs. Customization also reduces the total cost of marketing campaigns (Socketware, 2001). Ads can be sent only to those potential customers who are likely to respond.

A further benefit of the Internet channel to deliver ads is the ability to measure the success of marketing messages in real time (Socketware, 2001). Messages are coded so that the company can record whether recipients read the message, click on a link, browse the site, and/or purchase a product. They can also measure the size of the purchase and the profitability of the interaction. The response profile of each segment can be modeled and ads that are not eliciting desired targets can be redesigned immediately. Marketing campaigns can be distributed incrementally so that several improvement cycles can be accomplished before the majority of potential customers receive an ad. Jacob Nielsen provides a simple illustration of this process for advertising a conference in Nielsen (2001).

Context Marketing

In addition to demographic-based segmentation, ads can also be customized based on the potential customer's current task. Context marketing provides an additional and even more effective method through which to customize the marketing message. Customers' responsiveness to various types of advertising is often determined as much, if not more, by their current goals and objectives as by their age or income (Revenio, 2000). Combining customer models with task models increases the possible level of customization. For example, a shopper purchasing a book as a gift may be more receptive to an ad for wrapping paper than a related book. Other task-related characteristics can also improve ad customization. If the shopper is in a rush, he/she will respond differently to advertising as well as be receptive to different types of ads (Li, Kuo and Russell, 1999). Context marketing only works if the site can learn or predict characteristics of the task from the customer's clickstream behavior or by directly asking for the information.

Developing task models can be challenging. While demographic information can be collected over time, task information must be collected separately during each visit. Some customers may repeat certain tasks regularly, but in most cases, the task must be identified from behavior during that site visit. Besides simply asking shoppers about their current task when they arrive at a site, navigational behaviors can also be used to predict task characteristics.

Rozanski, Bollman, and Lipman (2001) identified several navigational behaviors that can be used to predict customer task requirements. They used factors such as the total time spent on-line, the average time per page, the percent of the total time spent within each product category, and the familiarity of that customer with the site to divide customers into seven context-based segments. Using these segments, they were able to identify when customers were most receptive to marketing messages and most likely to respond favorably. The importance of using context specific parameters was shown when 44% of their participants were observed to engage in all seven behaviors at some point during their study. Sending a message to a customer who is engaged in a task that makes him/her unreceptive can have negative consequences. Some of the segments are not conducive to any marketing efforts and customers should be left alone during these times. Using customer models alone can not differentiate among these context-based segments.

Implications for Usability

Marketing researchers have been developing statistical methods to build consumer models for many years (Churchill, 1999). Concurrently, Human Factors practitioners have been developing cognitive task analytical methods to model the behavior of system users (Schraagen, Chipman, and Shalin, 2000). Combining these two efforts can lead to superior models for consumers shopping through the Internet channel. Resnick (2001) suggested that when consumers are shopping on-line, much of their behavior can be described by Recognition-Primed Decision Making (RPD) models (Klein et al, 1993). Customer needs change significantly when shopping on-line. Issues such as trust (Egger, 2000), information privacy (Steinfeld and Whitten, 1999), and financial security (Rees, 2001) become more important. Issues such as service quality, which are addressed by marketing research, are also critical (Montania and Resnick, 2001a), but the manifestation of these concerns and the ways to alleviate them in the Internet channel are qualitatively different. The methods used by Human Factors practitioners to elicit user needs have several advantages over marketing focus groups. Focus groups are limited to needs that customers are consciously aware of and can articulate. They discover most of the functional needs, but are not as effective at discovering system performance needs. In order to develop truly customized 1to1 marketing campaigns, consumer models must be complete, including functional and performance needs. Using Human Factor's user needs elicitation techniques to develop 1to1 marketing consumer models leads to more comprehensive models than marketing techniques alone.

Testing is another area that 1to1 marketing efforts can benefit from Human Factors techniques. Iterative design and user testing is a strength of the Human Factors profession. The iterative nature of real time testing in 1to1 marketing campaigns requires sophisticated testing protocols. Human Factors is uniquely qualified to provide this expertise. Using context-sensitive empirical testing methods (Resnick, 2001b), campaigns can be modified to accommodate a wide range of customer needs. Testing can investigate factors such as balancing the frequency of ads so that they arrive as often as necessary to create brand image but not too often to overwhelm and alienate customers.

One area that can cause conflict between Human Factors and Marketing efforts is the permission strategy. Marketers generally prefer to add customers to the mailing list and allow them to opt-out if they do not wish to receive mailings. Similarly, they prefer to default the registration page check box to receive ads and require customers to uncheck it. On the other hand, research has shown that customers prefer that the default is not to receive ads and they will opt-in only when there is a high perceived value (Socketware, 2001). Developing a compromise strategy is needed to maximize the effectiveness of the campaign.

Design Considerations

The ads themselves would also benefit from a large dose of Human Factors. Many issues in the design of emailed ads can be improved through the application of Human Factors design concepts and testing. Design of the "From" and "Subject" lines to maximize recognition of the source and the permission is critical. Determining which pages to include direct links in the message can be based on the expected task flow of the customer's shopping behavior. Wording should be designed by a combination of marketing-oriented copy writers and usability-oriented technical writers. A similar partnership of marketing and usability should be used to select color schemes and graphics.

Summary

The design and development of 1to1 marketing strategies and campaigns has largely been left to marketing departments. However, as marketing migrates to the Internet and true 1to1 segmentation based on customer and context models becomes possible, the contributions of Human Factors cannot be underestimated. In order for companies to maximize the value they receive from their Internet marketing campaigns, it is critical for Human Factors practitioners to participate.

References

1. Churchill G.A. (1999). Marketing Research: Methodological Foundations, 7th Edition. Dryden Press: Orlando, FL.
2. Cuthbert M. (2000). The Six Basic Types of E-Shoppers. E-Commerce Times. 9/9/2000. http://www.ecommercetimes.com/perl/story/?id=4430.
3. Egger F.N. (2000). Towards a model of trust for e-commerce system design. In Proceedings of CHI 2000. Association of Computing Machinery.
4. Klein G.A., Orasanu J., Calderwood J. and MacGregor D. (1993) Decision Making in Action: Models and Methods. Ablex Publishing: Norwood NJ.
5. Li H., Kuo C., and Russell M.G. (1999). The impact of perceived channel utilities, shopping orientations, and demographics on the consumer's online buying behavior. Journal of Computer Mediated Communication, 5, 2, 1-20.
6. Montania R. and Resnick M.L. (2001). Promoting on-line customer confidence through page design. In Usability Evaluation and Interface Design Volume 1. M.J. Smith, G. Salvendy, D. Harris, and R.J. Koubek (eds). Lawrence Erlbaum Associates: Mahwah, NJ.
7. Nielsen J. (2001). Designing web ads using click-through data. Alertbox. September 2. Retrieved on 9/3/01 at www.useit.com/alertbox/20010902.html.
8. Rees J. (2001). Transaction security in B2C eCommerce: Perceptions and Reality. In Usability Evaluation and Interface Design Volume 1. M.J. Smith, G. Salvendy, D. Harris, and R.J. Koubek (eds). Lawrence Erlbaum Associates: Mahwah, NJ.
9. Resnick M.L. Recognition Primed Decision Making in E-commerce. Proceedings of the Human Factors and Ergonomics Society 45th Annual Conference, Human Factors and Ergonomics Society, 2001a.
10. Resnick M.L. Task Based Evaluation in Error Analysis and Accident Prevention Proceedings of the Human Factors and Ergonomics Society 45th Annual Conference, Human Factors and Ergonomics Society, 2001.
11. Revenio (2000). The Imperative for Dialog Marketing. White Paper. Revenio.com . Burlington, MA.
12. Rozanski H.D., Bollman G., and Lipman M. (2001). Seize the Occasion: Usage based segmentation for Internet marketers. White Paper. Booz Allen & Hamilton.
13. Schraagen J.M., Chipman S.F., and Shalin V.J. (2000). Cognitive Task Analysis. Lawrence Erlbaum & amp; Associates: Mahweh, NJ.
14. Socketware (2001). The e-mail marketing handbook. Accucast.com . Atlanta, GA
15. Steiger P. (2000). Needs-oriented e-commerce systems. In Proceedings of CHI 2000. Association of Computing Machinery.
16. Steinfeld C. and Whitten P. (1999). Community level socio-economic impacts of electronic commerce. Journal of Computer Mediated Communication, 5, 2, 1-15.