A recommended approach

Editor’s note: Richard Cohn is president of Cohn Research, Princeton , N.J. John Leggett is president of Slipstream Group LLC, a Dover, N.H., research firm.

Technological advancements in accessing customer transaction information and improved analytical processes are having profound effects on many business activities. Sales and marketing activities are not exempt from these advancements and may become significantly transformed by these changes.

These new opportunities to link customer information (e.g., tracking the process from the customer’s exposure to marketing communications to purchase) and to analyze it (e.g., mining customer data to identify patterns of behavior) are assumed to be good for marketers because it increases their understanding of the customer.

A customer-centric approach may be a true advancement in product development and service activities, but it has several shortcomings in sales and marketing. The first shortcoming is the limitation of knowing only what information is in the customer database. For example, if the purchase response to a direct mail or Web-based marketing communication is tracked, we only know which customers were exposed to these specific communications and their subsequent purchases. We do not typically know what other marketing communications from both our company and competitors had an impact on the purchase decision.

The second shortcoming of a customer-centric approach is that marketing insights come from understanding not only the customer but also the non-customer. After all, it is the non-customer or the potential customer that should be the true target of the marketer’s interest. Understanding the differences between the company’s customer and its non-customer is often the key to sales growth.

The marketer’s goal is to understand the market and the forces that shape it. The customer-centric approach may keep the marketer from seeing the forest because of the trees. This article argues for analytical approaches that are more comprehensive than a customer-centric approach. Two analytical approaches are described. One is to understand the impact of marketing communications on brand purchases1. The other is to understand the basis of brand choice. These analytical approaches provide more complete and more actionable understanding of the sales and marketing environment than could be gained from even the most sophisticated mining of customer data.

These two approaches are not new. They were developed and used in large-scale proprietary research programs during the past 12 years. The understanding gained from these research programs was used almost exclusively for high-level marketing decisions (e.g., assessments of new product introductions, advertising campaigns, and new marketing channels). In the past, the research was not used to increase rank-and-file sales and marketing staffs’ understanding of the market. In the final section of the article we argue that this use, keeping the staffs enlightened and up to date, has even greater value than for major single event decision-making.

Understanding marketing communications and their impact

Four approaches

The need to understand the impact of all marketing communications to accurately assess the impact of any marketing communication is illustrated with the following scenario:

On Monday, Nick is watching TV and sees a commercial for Brand A. The commercial is effective in stimulating Nick’s interest in buying the product and he is impressed with Brand A’s attributes. On Tuesday, Nick visits Brand A’s Web site to learn more about the product. He also reads some product reviews from a neutral journalistic source (e.g., Consumer Reports). On Wednesday, Nick gets a mailing that promotes Brand A and he talks to his friend who recommends the brand. On Thursday, Nick goes to the store and discusses the product with a saleswoman. She convinces Nick that Brand A does have great attributes and that he can get Brand B with the same attributes for less money. Nick purchases Brand B.

We could call this scenario a “linear chain of marketing communication exposures” but the more common term “shopping” will do fine. Nick’s shopping experience, a series of exposures to marketing communications, is summarized in Figure 1.

Consider four analytical approaches for understanding the impact of marketing communications on Nick’s behavior. We will reject the first three approaches, but they are worth reviewing because they illustrate problems with past and current approaches.

Approach I is the “old old-fashioned market research approach.” Exposure to marketing communications is measured independently of sales (e.g., Nielsen ratings of the Monday television program that carried the advertising, number of hits on Brand A’s Web site, number of pieces in the mail drop). There are only aggregate counts of exposure to specific marketing communications and sales and no linking of the two at the level of the individual consumer. We just know how many people watched television, visited the Web site, received the direct mail, and bought Brand A. There could have been high levels of advertising exposure, Web site visits and direct mailings. However, the marketing staff would incorrectly conclude that the communications were ineffective (i.e., did not receive a positive reaction and did not result in Brand A preference) because by the end of the week all the Nicks in the market ended up buying Brand B. This “old old-fashioned market research approach” is undoubtedly still the most common practice.

Approach II is the “new old-fashioned market research approach.” This approach uses traditional telephone interviewing tracking surveys to measure exposure to a single marketing communication type (typically television or print advertising) and brand purchase intention. The purpose of this approach is to assess the single marketing communication type (e.g., a television commercial advertising campaign). If Nick is interviewed on Tuesday or Wednesday and asked whether he saw the Brand A television commercial and his brand preference, the survey results will be an accurate assessment of the advertising. If Nick is interviewed on Thursday or any day thereafter, his answers will mislead Brand A marketers about the effectiveness of the advertising. Nick will say he saw the Brand A commercial and he prefers (actually bought) Brand B.

Approach III is the “new online research approach.” This approach does not survey Nick. It only identifies Nick as a visitor to the Brand A’s Web site on Tuesday. Since there is no subsequent Brand A purchase (which would add Nick to the Brand A customer database and perhaps link the purchase to his previous visit while shopping) there is not much information gained from this approach. At worst, marketers might incorrectly assume that the Web site information is not effective because it did not result in a sale to Nick when, in fact, he had both a positive reaction to the Web site information and it strengthened his preference for Brand A.

What would the “new online research approach” tell the marketer if the result of Nick’s shopping was different? What if he did not speak with the saleswoman on Thursday and he bought Brand A? This would result in a customer record with an indicator that he had visited the Web site before the purchase. Would the marketer be correct in attributing his purchase to the Tuesday Web site visit and in concluding that the Web site had a powerful impact that caused him to purchase Brand A? The Tuesday visit may have been a cause but was it was not the proximate cause (Nick’s Wednesday mail and his friend’s recommendation came later). The Web site visit was (perhaps) not even a direct cause (his visit to Brand A’s Web site caused him to go to neutral journalistic source for product reviews that were more convincing). This “new online research approach” not only can make the marketer not see the forest for the trees, it could make him build marketing programs with the wrong lumber.

Our recommended approach, Approach IV, the “comprehensive marketing communication model approach,” uses telephone interviewing tracking surveys of persons in the product market (both Brand A customers and non-customers).

The approach is based on simple premises about the correct way to determine the impact of a marketing communication on brand purchase intent.

  • To assess the impact of a single marketing communication, the relationship between exposure/reaction and brand purchase intention must be analyzed in the real world content of multiple influences2.
  • These multiple influences include a comprehensive set of the brand’s own marketing communications, competitors’ communications, non-advertising communications (e.g., press reports, product reviews, word-of-mouth commentary), and current brand usage experience.

The simple relationship between exposure to a marketing communication and purchase intention (or actual behavior), either at an aggregate (Approach I) or individual customer level (Approaches II and III) is inconclusive and can often lead to incorrect conclusions about the impact of the communication and to bad marketing decisions. Other marketing communications (from the company or its competitors) can overpower the positive effect of a good communication or mask the negative impact of a bad communication.

As indicated in Figure 2, these different marketing communication types can influence different ratings of the brand’s attributes. For example, the television advertising may have a positive effect on Nick’s perception of Brand A’s prestige, while the Web site information and product reviews may affect his perceptions of the brand’s availability and reliability. The salesperson contact may affect his perception of the cost or relative value of the brand. The comprehensive analytical approach uses a multivariate technique to analyze the relationships between the brand’s specific attribute ratings and the overall brand rating. Comparing this rating with other brands’ ratings determines brand purchase intent.

The model also includes current usage as a determinant of both brand attribute ratings and brand purchase intent. Personal experience or lack of experience with the brand has an obvious and important effect on brand evaluation and brand preference that needs to be taken into account in any assessment of the impact of marketing communications3.

Examples of the comprehensive approach’s value

Nick’s shopping scenario is loosely based on an important marketing event that was analyzed using this comprehensive approach. The introduction of the personal computer for home use was a marketing challenge for the OEMs. While most of the manufacturers were experts in industrial (i.e., B2B) marketing, selling the product to residential end-users (i.e., B2C marketing) was definitely a new experience in the early 1990s that called for new products, new communications, and new distribution channels. With everything changing at the same time, it was important to be able to interpret the success or failure of each change from the industrial marketing standard so that appropriate and rapid adjustments could be made.

One of the major OEMs had been using the comprehensive approach for several years in its industrial marketing programs so the application to the new home market went smoothly. The core interview for the tracking surveys was adapted to the home purchase decision-maker. In those pre-Internet days, the communications were primarily broadcast advertising and heavy doses of product reviews and word-of-mouth commentary. Current brand usage (based on office-based experience) also played an important part in determining brand purchase intention.

The initial results indicated a positive, although not blockbuster, reaction to the advertising communications and the product reviews. Yet brand purchase intention and actual sales were not reflecting these positive marketing program results. The comprehensive approach revealed the culprit. The brand’s distribution channels had for the first time included independent retail outlets. The research indicated that salesperson contact was having a negative effect on ratings of the brand’s cost (relative value) attribute. Like Nick, potential customers were being drawn into the retail outlets with a predisposition to buy Brand A and then convinced by the retail outlet’s sales force that Brand B was just as good and less expensive than Brand A. This insight lead to new communications with a value focus and a change in the relationship between the manufacturer and the independent retail outlets. The other three analytical approaches could not have provided this insight.

An even earlier anecdote illustrates the importance of measuring the impact of competitors’ marketing communications. An early version of the comprehensive approach was used in the 1980s to assess the impact of a home stereo equipment national advertising program. The advertising focused on several attributes of a good home stereo system but made little mention of the system’s sound quality. Purchase interest in the brand did not improve with the new advertising and the sound quality ratings of the brand began to decline.

The culprit was identified as the major competitor’s soaring rating on sound quality. The high levels of exposure and positive reaction to the competitor’s advertising that focused on sound quality overwhelmed the other brand’s communications. Copy research on both brands’ campaigns indicated that each brand’s advertising was effective in communicating the intended message and stimulating purchase interest. However in the real world of multiple influences, the competitor’s communications were having the greater impact.

The comprehensive approach gave the marketer a true understanding of his market and the forces that shaped it. He did not incorrectly assume that his advertising was a bad execution. He understood it was competitive forces that were driving brand purchase intent. This understanding gave him a range of actions other than the incorrect one of firing the ad agency.

The comprehensive approach to measuring the impact of communications is really a tool bag that marketers can draw from when they need to diagnose or fix a problem. Since the information is collected on an ongoing basis through tracking surveys, the tool bag is readily available. It allows the marketer or his colleagues in advertising, sales management or public relations to assess the impact of current communications and decide what, if anything, needs to be done. The information has been valuable to public relations staffs in the face of negative press reports. The tracking nature of the approach provides trend analysis with a built-in pre-event and post event comparison.

The comprehensive approach to measuring the impact of marketing communications has important advantages over the other approaches in both its accuracy and range of uses throughout the company. It also produces a by-product that several companies find equally valuable: a simple but powerful way to understand brand choice.

Understanding brand choice

Understanding why Nick bought Brand B is not as simple as concluding that the saleswoman’s comparison of Brand A and Brand B was the last (most proximate) communication he heard. The sequence in which Nick received communications about the brands may not be that important.

After Nick heard the saleswoman and all the other communications he made a choice between Brand A and Brand B. It is obviously important to the marketer to understand that choice.

The approaches to understanding brand choice are even more varied than the approaches to understanding the impact of marketing communications. They range from interrogating consumers right after the purchase to more formal qualitative (focus group) and quantitative (choice modeling) approaches. Each of these can be highly valuable to the marketer depending on his current state of knowledge and specific interests.

The recommended approach, choice grid analysis, is useful because of its intuitive simplicity and comprehensiveness. The graphic form of the analysis is easy to understand.

Three basic types of information are required for choice grid analysis.

  • The consumer’s ratings of the importance of specific attributes in choosing a brand.
  • The consumer’s first and second brand choices.
  • The consumer’s ratings of his first and second brand choices on specific attributes.

A choice grid uses the information on the importance of specific attributes (e.g., cost, prestige, reliability, availability, and service) and the ratings of first and second choices (Brands A and B) on the specific attributes.

For example, choice grid one (Figure 3) displays the importance of specific attributes on the horizontal axis and the difference in attribute-specific ratings of Brand A and Brand B on the vertical axis.

(The horizontal axis scale, labeled attribute importance, is the percent of consumers that say the attribute is “very important” in their decision of which brand to purchase. The vertical axis, labeled brand differentiation, is the difference in Brand A’s and Brand B’s ratings on the attribute. The difference in rating is scaled in terms of the test statistic value from a paired t-test (each survey respondent has rated both brands on each attribute). In this simplified example, Brand A is considered better than Brand B on those attributes towards the top of the chart, equal to Brand B on those attributes towards the middle of the chart, and inferior to Brand B on those attributes towards the bottom of the chart.)

Choice grid one is based on interviews with consumers who prefer Brand A to Brand B. Among these consumers, two attributes (cost and service) stand out as being very important in their brand choice. They think that Brand A is better than Brand B on one of these important attributes (cost) and that the two brands have roughly equivalent service. Note that those who prefer Brand A think that the brand has more prestige, but that this is a relatively unimportant attribute to them.

While this choice grid tells us what determines preference for Brand A, there is an equally important choice grid that indicates what drives preference for Brand B.

Choice grid two (Figure 4) is based on interviews with consumers, like Nick, who prefer Brand B to Brand A. In this simple example, they are similar to consumers who prefer Brand A in the importance of most brand attributes. The big difference between those who prefer Brand A and those who prefer Brand B is their perceptions of the relative cost of the brands. Those who prefer Brand B also think that the brands are not equivalent in service quality.

(Differences in the importance of brand attributes between consumers who prefer different brands are uncommon if the product has a commodity status. Some marketers are successful in convincing consumers that their brand not only possesses more of a particular attribute than does competitive brands but also that this attribute should be given more weight in their brand choice decision. In this example, those who prefer Brand B to Brand A are slightly more price sensitive and concerned about service than are those consumers who prefer Brand A.)

A complete understanding of what determines choice is gained by comparing the two choice grids. The marketer examines differences between the choice grid of those who prefer Brand A and the choice grid of those who prefer Brand B. Marketing communications should focus on these attributes. In this example, the perceptions of the cost advantage and the service advantage of Brand B are critical. Note that it would do little good to focus communications on Brand A’s advantage of prestige because both those who prefer Brand A and those who do not already recognize its superiority.

Through a series of choice grids that compare Brand A with different competitive brands, the marketer can develop a comprehensive understanding of brand choice and provide the company’s sales and marketing staffs with the information they need to deliver effective sales presentations and other forms of marketing communications.

An example of the value of a choice grid analysis

A dramatic example of the value of choice grid analysis involved both marketing and product development decisions. The leading manufacturer in an industry was preparing to introduce a new version of its product to attract the high end of the market. Somewhat coincidentally the company had started the comprehensive approach to studying the impact of its marketing communications and, as a by-product, was collecting the information for choice grid analysis.

Although the marketers had a gut feel for the strengths and weaknesses of the brand, they were surprised by the choice grid analysis. They knew their strengths but were unaware of the importance of their weaknesses in determining preference for rival brands.

A key weakness was the attribute of quality of workmanship. Testing of the soon-to-be-introduced high-end product indicated that it might fail at a rate that would reinforce the perception of poor workmanship. This is an especially significant problem for a flagship product.

The choice grid analysis caused the company to delay the product introduction until it and the word-of-mouth commentary it would generate would improve the perception of workmanship quality. Comparisons of choice grids over the next few years that showed a steady improvement in perceived quality of workmanship supported the wisdom of the company’s decision.

Using this understanding to drive sales and marketing programs

The use of the comprehensive approach to understand the impact of marketing communications and choice grid analysis to understand brand choice has proved valuable in executive level decision-making. Marketing communications have been altered, distribution channels changed and product development enhanced by the improved understanding of the market and the forces that shape it.

It is very likely that this improved understanding did not reach the lower levels of sales and marketing staffs, including sales and service representatives who have face-to-face contact with current and potential customers. It is these staff members who need the most actionable and up-to-date information on the market and their brand’s position in it.

The dissemination of the information from these approaches to a wider audience of sales and marketing staff is the next important value enhancement. Regular communication of the information through easy-to-access and flexible channels such as online communications can extend the use of the information from major strategic decisions to lower-level staffs’ tactical decisions that, when added up, determine the success of many companies.

While there may be concerns with the wider dissemination throughout the organization, such as the security and possible misinterpretation of the information, these are not insurmountable barriers to expanded use. There is a powerful competitive advantage to having all the company’s marketing and sales forces know why customers prefer their brand and why non-customers prefer another brand. 

 Notes

1Marketing communications should be broadly construed to include all relevant types that can affect brand choice. These include traditional media advertising, new media (e.g., a brand’s Web site), contact with salespersons, and communications that originate from sources other than the brand (e.g., press reports, journalistic product reviews, word-of-mouth commentary).

2 The measurement of reaction to each type of communication is as important as measuring mere exposure. Many types of communications, even advocacy types such as advertising, direct mail and salesperson contact, produce a neutral or negative reaction. It is critical to know the percentages of the market that react in different ways to each communication and to determine how many of each of these exposure/reaction subgroups intend to purchase the brand.

3 The current usage determinant (i.e., whether the person is a Brand A customer or non-customer) often leads to a more complex analysis than is indicated by the model. In many applications of the model, levels of exposure and reaction to the individual marketing communications and their relationships to individual brand attributes differ significantly between current customers and non-customers (i.e., there are interactions between the variables in the model). When this is evident, the relationships in the model are analyzed separately for subgroups of customers and non-customers.