How to quantify the elements of brand and user imagery that drive purchase and create loyalty

Editor's note: Dr. Glenn Livingston and Sharon Livingston are partners in Executive Solutions, a New York City research firm.

For decades, the marketing com-munity has been aware that there are emotional and psychodynamic factors that drive brand selection and loyalty. Even in today's price-sensitive economy, the imagery attached to brands goes far beyond product attributes, functional benefits and price to sell products. All products and brands develop personas in consumers' minds. All project varying user images, which differ by audience. Consumers tend to buy products with imagery that is either consistent with their positive view of themselves, or which conveys a plausible aspirational model (something they would like to be and believe they could conceivably achieve).

While the importance of product imagery is clear, it is tricky to assess via direct questioning techniques. All but the most creative respondents are hard put to provide rich, detailed answers to straightforward questions such as "If this soda were a person, what kind of person would it be?" There are many reasons for this difficulty, but primarily, respondents can't answer these kinds of questions because consumers don't realize (or don't want to admit) that advertising images affect their purchase decisions. Indeed, most consumers want to believe that they purchase based solely upon rational facts such as price, value, taste and performance. Moreover, since consumers tend to deny that product imagery affects their decisions, they can become anxious that their answers to direct image-related questions are a reflection of their personality. The result of all these dynamics is a relatively quiet respondent, who gives sensible, general, barely useful responses.

Despite these inherent difficulties, many researchers (who are unaware of alternative approaches) attempt to ask direct questions to assess imagery in focus groups and in-depth interviews. However, a handful of qualitative market researchers have borrowed techniques from psychology called projectives in order to obtain richer, more detailed descriptions of product imagery.

Using projective techniques to assess imagery in qualitative research

Succinctly, a technique is projective when it indirectly encourages the expression of psychologically motivating material (imagery) which the respondent is otherwise unaware of. Most projective techniques do this by presenting the question so that the consumer believes her response is part of a game which could not possibly reflect on her personality. Projectives allow research participants to sit back, relax, and to view their responses as if they were watching a movie screen, unaware, for the moment, that they wrote the film and that they hold the projector.

Returning to the soda can example, instead of simply asking the respondent "If this soda can were a person, what kind of person might it be?" the moderator positions the question as an experience. She tells respondents they are about to engage in a fun exercise, uses some sort of relaxation technique, helps them to imagine the soda can in their mind (as opposed to directly looking at it) and then says something like "Now imagine you see a hand reaching for the Diet Sunkist . . . what does the hand look like? Describe it in detail. Now, what about its owner? Their occupation? etc." (She continues to get a rich description of the image).

While on the surface, this question may seem quite similar to the more direct question asked above, there are some very important differences in the way it was presented. The primary differences are 1) the degree of intellectualization required of the respondent, and 2) the emotional state the respondent is in when the question is posed.

A projective technique doesn't require intellectual reasoning. For example, the respondent is instructed to imagine a hand, then to imagine the rest of the person. Properly presented, projectives are experienced like a game - like playing make-believe as a child. This is markedly different from the direct, rational question "If this soda were a person, what type of person would it be?" To answer that question, most respondents feel they need a rationale to support their conclusions, which severely restricts their ability to respond. Projectives remove the need for rationale, and make it much easier to elicit potent imagery which the respondent might not really understand (and therefore cannot rationalize). A more detailed example of a projective technique follows.

The moderator takes respondents through a brief relaxation exercise, followed by guided imagery where they imagine seeing a door with a soda can on it. When they walk through the door, they are told they will find themselves in an entirely different scene. After having them explore that scene thoroughly, the moderator tells participants that they will find a person in that scene. They can observe a typical moment in that person's day, knowing that this particular moment is just a snapshot. After thoroughly observing the person in the moment, they can imagine what their day was like before and after that moment. When they open their eyes, respondents are asked to talk about the people they observed. Following are two sample responses. (The results are for illustrative purposes only, and are based upon only one respondent's answers.)

Diet Sunkist - "I saw a young woman in her 20s drive up to a local convenience store. She was driving a red sporty convertible (I guess it was a Mustang) with the top down. She had just finished a game of tennis and was very thirsty. She wanted something light, sweet and sparkly. She went into the store and got the Sunkist. At the register, the guy flirted with her. She smiled at him, paid for the soda and jumped back into her car. It was a Sunday, the weather was warm, but not too hot. She took a long drink of her soda, felt refreshed and drove off to spend the afternoon with some friends."

Diet A&W Root Beer - "A man in his mid 40s pulls up to the same store in his pick-up. He feels depressed and upset because he had an argument with his wife this morning. She told him he was getting fat and flabby and looking just like his father. They had a blow up about it. He gave her a hard time, but he really felt bad. He walked into the store wanting to soothe himself with some ice cream. On the way to the freezer, he passed the soda case and noticed the Diet A&W. He remembered how much he used to love root beer when he was a kid. He picked up the diet soda and paid for it. He opened it up and took a swig and was surprised at how good it tasted. He got back in his truck and headed home, feeling hopeful and excited to tell his wife that he had taken a first step in the right direction."

A little imagination should convince the reader that projective exercises like this one can yield a wealth of imagery, and frequently some very useful creative insights. However, many in the field have raised concerns over the interpretation of projective techniques. Their concerns are:

  • The complicated manner in which the questions are asked, and the indirect manner in which they are answered, makes it very difficult to translate responses into concrete, useful marketing information.
  • The commonly agreed upon purpose of qualitative research is to develop hypotheses which later need to be validated in quantitative research. Traditionally, the rich imagery uncovered in projectives has been very difficult to quantitatively verify. The most common practice is to revert back to direct questions about attributes in a quantitative questionnaire.
  • Because of the difficulty in quantifying projective methods, it has also been difficult to determine which specific aspects of the imagery elicited from these techniques actually drive purchase interest.

How can projective imagery be validly assessed quantitatively?

Traditionally, imagery is quantified via simple attribute check lists or semantic differentials. These are extremely limited direct methods which leave little or no room for the respondent's free associations and imagination. They are also subject to the same response-inhibiting effects as direct questions about imagery in qualitative research.

However, it is possible to use the same types of projective techniques quantitatively, in a reliable and valid manner, with just a few simple modifications. A description of a sample quantitative projective procedure follows:

The technique to be used is first pretested and refined in one-on-ones for the particular product and target to be assessed. Then, the moderator is videotaped giving the instructions for the projective technique. Qualified respondents are recruited to a central location. There they view the videotape, which also instructs them to write down their projective experience as a detailed story with a beginning, middle and end (a field interviewer sits outside the room to answer any questions). They are also asked to respond to a series of written prompts regarding specific characteristics of their fantasy (e.g., about how old was the person they imagined, what sex, etc.). Any more traditional quantitative questions (like purchase interest) are presented last.

Respondents' stories are analyzed using a variety of scales which have been proven reliable and valid for analyzing fantasy material. For example, many psychological scales used for 25 years in the study of dreams and daydreams can be applied. These include scales of induced mood, creativity, masculinity and femininity, assertiveness, aggression, anxiety, satisfaction and many others. Of course, the traditional attributes are available, with the pointed difference that they were elicited via free associative/non-intellectual methods - which is the way that imagery influences purchase motivation.

The result is that every dimension of imagery thought to be important in a particular category is quantified! Statistical analyses can be run to determine which dimensions predict purchase interest, and which are less important.

Here is an example of the use of quantified projective imagery:

A food manufacturer wanted to test two packaging options for a new line of fat free, low sodium soups. Because lowered fat and sodium levels are associated with reduced taste perceptions, brand management strongly believed that the product packaging had to create the impression of a zippy, zesty, spicy and delicious product. The target was the primary grocery shopper aged 25-54.

Both packaging options were taken to qualitative (a third was actually eliminated there). There, projective techniques like the one described previously revealed three very different sets of imagery:

Option A: For the most part this version created the impression of a user who was a young, urban woman. The majority described her as a career woman in her late 20s, concerned with keeping fit as well as feeding her family healthy things. She worked out regularly, had a few close friends who understood that she just barely had time for them, and was happily married with small children.

Option B: Created the impression of a slightly overweight man in his 40s or 50s. His wife bought the soup because she is trying to get him to take better care of his high cholesterol. Most felt that the wife didn't consume the soup herself and didn't feed it to her teenage children.

Given the vastly different images created, the projective was submitted for quantitative verification. Four hundred interviews were conducted (one in each of four geographically dispersed cities). In addition to the projective, traditional measures such as purchase interest were taken. Last, respondents were asked what element of the packaging first drew their attention.

Stories were scored for activity, satisfaction, overall mood, imagination and anxiety. (Anxiety refers to respondent mentions of types of anxiety - guilt, separation, shame or other non-specific forms - in their stories. An example of guilt anxiety related in a story is: "The woman I saw was in a candy store. She bought some jelly beans even though she knew she was getting fat.") Respondents were also asked how old the person behind the imaginary door was, what their gender was, how much they liked this person and how similar the imagined person was to themselves.

Package A yielded purchase interest almost identical to package B. This was the case across demographic segments.

With regards to imagery, however, package A was significantly different from package B in five respects:

  • The package A user was seen as significantly younger than the B user.
  • The package A user was seen as significantly more active than the B user.
  • Package A elicited significantly more anxious stories than package B.
  • The package A user was described as significantly more dissimilar to the respondent than was package B.
  • Respondents liked the package A user significantly more than the package B user.

The quantitative results matched the qualitative in that package A created the image of a young, active person whom the respondents seemed to like. However, this did not seem like enough to base a packaging decision upon, since package A also created more anxiety than package B, and package B made respondents think of users who were more like themselves. Clearly more statistical analysis was required.

First, it was necessary to know which of the dimensions assessed actually were correlated with purchase interest. The answer turned out to be activity, youth, liking, and anxiety. The more active and younger the imagery was, the higher the respondent's purchase interest. Likewise, the more they liked the imagined person, the higher was their purchase interest. Perceived similarity to the respondent made no difference. However, the more anxious the elicited stories were, the lower was purchase interest. This was the largest correlation.

Statistical adjustments (via an analysis of covariance) were made to remove the effects of the anxiety on purchase interest. Under these conditions, package A clearly elicited much higher purchase interest.
At this point it seemed that package A would be the preferred option, if only the cause of the anxiety could be determined and removed. An exploratory CHAID analysis with the anxiety score as the dependent variable revealed that respondents who gave anxious stories were much more likely to say that the first thing that drew their attention on the packaging were the words "low sodium." Many more of these people had seen package A than package B.

Analysis showed that package A's purchase interest was deflated by anxiety caused by the prominence of the words low sodium. (Both low sodium and fat free were indeed more pronounced in package A.) It was recommended that the client de-emphasize low sodium and go with package A. It was also suggested that the client conduct more qualitative research to determine why reminding consumers about the sodium content in these soups caused anxiety.

It is noteworthy that without the quantification of the projective techniques in this example, there would have been little solid evidence upon which to base a packaging choice. (Recall that overall purchase interest was equal.) Because of the availability of quantitative projective dimensions, it was possible to isolate what was bothering people about package A (and suppressing otherwise higher purchase interest).

Summary

All brands and products project an image, which is at least partially responsible for variance in purchase interest. Because imagery is difficult to assess by traditional direct questioning methods, projective techniques have been borrowed from the behavioral sciences. Projectives are simple psychological exercises which encourage the expression of unconscious motivational material while protecting respondents' self perception (i.e., their need to believe that they are not motivated by such things as product imagery).

Until recently, projective techniques have been used only in qualitative research, and only by a handful of skilled moderators. Use in this manner has been criticized because there has been no way to quantitatively test hypotheses generated in focus groups by projectives, and their interpretation has therefore been open to wild speculation.

This article presented a simple method for the large scale administration and scoring of projective techniques. In addition to rendering the results of such techniques testable, the benefits of projective quantification include 1) the addition of several rich dimensions to more traditional quantitative tests, 2) the ability to determine which particular aspects of product imagery drive purchase interest in a given category, 3) the ability to quantitatively isolate what aspects of a product are creating a particular image or feeling.