April 1985, the management of Coca-Cola Co. announced its decision to change the flavor of the cotnpany’s flagship brand. The events that followed from this decision, as well as the faetors which led up to it, have been reviewed, discussed, and extensively analyzed in the popular press, the trade press, and in marketing textbooks. Two books and at least two marketing cases have been written on the events surrounding the flavor change decision.
Also, a well-known, but somewhat older Harvard Business School marketing case deals with some of the key events which led up to the decision. Despite the extent of this attention, more can be learned from this dramatic pieee of marketing history. Pepsi began communicating these findings to consumers through “Pepsi Challenge” television ads . showing taste tests where Coca-Cola drinkers expressed preferences for a cola which was then revealed to be Pepsi, This campaign contributed to Coca-Cola’s slow, but steady decline of market share in the soft-drink category.
This erosion was most apparent in foodstore sales, which reflect consumer preferences more directly than do vendingtiiachine or fountain sales. By 1977, Pepsi had actually pulled ahead of Coke in foodstore market share. Although publicly expressing a lack of concern about the Pepsi Challenge advertising, Coca-Cola’s managetnent privately was quite worried because blind taste tests by the company’s own market research department had confirmed Pepsi’s claims. Secretly, Coke’s management began researching the possibility of reformulating Coca-Cola to respond to the apparent changes that had occurred in consumer tastes.
By 1984, researchers had arrived at a new formula for Coke whieh, in blind taste tests, beat Pepsi by as much as six to eight percentage points. In addition to beating Pepsi, cola drinkers chose this new formula over the old Coke formula by 55% to 45% in blind taste tests and loyal Coke drinkers chose it over the old Coke formula by 53% to 47%, In taste tests where the drinks were identified as “new Coke” and “old Coke,” cola drinkers preferred the new formula over the old formula hy 61 % to 39%, Well aware of the importance of the reformulation decision.
Coke’s management made sure that the taste test results were checked and coiToborated in every major market in the country. Overall, Coke’s market researeh on the reformulation was one of the most exhaustive market research projects in history; It cost $4 million and included interviews with almost 200,000 consumers. After the decision to reformulate Coke was made, CocaCola chairman, Roberto Goizueta, termed the decision, “one of the easiest we have ever made,” according to Hale N, Tongren in his book. Cases in Consumer Behavior.
On April 23, 1985, Coke announced the reformulation with a grand flourish, slaging a multicity satellite press conference in New York, Atlanta, Chicago, Houston, Los Angeles, and Toronto. The next day, a front-page article in The New York Times reported: “The Coca-Cola Company said yesterday that il had scrapped the formula for the world’s best-selling soft drink. The recipe, concocted 99 years ago, has been piaeed in the vault at the Trust Company of Georgia Bank, never to be used again, said Roberto C. Goizueta, chairman of Coca-Cola.
“We have a new formula for Coke,” he added. ” In addition to the extensive publicity. Coke announced that the new Coke would come in a new ean, with updated red and The embarrassing failure of Coca-Cola’s attempt to change the flavor of itsflagshipbrand has hecome a textbook case of how ma ket research can fail. The lesson usually drawn is that Coke’s researchers asked respondents the wrong questions.
However, a careful examination of the events surrounding the reformulation attempt suggests an alternative explanation: that the error res ed from the standard market research practice of considering focus groups to be only a form of preliminary research and not app ciating their unique ability to predict the effects of social influence. The Attempt to Reformulate Coca-Cola T 22 he 87-year old rivalry between CocaCola, the traditional market leader, and Pepsi Cola, the perennial runner up, took an unexpected turn in the mid1970s.
Pepsi’s consumer research discovered in blind taste tests that a majority of consumers preferred the taste of Pepsi over thai of Coke. In fact, even a majority of loyal Coke drinkers reported preferring Pepsi in the tests. Decetnber1992 silver graphics replacing the traditional red and white look. Clearly Coke had decided to make sure that consumers would be aware that Coca-Cola’s flavor was being changed. The initial reactions of most consumers appeared to be positive.
Many bottlers reported that sales of new Coke were greater than expected and. uring the first few weeks after the new Coke introduction, the company’s weekly survey of 900 respondents showed consumers preferring new Coke over old Coke by a margin of 53% to 47%. However, during this period, there was also intense media coverage of those consumers who did not like the new Coke and were angry about the change. In a number of cities, old Coke loyalists sponsored protest rallies and boycotts and received widespread media attention. By June, it was becoming apparent to Coke’s management that consumer dissatisfaction with the reformulation was increasing.
The stream of angry letters and phone calls was becoming a flood, and weekly tracking surveys confirmed that consumers were becoming increasingly negative about the change. In a survey conducted during the first week of July only 30% of consumers interviewed reported preferring the new Coke to the old. On July 10. the company announced its decision to respond to public pressure and bring back the old Coke formula. It would be available in the form of a product with the name “Coca-Cola Classic,” and was intended as a flanker brand to satisfy those consumers who wanted the original taste as an option.
The reformulated soft drink was to be known simply as “Coke” and would remain the flagship brand. However, sales of new Coke eroded rapidly. In August, sales of Classic began to exceed those of new Coke, and by the end of September, Classic had a 70% share of the combined volume of the two products. Over the next few months, large fountain accounts, such as McDonald’s, began switching back to the old formula. In 1986. Classic outsold new Coke by more than 8 to 1. despite the promotion of new Coke with over $48 million of toprated television advertising.
Although Coke’s advantage over Pepsi in the sugared cola category had decreased slightly compared to 1984. Coke’s advantage over Pepsi in the overall market had increased, mostly due to the continued success of Diet Coke and tbe introduction of Cherry Coke, In response to questions about which was now the flagship brand, Goizueta explained that the company had adopted a “megabrand strategy” in which some promotion expenditures were directed at enhancing the image of the company’s entire line of cola drinks.
During the following years, the market share of new Coke continued to decline. By 1990. new Coke’s share had dropped to . 6%, In April 1990, the company began test marketing new Coke under the name ‘Coke II. ” It was advertised as having “real cola taste” with “the sweetness of Pepsi,” according to the Wall Street Journal. Although Coke’s management has claimed that these tests were successful, they appeared to have developed a considerable degree of caution on the question of supporting Coke T with a national rollout.
T marketers have used these events to disparage the discipline of , marketing research I The Wrong-Question Explanation T back the old formula. Coke’s president. Donald Keough told Time magazine, “… you cannot measure it any more than you can measure love, pride, or patriotism,” However, most observers did not attribute the failure of Coke’s research in this instance to an intrinsic limitation of the capabilities of marketing research. Rather, they judged that the research was conducted or interpreted incorrectly.
Although some have argued that Coke’s research error was to overgeneralize from inexact taste test results, the vast majority of people who have publicly voiced an opinion concerning where Coke’s research efforts went wrong espouse what could be called the “wrong-question explanation. ” In this view, the reason that Coke’s marketing research did not detect the consumer outcry which resulted from the reformulation was that they did not make it clear to the taste-test respondents that if most people chose the new Coke flavor, then the traditional Coke flavor would no longer be available.
In other words, rather than ask. “Which flavor do you like better,” consumers should have been asked a more relevant question. This question might have been something like. “How would you feel if we discarded CocaCola’s current taste and replaced it with this new taste? ” In the years since the introduction of new Coke and the reintroduction of CocaCola Classic, the wrong-question explanation is the one that has been most often repeated. It appears to have become the standard explanation for this highly publicized failure of market research. e widespread awareness of these events has led many observers to search for an explanation.
The particular question which most frequently has arisen is why Coke’s extensive market research was unable to provide management with better guidance in the reformulation decision. Anyone who is or has been a marketing research practitioner will not find it surprising that some marketers have used these events to disparage the discipline of marketing research.
For example, in Marketing News, one ad-agency manager attributed Coke’s mistake to “executives who solely rely on statistics. When announcing the re introduction of old Coke in July 1985, the top people at Coke suggested that research is not capable of measuring the types of consumer feelings that resulted from the attempted reformulation. When describing the emotional outpouring which led them to bring An Alternative Explanation S ince the intense publicity has died down, some further details of the research behind the new Coke decision have come to light. In particular, it is now known that Coca-Cola’s market research department did indeed ask the right question.
Considerable attention was devoted to testing consumer reactions to the idea of changing Coke’s flavor. These tests and their results are described in Thomas Oliver’s book. The Real Coke. The Real Story: “In 1982 the Coca-Cola Company conducted 2,000 interviews in 10 major markets to investigate further the public’s willingness to accept a different Coke, “Consumers were shown storyboards, comic-striplike mock commercials.
One board said Coke had added a new ingredient and it tasted smoother, while another December 1992 2 3 said the same about Pepsi. To compensate for the impact of first impressions, 1,000 people were shown the Coke statement first and the other 1,000 saw the Pepsi announcement first. )
Then Coca-Cola asked consumers a long series of questions about what their reactions to such a change would be. Would you be upset? Would you try the new drink? Would you switch brands immediately? “‘We estimated from the response that 10%-12%i of exclusive Coke drinkers would be upset, and that half of those would get over it, but half wouldn’t,’ said |Roy] Stout [director of marketing research I. … While the interviews pointed to people’s willingness to try a new Coke, Zyman [vice-president of marketing] and Stout discovered through other tests that many people just didn’t believe anyone could or should tamper with the king of the colas.
To hear debate on the issue, Coca-Cola’s research department used focus groups, a favorite marketing tool. “… Some of Stout’s focus groups were shown a storyboard depleting a proposed commercial. One said Brand X soft drink was going to be improved. ‘Fine. the group replied, and they were equally sanguine about a proposed improvement in Pepsi, But when it came to changing Coke for the better, the resounding response was NO. ‘It was like saying you were going to make the flag prettier. ‘ said Zyman.
A similar response came in 1983 from a group that included some exclusive Coke drinkers. This group agreed that Anheuser-Busch could change Budweiser but in no way should the Coca-Cola Company try to improve Coke. ” Oliver goes on to describe how these results led Coke managers to continue to support the work of their technical people to come up with a new flavor for Coke that would beat Pepsi.
By September 1984. product research had developed sueh a flavor, and taste-testing market research began moving into high gear. What Oliver describes sounds very much like what would be considered standard market researeh proeedure for the development of a new product or the modification of an existing one. The ideal would be to begin with focus-group testing of the product concept. Then, a survey would be conducted, using individual interviews with a large representative sample of consumers, to verify and quantify the results of the focus groups.
If the new product survived these concept tests, then the testing of an actual product (or prototype) would begin. 24 December 1992 Coke’s only deviation from this standard sequence is that the quantitative survey of individuals appears to have been done before rather than after the focus groups. But this is a minor point. What we see here are very normal market research procedures. Coke’s researchers started out asking the right questions, and in the right way. The results of the foeus-group phase and the survey conflicted.
The researchers trusted the survey, whieh comprised a large number of interviews spread over a wide, and presumably representative, area. – he focus group is a unique source of information about how the consumer will respond in a situation where there will be an awareness of the views of other consumers A wave of focus groups usually involves less than 200 respondents and is conducted with minimal attention to obtaining a representative sample.
It is standard market research practice to trust survey research over focus groups, and this praetiee appears, even to the nonresearcher (i. e.. he executive decisionmaker), to make good sense. Moreover, Coke’s research did a pretty good job of predicting consumer response, at least initially. Recall that when the reformulation was first introduced, the consumer response was favorable. But by the end of May 1985, it had begun to change. It was this that Coke had not expected. The New York Times’ report on Coke’s announcement of the reintroduction of old Coke began as follows: “When the CocaCola Company introduced a reformulated version of the world’s best-selling soft drink on April 23, it was well aware that it might alienate some faithful Coke drinkers.
The company, however, expected that alienation to fade. It was completely unprepared for how it would spread and deepen in the two months following the debut of the new Coke. ” It is this change in eonsumer opinion, and only this change, that Coke’s market research had failed to predict. With the benefit of hindsight, the locus of their research mistake becomes clear. It was to respond to the conflict between the Marketing Research results of the focus groups and the survey of individuals by trusting only the survey.
As it turned out. ne can see that both procedures had provided important information. When new Coke was first introduced. people made individual decisions on it. and most at least acquieseed to the change. This is what was predicted by the individual interviews which indicated that only {)%-2% of consumers would be upset. But over time, as the majority of the population bad the opportunity to be stimulated by media reports and other social interactions with angry Coke loyalists. most changed their minds. This is what was predicted by the focus groups.
Given the IO%-12% figure from the quantitative survey, a typical eight- to 12-member focus group is likely to have at least one angry loyalist as a member. The focus group results showed that, in this situation, exposure to the views of angry Coke loyalists is likely to sway the others in the group to their position. By July 1985, Coke executives had sensed that this social interaction was a major factor in causing their problems; it was reported in Advertising Age that Coke officials were blaming the press for “fanning public discontent. Of course, by then it was too late. Coke had already Ignored the research that told them how the market would respond to a flavor change carried out in a public context.
Moreover, by the summer of 1985, Pepsi had also come to appreciate the role of publicity in causing Coke’s problems, and then actually acted on that understanding. As David Gilman, manager of public relations for Pepsi Cola International, described it in Public Relations Journal: “After new Coke was introduced here in the States, Coca-Cola said that it was planning to announce the new formulation overseas.
But months went by, and they never made the announcement to the foreign market. This didn’t sit well with us, so as a public service we conducted press conferences in 18 different countries on Aug. 20. In effect, we made Coke’s announcement for them. ” In this view, the real lesson to be learned from the new Coke affair is that the focus group technique is more than just a means of getting a quick and vivid look at consumer opinion. It is a unique source of information about how the consumer will respond in a situation where there will be an awareness of the views of other consumers.
Even if Coke’s researchers had included in all of their individual interviews and taste tests a question directly asking con- sumers how they would respond if only tbe new Coke flavor were available, it is very unlikely that the results would have predicted the shift in opinion which eventually occurred. Without information on how other consumers will respond, individuals have no means of predicting how their own feelings will change after being exposed to the responses of others. Viewing the events surrounding Coke’s attempted reformulation in this light, one is less likely to fault Coca-Cola’s management.
After reintroducing the old Coke, these executives were subjected to unflattering suggestions that they had failed to appreciate the importance of Coca-Cola in the American psyehe or that they had absentmindedly neglected to ask consumers an obviously relevant question. Rather, with the perspective of some distance, it appears that they indeed appreciated the importance of consumer response to the idea of reformulating Coke and used very acceptable market researcb procedures to try to predict consumer reaction to this idea.
Moreover, in contrast to many executives who devote resourees to market research studies and then ignore the results of these studies in favor of their own personal intuition, it appears that Goizueta and other top Coke executives actually trusted and used their market research results. It is a flaw in accepted market researcb practice, and in the understanding of consumer behavior that supports this practice, which must bear at least some of the responsibility for the embarrassing failure of the Coca-Cola refonnulation.
Although the interpersonal interactions in a focus group setting cannot be expected to be identical to those that will occur under natural conditions, they can be expected to yield a far better indication of the effects of group interaction than methods where consumers are interviewed individually. In individual interviews, whether qualitative or questionnaire-based, there is no opportunity at all for individual opinions to be influenced by the views of other eonsumers.
This distinctive strength of the focus 1 is the active awareness of the opinion of influential others which is the key factor products and felt that it was devious and insulting to try to pass it off as a new product. This invariably led to a change in the views of the other group members, and the consensus of most of the focus groups was negative. Note that the difference between the individual and group results in this ease was due to the ability of the focus group to bring to awareness a viewpoint which was probably not even considered during most of the individual interviews. Apparently, such also was the case with new Coke.
When interviewed individually about the proposed refonnulation, the vast majority of the respondents were not even eon. scious that some others would express feelings of loss and anger. However, the ability of the group setting to make them conscious of such views as they were formulating their opinions made it possible for them to be influenced by this opinion. During the introduction of new Coke, it was largely the media rather than personal interactions which caused this consciousness, but the effect of this awareness on the opinions of most consumers was the same.
Predicting the Effects of Social Influence The failure of market researchers to appreciate the important and unique information which focus groups provide may be one reason behind what Edward Tauber has argued is the relatively poor validity of eoneept-testing research. However, the recent rise in theoretical work on qualitative methods has set the stage for a rethinking of the traditional role of focus group methods. Consumer behaviorists have pointed out that focus groups and individual interviews each have distinctive strengths and weaknesses, and that each is appropriate for different researeh tasks.
The opportunity for respondents to interact with others makes focus groups the best method for measuring the effects of interpersonal influence. group method is not important when individuals can accurately anticipate the views of influential others and are actively conscious of these views when they are being interviewed. For example, suburban homeowners are likely to be highly and accurately aware of how their neighbors would react to a new line of exterior house paints featuring iridescent colors.
In such eases, individual and group methods will yield the same results and the distinctive strength of focus groups is not important. However, when individually interviewed consumers either cannot or do not accurately anticipate the views of others, then the ability of a focus group to bring tbese views to awareness becomes an important advantage. It is the active awareness of the opinion of influential others which is the key factor here.
Consider the case of medical doctors being interviewed by a drug company concerning a new product that dispensed two types of birth-control pills which needed to be taken together. When interviewed individually, physicians indicated a generally positive reaction to the new produet. When interviewed in focu,s groups, physicians initially expressed positive reactions. However, at some point in the typical group, one physician would express a very negative view to the effect that he or she perceived this new product as nothing more than a repackaging of two existing The Validity of Focus Group Results This new appreciation of the distinctive capability of the focus group method leads to the following practical advice.
In new-products research, rather than have focus groups be the preliminary research and the individual interviews of a quantitative survey be the confirming research, conduct preliminary research using both individual and group methods. If the two forms of qualitative research agree, then the confirmatory research ean proceed using the traditional survey methods. However, if the results of the preliminary individual interviews and focus groups diverge, then the researcher needs to consider how much awareness of the views of influential others will exist in the marketing situation the research is being used to predict.
The researcher should carry out confirmatory research using a procedure that provides tbe respondent with a level of awareness of tbe views of others comparable to that which is expected to oecur during tbe rollout of the marketing program in question. If the researcher expects that very few consumers will be aware of the reactions of other consumers to the introduction of the new or altered product, then confirmatory research using individual December 1992 2 5 interviews should yield the most accurate prediction.
But if the researcher expects that many consumers will be aware of the reactions of others, then the focus group results are more likely to be correct and should be confirmed with a research technique that gives respondents a realistic awareness of the views of other consumers. At least three factors can be used to estimate the level of avi’areness of the opinions of others:
The visibility of product. Products that are highly visible, such as automobiles, beer, clothing, and magazines displayed on the living-room coffee table, are very likely to be associated with high awareness of the views of other consumers. nvolves the alteration of an existing product, then even if the product is associated with high visibility, importance, or difficulty of decision, there are many cases where management can choose to roll out the program with little or no consumer awareness.
Coke’s reformulation was almost certainly one of those eases. A gradual phasein of the new flavor with no announcement or other changes would very likely have gone unnoticed. Pepsi had been quietly reformulated from time to time, and some believe that Coke had in f’act, secretly modified its formula everal times during its first 99 years of existence. Consider also that researcher David Pierce found that, even after all the flavor-change.
The importance of the product. Consumers will tend to seek out tbe opinions of influential others when making decisions about risky and otherwise important products such as medical services, day-care centers, and banking services. The difficulty of the decision regarding the product. Even if the product is not especially important to consumers, they will tend to seek out the opinions of others if they regard the product as complex (e. . , personal computers, audio equipment) or if it is a service that is difficult to evaluate before purchase.
For products deemed high in any of these three factors, the results of group interview methods are more likely to be accurate than those of individual interview methods in cases when the results of the two research methods conflict. For the introduction of new or altered products that are not highly visible or important and do not involve a difficult decision, the marketer can often exert some control over the degree of the consumer’s awareness of the opinions of others.
This makes it possible for the marketer to resolve conflicting results of individual and group research by designing tbe marketing program to create the conditions which produced the more favorable result. For example, the drug company that tested the new birth-control product was advised to inform physicians about the product through its detailers so that each physician could form an opinion on an individual basis, as opposed to launching the produet at a iiu-ge conference or other group setting.
Further, if the marketing program 2 6 December 1992 onsideration of the effects of social influence must become a standard part of the new-product development process publicity, consumers rated the new Coke flavor labeled as old Coke to be more desirable than the old Coke flavor labeled as new Coke, Even if it was later revealed—by Pepsi. for example—that Coke had secretly phased in a reformulation, it is unlikely that there would have been any serious consumer reaction.
Consumers’ awareness that they had already been drinking the new formula with no decrease in their enjoyment would probably have made it somewhat difficult for them to summon the indignation necessary to voice public complaints. careful attention should be given to the social context of the consumer decisions that will determine the product’s success or failure. Although focus groups are the best research method for revealing critical social interaction effects, there is a need for methods which make possible the confirmation of the insights of focus group research.
One approach would be the development of procedures for conducting group research on a large representative sample of consumers. For example, such “confirmatory focus groups” might be larger and less homogeneous than traditional exploratory focus groups, and recruitment could be guided by expectations concerning the types of opinions consumers are likely to become aware of after the product is introduced.
A second approach to the development of research techniques to verify focus group results would be to experiment with individual interview procedures that make consumers aware of the full range of viewpoints expressed during the exploratory focus group pha. se and which are likely to eome into eonsumer awareness after the product’s appearance in the market. The real lesson of the failed new Coke introduction is that consideration of the effects of social influence must become a standard part of the new-product development process.
This can best be done by appreciating that focus groups can reveal important insights about social interaction effects that other qualitative methods are likely to miss, and by developing new research procedures that make possible the reliable and representative quantification of these insights. Atihough both the locus groups atid the survey provided indications that there would be consumer dissatisfaction, the survey results indicated that this dissatisfaction would be limited to a smalt segment of the market; the Twus groups suggested the dissatisfaction would be widespread.
Case conlributed by Edith Krieger of Psychonometrics Inc. , Southampton. Pa. Conclusions T be Coke reformulation attempt was a dramatic example bow consumer awareness of the reactions of other consumers can play a critical role in the success or failure of a new or altered product. It highlights the necessity for some explicit investigation of social-interaction effects during concept-testing research. At the very least, exploratory research should routinely be carried out using individual as well as group methods.
