Tobacco Documents Online

Ilk THIS REPORT IS C4NFIDElfrIAL TO THE BUSINESS OF 'HE CCFPANY: IT SHOULD BE CAREFULLY HANDIED. IS WT "OJIMSFERABLE TO ANOTNZR 1 N01 Y i DUAL, AND I S NOT T" BE PHOTOt:OP'1ED. If ths report has served Its purpose and is no Ionger needed, pisas• return It (a+sedlately to th• Central Fil• at the Research Center for record keeping purposes and destruct!on. P H I L I P M 0 R R I S U. S. A. RESEARCH CENTER RICHMOND, VIRGINIA CHARGE NO. & TITLE: Number : 87-015 Copy Number: Issued To: S. S U) ~ 0000011672 TYPE REPORT: 0 ANNUAL 0 SEMIANNUAL 0 COMPELETION DATE: July 8, 1986 PERIOD COVERED: REPORT TITLE: 1985 New Product Concept Evaluation DISTRIBUTION: C. Bates (NY) W. Claflin P. Gauvin W. Geiszler R. Heretick W. Houck K. Houghton R. Ikeda Peter Martin M. Jeltema M. Johnston J. Jones A. Kallianos A. Kassman H. Lanzillotti P. Martin L*: Meyer 0 SPECIAL Francis J. Ryan F. Ryan H. Spielberg J. Tindall F. West L. Wu J. Zoler (NY) Central File (2) KEYWORDS: Concept Testing, Mall Intercept, Liqueur Flavored, Designer Pack, Distinctive Flavor, Tinted Paper, Light Smoke Aroma, Plastic Pack, Ashtray Attachment, Consistent Strength, Different Shape, Individually Wrapped, Assemble Your Own, Nonstaining Filter, New Manufacturing Process, Flavor Change, No Visible Smoke, Nonburning, Premium, Supplement, Slim King-Sized, Thicker, Ultra Low Version, Nicotine Free, Adjustabl.: Tar, Only Burns When Puff, No Ashtray Odor, Breath Freshener, Attributes Accession

TABLE OF CONTENTS ABSTRACT 1 INTRODUCTION 2 PREPARATION 3 METHOD 4 Respondents 4 Procedure 4 RESULTS 5 Quantitative Data 5 The More Attractive Concepts 6 Moderately Interesting Ideas 7 Additional Concepts of Some Interest 9 Problems Ideas 10 Reproducibility of Findings 11 Yeasayers and Naysayers 20 Ranking Data 21 Interest in Trying Flavored Cigarette 24 Qualitative Data V 27 Concepts of Most Interest 27 Concepts Ranked Upper Middle 31 Concepts Ranked Lower Middle 35 ~ Concepts of Least Interest 39 ~ ~ SUMMARY AND RECOMMENDATIONS 42 ~ ~ ACKNOWLEDGMENTS 43 ~ FIGURES APPENDICES

ABSTRACT A concept testing procedure initiated in 1983 was repeated in 1985 with a new set of 26 novel cigarette product ideas. The concept statements were rated and ranked on triai interest by 693 smokers in eight shopping malls across the country. Each respondent provided qualitative information about top, middle, and bottom ranked ideas. Results inciicated that new product ideas addressing issues of social pressure (health and consideration of nonsmokers) evoked strong trial interest. Those concepts having what consumers considered to be beneficial attributes received higher trial interest ratings than those which were viewed as having no relative advantage over currently marketed products. The distributions of rating scores for the six benchmarks borrowed from the 1983 study demonstrated the reliability of data generated by the test methodology. Concept testing provides insight into the needs of today's smoker, and has direct application both to consumer evaluation and to product development.

2 INTRODUCTION In 1983 a methodology was developed to provide early-stage consumer evaluation of novel cigarette product ideas.(I) Concept testing, as conducted in Research and Development, is procedurally different from the research practiced in Marketing or Marketing Research divisions.(2) Whereas their testing involves a detailed product concept, complete with advertising claims and possibly brand name and image, our R&D research is closely related to early-stage idea screening. The concepts are concise statements of core ideas, serving as a set of stimuli to evoke responses from consumers. We emphasize what the smoker tells us regarding attributes, issues, and benefits as the important informa- tion derived from concept testing. These conclusions are then communicated to Development, to guide the direction of future products toward consumer desires. The concept test methodology is designed to precede, not to replace, prototype testing. In 1983 our concept testing was a new methodology that lacked an experience database to aid in interpretation of the quantitative results. We realized the potential for misinterpreting the data, and were skeptical of how the quantitative findings might be used. To reduce the impact of such effects on our results, and to learn more about our new methodology, we relied heavily on the qualitative information that smokers provided in their open-ended responses.(3) In 1983 we asked smokers for their reasons for ranking the top two ideas as being of most interest, and for classifying the bottom two ideas as being of least interest. In 1985 we repeated the basic concept test procedure used in the initial study but, because of the important insights gained through the qualitative information, we increased the number of open- ends to six: the top three ideas, the middle (13th), and the bottom two. The middle concept was included to avoid losing important consumer information about those concepts which were of neither strong interest nor disinterest. Based on what we have learned from the two tests, and subsequent marketplace events, we now feel confident that concept testing is an effective method for consumers to evaluate innovative product ideas. 1. Jones, Jan A. "1983 Mall Test of New Cigarette Product Ideas," March 7, 1984 (Accession Number: 84-054). 2. Memo from Jan Jones and Frank Ryan to Dr. Peter Martin, "Concept Testing as a Methodology," March 18, 1986. 3. An "open-end" question is asked in a way to evoke the most information from the smoker without offering clues as to what might be appropriate responses. In this study the open-ends were prompted by "why" questions, and smokers' responses were probed with "Why did you say that?"

3 PREPARATION In 1983 our list of 26 concepts included: new ideas not previously evaluated by the consumer, ideas for which the prototypes were curreatly under development, and a few ideas representing products previously marketed. This latter group of concepts served as benchmarks for the potential marketplace performance of new ideas (assuming the statements accurately conceptualized the products). In 1985 we included six of the ideas tested in 1983, unchanged in wording, selected from among the ideas of most consumer interest, of intermediate interest, and of little interest. These provided a new set of reference points. In addition to the six ideas unchanged in wording, we included five concepts that were modified based on what we had learned from the 1983 open-end responses and from other qualitative research conducted post-1983. For the remaining 15 concepts, we turned to several procedures for generating new product ideas. We derived input from ongoing development projects and from brainstorming sessions; in addition, a series of synectics groups was conducted among employees from various departments within Philip Morris U.S.A.(4) The ideas that were generated then went through a screening process to yield 26 concepts addressing a variety of attributes of potential interest to consumers. The concept statements evaluated in the 1985 study are listed in Appendix 1. 4. Ryan, Francis J. Generating Concepts for the 1985 Concept Evaluation Study, March 10, 1986 (Accession Number: 86-066).

METHOD RESPONDENTS Six-hundred and ninety three filtered cigarette smokers were recruited through mall intercepts in eight cities across the country: Buffalo, Cincinnati, Clearwater, Denver, Grand Rapids, Knoxville, Phoenix, and Spokane. The percentage of smokers in each quota group appears in Appendix 2. Respondents were screened for affiliations with marketing research firms, advertising agencies, or cigarette manufacturing companies, and those with affiliations were excluded from the study. The minor differences between cities are shown in Appendix 3. PROCEDURE The procedure was similar to that employed in the 1983 study. Respondents who met the security requirements and fulfilled the approximate quotas were escorted to the testing facility. The smoker was seated at a large 7-point scale (sort board) of trial interest, word anchored with "Not at all Interested" (1) and "Extremely Interested" (7). The respondent then was handed a stack of 26 cards (shuffled to randomize the order), each describing a new cigarette idea in a brief concept statement. After reading each statement, the smoker placed the card on the number that corresponded to how interested he or she would be in trying the new product idea if it were available. Within each rating, the smoker then ranked the ideas from the one of most interest to that of least interest. These ranks within ratings were then converted to an overall rank of 1 to 26. The smoker was asked to provide comments as to why particular concepts were placed in the rank positions of: first, second, third, middle (13th), 25th, and 26th. These open-end comments were recorded verbatim by the interviewer. The screener and questionnaire appear in Appendix 4. At the conclusion of the interview the smoker was asked how interested he or she would be in trying a flavored cigarette, other than tobacco tastes or menthol. Those who gave a trial interest rating of '5' or above were handed a flavor checklist of 44 flavor names. For each, the smokers were asked to chevk whether they were or were not interested in trying that flnvor in a cigarette.

5 RESULTS The data were analyzed using both a qualitative and a quantitative approach. The mean rating distribution for each new product idea reflects trial interest for a particular concept; a comparison of these distributicns suggests interest relative to other novel product ideas. In 1983 w,_ de-emphasized the quantitative information due to the lack of a database for interpreting the numbers. Since 1983, several of the new product ideas wer~_- further tested as prototypes, and a few went to test market. Information recarding consumer interest in these products has begun to build a database fJ,• quantitative interpretation. In the present study we included several --oncepts tested previously to serve as a reference group. PA comparison of data points from the two studies was made to assess stability of the measurG, or test-retest reliability. These results will be discussed in the following section on Quantitative Data. The qualitative information provides consumers' rationales fDr their quantitative responses. The over-4000 open-end comments we-e coded and interpreted for key issues influencing the ratings and rank=ngs of each concept statement. The rating scores and relative rankings of ideas may be influenced by a number of factors. The reasons given for taese rankings, however, are clues which may be important in prototype development and future testing. A brief summary of the qualitative data is presented here; a more thorough discussion of actual verbation responses has been reported separately (Accession No. 87-013). QUANTITATIVE DATA In presenting the results we will first show the distributions of trial- interest rating scores for each of the concepts, beginning with those liked best and progressing through those which evoked very little interest. Next we show some distributions for ideas evaluated both in 1985 and in 1983 to Y see how stable the opinions were, and progress to a few items which showed different distributions among different demographic and smoking history subgroups. These distribution differences are so slight that we then abandon the distribution format in favor of presenting only the means for different subgroups. Because exhaustive tabulations of subaroup means would serve little purpose when tabled across many groups and 26 concepts, we restrict these presentations to the relatively few cases where we had some a priori reason to expect different rating scores among specific types of people plus those cases which others have asked about during several presentations of the data to different management and development groups in Richmond and New York. Finally, we present some of the ranking data to give the reader a feel for the way individuals evaluated the importance of this idea compared to that idea. This is the only head-to-head comparison the consumers made. We hav2 applied no inferential statistics to the basic findings. We feel that it would not only be misleading to do so, but that the departures of the rating scores from assumptions which underlie the most c,~mmon?y used inferential statistics--particularly in regards to the shape of the distributions--would make such common tests as those of t ar:,{ F technically

6 conceptu,~lized ideas in different words might have made any given idea more vr less attractive. We wish the reader to conclude that some ideas seem more attractive than others, but we do not wish to suggest that any one idea is best or worst. Instead we emphasize the fact that several ideas were very well liked, while several other ideas were unable to generate much if any interest at all. The More Attractive Concepts The best liked concepts addressed contemporary smoker problems of health and anti-smoker social pressures. This message is very clear. Three of the top ideas were health related. The first proposed a Nicotine- free cigarette; a second suggested that a beneficial supplement, such as a vitamin, be delivered in the smoke stream; and the third promised an Ultra- low Tar Version of a Popular Full Flavored Brand. The distributions of the rating scores for these three concepts are shown as Figures 1, 2 and 3. The percentage of the 693 respondents giving each rating score is measured along the Y-axis and the rating scores are shown on the X-axis. A rating score of 1 is given to a concept by panelists who are not at all interested in trying such a product. A rating score of 7 is given by a panelist who is very interested in trying -ach a cigarette. Note that these three curves are generally J-shaped, with many panelists (37 to 50% of those tested) saying that they are very interested in trying these ideas, progressively fewer expressing progressively lower degrees of interest, and about 159'o expressing no interest at all. For detailed comments on these ideas, see the qualitative section, which covers the individual explanations given by people who ranked (not rated) these ideas as among the best or worst. In examining the distributions here, and those to be presented below, it is possible to concentrate on the frequencies at either extreme of the scale. We are inclined to think that the lowest rating has a different message than the other ratings. It is certainly easier for the smoker to tell us that a concept is uninteresting than it is to make the gradations of interest reflected in the higher scores, and it may be that some proportionate score, such as the number giving a concept a rating of 5 or higher divided by the number giving a rating of 1, will prove to be the best index of an idea's attractiveness. In order to test such hypotheses, however, we will need data cn the eventual success of different ideas in the marketplace. For the present we can only point out that every idea we have ever evaluated in this procedure has had at least a "tail" on its J-curve, in which more people gave a rating of 1 than gave a rating of 2. Indeed, there is only one exception to the comment that there are always more ratings of 1 than ratings of 2 and 3 combined, (see below), so that the frequency of the 1 rating does seem important. Note that what is being evaluated is the interest in trying such a product. There is a big step between trial and a subsequent approval of a prototype illustrating the idea, and another big step between the approval and a decision to switch to a new brand based on the idea. A prototype of an interesting idea could be excellent, but the idea and the prototype could still not be powerful enough to break the habit of requesting an old

7 familiar brand. The case of the Nicotine-free cigarette serves to illustrate this problem. Our data show that a lot of people are very interested in trying such a cigarette, and we suspect that we can develop one which has a rich and rewarding taste. But if, as has been generally assumed, people need c;- want the t,)ost provided by nicotine, a nicotine-free product will prove "unsatisfying" in the long run, and may therefore be predestined to fail. Certainly the industry folklore would predict failure, but the industry folklore has not always been correct and so we are willing to reserve judgment on the matter. This is particularly true for those smokers who do not smoke out of any "need" for nicotine, but instead smoke out of habit. Indeed, we are not impressed that there is any single reason people smoke, but that there are many reasons, so that many people may be attracted to a low nicotine product. The ultra low tar concept had been evaluated very highly in 1983, when it was one of the top two ideas tested. In 1985 it did about as well as it had done earlier, but it was not as relatively high as it had been, because this time it had to compete with some other attractive notions which addressed anot"ar smoker need: handling public criticism of the smoking act. Two of these top-rated concepts related to sidestream smoke and one to ash residuE. The issue of a "light smoke aroma" had fewer ratings in the not- at-all-interested category than any other concept yet tested. (See Fig 4.) Somewhat ahead of it in high interest and only slightly benind it in non- interest was the related notion of "no ashtray odor." (Fig 5.) Also well rated was the notion of "no visible smoke", which was qualifir~ within the concept description and generally understood by the smokers to mean "almost no visible smoke." (Fig 6.) That these three ideas did as well as they did, given that they deal with what would seem to be inevitable by-products of the combustion process and that they may have suffered somewhat in believability as a consequence, reflects the smokers' desire to be nonoffensive to others. Our critics tell us that smokers are selfish people who don't care whether they offend others or not by their evil self indulgence. The smoker interest in these three areas tells us that our critics are wrong: smokers do not want to be offensive and are actively interested in new products which would lessen the criticisms. Collectively, one or another of these three related ideas was ranked first in importance by over 13% of those interviewed, (see below), while the three health- related ideas above garnered almost 44% of the votes for the most interesting idea. The concept of a cigarette that "only burns when you puff on it" appears closely related to the smoke and ash ideas above and was somewhat similarly received (Fig 7), but, as will be seen later, it had some other appeals which suggest that we treat it separately. Moderately Interesting Ideas The J-curves shown so far have had relatively few people rating ideas as a 1 or a 2, which axpress very little interest in the concepts. Now we come to six ideas which have more U-shaped distributions, with over 20% of the smokers uninterested and under 30% extremely interested in trying them. Again we treat them as a group rather than stress the differences in their distributions. In general, these are product related ideas rather than

8 them as a group rather than stress the differences in their distributions. In general, these are product related ideas rather than ideas related to larger issues: a plastic pack (Fig 8); a new manufacturing process (Fig 9); a last- puff breath freshener (Fig 10); a nonburning product akin to Favor (Fig 11); a nonstaining filter (Fig 12); and a cigarette which delivers a consistent strength (Fig 13). Our decision to group these particular ideas together is partly based upon convenience of presentation and partly upon the fact that they appear to tap different consumer motives than those above. The 26 concepts received widely different receptions, and the distributions of the ratings look like a continuum ranging from a pronounced J-shape through these U-shapes to remarkably mirrored-J-shapes. The groupings -,re therefore rather arbitrary. However, we are now well into the list of 26, and it does not appear that these 6 concepts were rated like the Nicotine-free idea or the Light Smoke Aroma notion, but may reflect marginal interest. Whether ideas in this range (or any other range) can succeed in the marketplace is unknown. Existing information from interviews with consumers suggests that Favor, for example, is getting a lot of trial interest in the Southwest, where its market has just been extended, and it is trial interest which we are measuring here. However, Favor has generally disappointed most who have tried it, and its repeat purchasing is low. That does not mean that a better execution of the idea could not succeed, and in a sense the success or failure of any product is irrelevant to the understanding of this concept testing procedure. The point to be made is that a middle range idea was tried by many people, which implies that ideas which are better received tha: the concept of a nonburning cigarette would be tried by even more people. The distribution for the notion of a Consistent Strength cigarette is a little different from those of the other concepts. It is relatively "flatter" than the other distributions. This concept received more ratings of 4 or 5 than any other concept did, and was sec-:nd in the number of 3 ratings by a half a percent. It is clear that there is some interest in the ,idea, but what it means to have a lot of moderate scores is not at all clear. One economical interpretation would be to conclude that consistent strength is a fundamentally desirable attribu~te, but not one of extraordinary interest. This brings us to make a comment on the effect of the consumer's degrees of freedom in assigning rating scores as opposed to ranking scores. Any concept can be assigned any rating score from 1 to 7. The scores assigned are theoretically quite independent of the scores assigned to any other idea. In practice, however, the consumer tends to remember how well he has rated prior ideas and to make later assessments with this in mind. This is not true of the ranking data, for obviously once one idea has been ranked #1, the best of the lot, no other idea can be ranked #1. The presence of several very attractive ideas on a given test will therefore change the possible rankings of the remaining ideas. This is what the Ultra-low Tar Version ran into this year. The rating scores don't have this problem. They do have a counterpart of it, however, in which the consumer thinks "I'm not very interested in this, so I can't give it a 7, but on the other hand I'm not completely uninterested in the idea, either, so I can't give it a 1, so I guess I'll give it a 4 or a 5," or uses some similar line of reasoning to assign his statement of mo&rate interest. Another problem with ratings is that their very independence permits the smoker to be lazy in making evaluations, giving many ideas the same score instead of applying some discriminative effort to the task. A real "yea-sayer" could rate everything as wonderful, and give each concept a 7. However, rankings require discrimination, and force the smoker to break such ties.