Journal of Advertising Research Advertising Research Foundation 641 Lexington Ave., New York, NY 10022; (212) 751-5656

November/December 1997

What Do You Want Your Brand To Be When It Grows Up: Big and Strong?

Nigel S. Hollis and Andy Farr

In April 1997 the long awaited clash of the titans took place at the Advertising Research Foundation Annual Conference. Larry Light, the champion of brand loyalty as the main mechanism of brand success (1993), took on Professor Andrew Ehrenberg, champion of Double Jeopardy (the empirically based generalization that big brands have more brand-loyal users than small ones simply because they are big (1997). Light attempted to demonstrate that consumers could possess a clear attitudinal predisposition toward a brand that resulted in stronger sales and profit levels for the brand concerned. Ehrenberg replied by demonstrating that the behavioral loyalty of most brands within a particular category could be predicted on the basis of their penetration alone, without resorting to more attitudinally based concepts of brand equity. Each of the contenders argued their case with eloquence and passion, but at the end of the day the two seemed as opposed as ever. However, in the opinion of the authors, neither viewpoint seems to give a complete view of what most of us in marketing and related disciplines observe in real life. In many cases, behavioral loyalty is directly related to brand size within a category and appears to be remarkably resistant to change. Much marketing effort, however, is spent trying to buck that trend, and success in doing so is often attributed to building attitudinal equities, the perception that a brand is better or more appealing than its competition. This suggests that in real life both theories can apply. In most product or service categories brand size does have a strong influence on the behavioral loyalty of consumers. In many cases, however, this does not prevent a brand from temporarily, or permanently, deviating from that relationship, for good or ill, as a result of changing attitudinal affiliations.

THE SCOPE OF THIS PAPER

In this paper we seek to demonstrate that brand size alone does not dictate behavioral loyalty and that attitudes toward a brand consumer equities do have an important role to play in determining a brand's success. The paper is divided into three main sections.

First, we will review the fact that there are strong Double Jeopardy relationships in attitudinal as well as behavioral data. We will argue that there are two basic mechanisms that cause Double Jeopardy relationships within categories: push and pull. Push mechanisms are the result of marketing activity's direct influence on consumer decisions at the point of purchase. Pull mechanisms are a result of marketing efforts that establish a longer-term positive consumer predisposition toward a brand.

Second, we will show that stable deviations from category relationships do exist. These are the result of the particular brand's strengths and weaknesses relative to the competition. While stable deviations are more common, trends in deviations over time are observed for both new and established brands. We will argue that brands grow by temporarily deviating from the normal category relationship to achieve a better position relative to other brands. We believe that there is a need for the marketing team to better understand how they might break the existing status quo if they are to grow their brand effectively.

The BrandDynamics™ System, a survey research tool for explaining behavioral and attitudinal loyalty, was created with just that objective in mind. In the final section we show that a BrandDynamics™ measure of attitudinal loyalty called Brand Strength has a clear relationship with longer-term changes in market share. This relationship is far stronger than would be expected on the basis of size alone. The final conclusion is that brands can be both big and strong.

SECTION 1: ATTITUDINAL DOUBLE JEOPARDY RELATIONSHIPS

Making the distinction between behavioral and attitudinal loyalty

In our introduction we make a distinction between behavioral and attitudinal loyalty. It is important to recognize that these are indeed two separate concepts, and we highlight the distinction now since it will be of relevance throughout the following discussion. Behavioral loyalty refers to observed buying behavior, where loyalty is inferred from the patterns in that behavior. If a person buys the same brand within a category more than a set proportion of times they are considered loyal to that brand. No attitudinal loyalty need be involved, although it is often assumed. On the other hand, attitudinal loyalty is inferred from what people say, not what they do. In response to survey questions or in conversation they might refer to a brand as the best, the only one they would consider buying and so on, and from this we assume a strong attitudinal loyalty toward the brand. It does not, however, mean that they will necessarily buy that brand. The fact that people do not necessarily follow through on their stated intentions has been a long-standing frustration of survey researchers. This is an issue that we specifically sought to address in developing the BrandDynamics™ System, and we will return to describe how later in the paper.

Attitudinal double jeopardy

Over the years Ehrenberg and colleagues (1995) have consistently documented the presence of Double Jeopardy in a wide variety of categories. There can be little doubt that this is a widespread behavioral phenomenon. What is less well recognized is that Double Jeopardy is also apparent in attitudinal data. This was noted by Ehrenberg (1969) in his paper 'Towards an Integrated Theory of Consumer Behavior' but has not been as well documented as its behavioral counterpart. Like that phenomenon, however, Double Jeopardy effects in attitudinal data are prevalent and can cause misleading conclusions to be drawn if not properly understood. For instance, a brand might appear to score well on a particular image attribute by comparison to other brands in its category but actually be under performing given the proportion of consumers who use it.

Identifying category relationships in image data

Gordon Brown documented the consistent relationship between brand size and brand image data in 1986. He also demonstrated how to identify the deviations from this underlying pattern and showed that it was these deviations that often led to the most useful marketing findings. We have found that it is how well a brand performs relative to both its own size and the competition that reveals the true strengths and weaknesses of a brand. Similarly, we have now extended this analysis to look at awareness and claimed buying behavior, either in terms of reported past behavior or future intentions. The findings from this type of analysis have not been reported in print before; thus we will describe the technique and then show how it can be used to shed light on the fundamental relationships between brands in a category.

Identifying category relationships in awareness and usage data

A consistent relationship can be identified in most categories between survey measures that might reasonably be expected to be related, for instance, brand awareness and claimed trial, claimed brand familiarity and purchase consideration, and claimed trial and recent usage. Similar relationships were reported concerning brand awareness measures by Laurent, Kapferer and Roussel in the 1995 MSI papers on empirical generalizations referred to earlier. To some degree, the nature of the relationship is that of the chicken and the egg; it is not clear which comes first. Intuitively, however, it makes sense that for most goods and services awareness is likely to precede trial (if only by a few seconds in some cases), and that for more involved purchases the acquisition of a certain degree of familiarity is likely to precede strong purchase consideration. As we shall see, our data supports this direction of causality.

The relationship between survey measures of the type described is nonlinear. This results in a concave relationship like the one for a US cough and cold category shown in Figure 1 (each point on the graph represents a brand). From Figure 1 it will be apparent that the more people who are aware of a brand the more likely they are to claim to have tried it. The relationship is not proportional, bigger brands (in awareness terms) get more than their fair share of trial. It is a classic Double Jeopardy relationship. We have found that a simple logit transformation of the two sets of data under study allows us to fit the relationship between the two measures using linear regression. Figure 2 shows the same example but with the data transformed and a trend line fitted.

As you can see the relationship between brand awareness and trial in this category is strong (r² of 0.97), and there is relatively little deviation from that relationship. Table 1 summarizes the correlations observed across a variety of US packaged-goods categories (percentage scores) and shows similar data for a set of automotive segments (most relationships are based on top two box scores from five-point scales). This summary is not based on a census from across all our studies (merely the most recent), but most do show strong relationships between the measures where the brands are of a fairly homogeneous nature. Weaker relationships are observed where sets of brands are differentiated by some fundamental feature, for instance, price and quality in a liquor market or import versus domestic brand in the automotive category. In fact, within segments automotive cases still tend to deviate from the underlying category relationships more than do the packaged-goods cases. This is probably reflective of greater actual product and pricing differentiation and higher consumer-purchase involvement as much as the difference in survey measures compared.

TABLE 1: CORRELATIONS ACROSS US PACKAGED-GOODS AND AUTOMOTIVE CASES

So, strong category relationships do exist in both behavioral and attitudinal data, with Double Jeopardy working in favor of the bigger brands. The unanswered questions are: Why do these relationships exist, and why do they benefit the bigger brands?

Why do strong underlying category relationships exist?

Most authors have been content to identify that Double Jeopardy relationships exist and have left the causes of these behavioral or attitudinal relationships unexplored.

A brief review would suggest that there are two basic ways in which Double Jeopardy might come about. These can be referred to as push mechanisms and pull mechanisms and might be expected to operate simultaneously within the same category.

Push mechanisms: forcing consumers to buy a brand

The push explanation would suggest that once a category arrives at a steady state there are various structural advantages that attach to bigger brands that keep them big (and so keep small brands small). Essentially these advantages influence people at the point of purchase by steering their choice to the bigger brands in a category.

One of these potential advantages, which is itself an expression of McPhee's (1963) original supposition that Double Jeopardy is simply a statistical artifact derived from differences in exposure, is discussed by Reibstein and Farris (1995). They suggest that there is a two-way causality between share and distribution, such that bigger brands are more likely to be stocked, guaranteeing purchase by existing buyers, but also allowing them to gain from consumers making a 'compromised' choice because their brand is not stocked. Distribution is just one of the potential structural factors that might cause Double Jeopardy. Other mechanisms whereby big brands have an inherent advantage are: greater shelf-space allocation; more in-store promotions; greater likelihood of getting line extensions listed; greater advertising budgets, etc. All of these reflect ways in which big brands might push consumers into choosing them rather than smaller brands. This mechanism seems likely to affect behavior almost irrespective of peoples' predisposition toward the brands; moreover, there is reason to expect that any action that seeks to overthrow the status quo will be rapidly met by counteraction by other brands within the category (1995).

As well as identifying some of these possible structural causes of Double Jeopardy, Reibstein and Farris (1995) also recognize that 'when search loyalty is low the effect of distribution on share is greatest.' In other words, the push explanation of Double Jeopardy will have most sway in categories where the brands are readily substitutable and loyalty is low. Double Jeopardy, however, has been shown to exist in all types of categories, including services and durables, where products would seem to be less substitutable and search loyalty high (although as we saw above, in the automotive example, the relationship may be weaker). This leads us to consider whether there are purely attitudinal causes of Double Jeopardy. Are the patterns described above simply the reflection of constraints imposed on behavior, or are there reasons why people might want to buy bigger brands more than small ones?

Pull mechanisms: making consumers want to buy

In one of the few papers to consider the possible attitudinal or pull causes of Double Jeopardy, Rindfleisch and Inman explore three alternatives (forthcoming). The first, the mere exposure hypothesis, suggests that repeated exposure to a brand or its advertising could lead to an increase in positive affect that is greater for bigger brands. The second, the information availability hypothesis, suggests that the availability and accessibility of information on competing brands will favor big rather than small brands. Lastly, the social desirability hypothesis suggests that consumers are often swayed by the sheer weight of popular opinion, real or perceived, and so will tend to choose the bigger brands in a category. While the first two of these cannot be considered in isolation from the push factors described above, they do suggest a pull-driven reason for Double Jeopardy's existence. Although Rindfleisch and Inman (forthcoming) acknowledge the shortcomings of their experimental approach, they conclude that 'Of our three hypotheses, social desirability appears to be the strongest theoretical explanation of the familiarity-liking relationship, as it demonstrated significant effects across all four measures of brand preference, including brand choice, future purchase intent, monetary allocation, and overall evaluation.' This suggests that even in high search loyalty categories that there may be good reason to expect a Double Jeopardy effect, evidenced by strong relationships such as those described above.

We believe that this brief review provides some important clues about why categories often exhibit strong Double Jeopardy relationships. However, the relationships are not perfect and important insights can be gained from examining the outliers to the basic relationship. Further, we will provide evidence that these deviations can provide insight into how brands grow over time.

SECTION 2: EXPLORING DEVIATIONS FROM CATEGORY RELATIONSHIPS

Deviations from category relationships: An important difference or just wobble?

Ehrenberg accepts that the Double Jeopardy relationship is not perfect in his analysis of behavioral data (Uncles, Ehrenberg and Hammond, 1995), but he argues that deviations within a category are caused by niche brands or statistical wobble (1997). He has also stated that it is not possible to forecast how a brand will change with regard to growth or decline from one year to the next (1997). While we definitely agree that Double Jeopardy affects attitudinal data as well as behavioral, we would have to disagree that deviations are limited to niche brands or simply the result of wobble. While not often observed, some brands do deviate from the established category relationship for prolonged periods of time. Much can be learned from how long that deviation is maintained and from whether or not the brand redresses back to the general category relationship. In addition, we find evidence that deviations from category relationships can be predictive of longer-term success.

What drives deviations?

Point-in-time analysis allows us to identify only that a brand deviates from the category relationship or not. In each case, there are clear reasons why such a deviation might be expected to exist. Table 2 lists some examples taken from the cases summarized in Section 1 above. In all cases, there was a strong category relationship between the measures (r² of 0.85 or better for packaged goods, and 0.75 or better for automotive).

TABLE 2

It is not too difficult to identify potential causes for the observed deviations. Cereal A is a long-established and successful brand with broad appeal (the result of successful line extensions and extensive advertising). Deodorant B has a strong advertised heritage as a value-priced brand and so achieves high levels of trial. Brand C is clearly targeted at a particular subsegment of the frozen-food market and so limits its appeal to those who are limiting their calorie intake. Shampoo D might be an attractive brand to many people but typically sells at a premium price, discouraging trial. In the case of the car examples each is the recognized leader (in sales terms) within its segment, the result of extensive innovation and marketing support.

In the cases highlighted above, there were obvious reasons why individual brands deviated from the category relationship: broad and segmented appeal, value or premium pricing. The existence of these deviations is intuitively sensible and repetition of the analysis over time typically shows little change in the deviation. Sometimes, however, the deviation will alter over time, and this gives us some clues on how brands grow.

Clues on how brands grow: New brands

Changes in deviation are most often observed in the case of new products. The typical pattern can be seen for a new detergent brand in Figure 3 and a new car brand in Figure 4. In these cases, the category relationship is defined by the brands other than the new one over the full period of observation. The new brand is plotted as a supplementary point over discrete time periods.

In both cases the new brands start out well below the category relationship as awareness and familiarity precede trial or checking out. As time goes by an increasing proportion of people adopt the new brand into their consideration set. The detergent brand establishes an awareness to trial conversion that is slightly better than the category relationship within a few months. In the case of the new car, a stronger than normal relationship is established between familiarity and consideration within a few months of the start of advertising. Actual ownership, however, lagged consideration, but as the purchase cycle took effect the brand became one of the best selling in its class. Note that the deviation has held at roughly the same level for several years, by which time the brand could be regarded as well established. The final deviation of a new product from the established-category relationship is determined by a variety of factors, for instance, the general appeal of the brand promise, the relative price point, the relative distribution strength, etc.

An important point to note is that changes in the awareness/trial relationship do not necessarily work through in the same way to other usage measures. For example, a new brand succeeded in raising brand awareness as a result of advertising from 50 to 70%. As a result, trial increased, exactly as predicted by the overall category relationship. Examination of the relationship between claimed trial and recent usage, however, showed a different pattern. As Figure 5 demonstrates, recent usage increased more strongly than the increase in trial alone would have predicted, so that in Period 2 there was a clear positive deviation from the category relationship. This was gradually eroded, and by Period 4 the brand was moving back closer to the expected relationship. We hypothesize that the marketing activity had brought new, perhaps more experimental, buyers into the brand's franchise but failed to keep more than its normal share in the longer-term.

Breaking the status quo by breaking the category relationship

On the basis of evidence like this we hypothesize that brands do not move smoothly along the category relationship but move up through temporarily deviating from it. Two types of deviation are observed:

1. Awareness led (as a result of increased presence due to advertising, distribution change or sponsorship).

2. Loyalty led, as demonstrated by changes in the trial and recent usage relationship (most probably caused by incentives, or some perceived advantage, either rational or emotional).

This is in keeping with Dru's recent book (1996) on disruption that suggests brands only make significant progress within a category by breaking the established status quo. They review some more obvious examples of this type of effect. US examples would be the introduction of Saturn (Hollis and Morrissey, 1997) or the launch of Pepcid AC (Millward Brown International, 1996), but we would argue that this mechanism is likely to happen to differing degrees within all categories. The scale and duration of the deviation may differ from small and temporary to large and long lasting, but in each case the brand grows by essentially breaking the existing category status quo as a result of marketing activity.

Observations of this type of change are rare for established brands, even across Millward Brown's extensive database of brand-tracking results. This is possibly because most deviations happen over shorter time scales and are not that large in absolute terms. Changes of this type may simply go unnoticed or be discounted as random wobble. We are currently implementing a more thorough search for this type of phenomenon and aim to share our findings in 1998. There is, however, good reason to expect that major changes in deviation for established brands will be rare. DeKimpe and Hanssens (1995) found that most categories (78%) were essentially stable in terms of market share. We can assume that this is likely to be true of attitudinal as well as behavioral data. Under conditions of stability brands will simply maintain the existing relationships across both behavioral and attitudinal measures. The marketing implications of this conclusion will be examined later.

The findings reviewed so far indicate that attitudes do have an effect on brand preference. We believe that the findings and hypotheses fit with the findings on Double Jeopardy reported by Ehrenberg. There can be little doubt that under conditions of stability attitudes and behavior are unlikely to change much from one year to the next. The marketing team, however, is usually charged with growing share and sales. They need to identify potential ways in which their brand might break the status quo in order to grow and become stronger. Next we review one survey research tool designed to meet that objective and show how it can be used to anticipate share change.

SECTION 3: USING BRAND DYNAMICS™ TO MEASURE BRAND STRENGTH

Providing insight into attitudinal loyalty

The available evidence suggests that Double Jeopardy can be created by both push and pull mechanisms. However, there is still a need to provide insight and understanding of the two mechanisms and identify how a brand might leverage deviations from the normal category pattern to its advantage. While there are several potential sources of reliable information on behavioral activity, such as panel and scanner data, most survey research commissioned to understand the effects of attitudes on purchase fails to provide a credible prediction of actual behavior. As reviewed in the November/ December 1996 edition of the JAR (Dyson, Farr and Hollis, 1996), this is why Millward Brown set out to create the BrandDynamics™ System. From our viewpoint, understanding the push causes of Double Jeopardy was incidental to our main purpose, but they still needed to be accounted for in the some way. Similarly we needed to recognize that there would be an attitudinal Double Jeopardy effect that would need to be taken into account before we could extract meaningful findings from the data. Let us briefly review how this is done, starting with how the push factors are dealt with.

Predicting behavioral loyalty from survey data alone: The consumer value model

Within the BrandDynamics™ System we recognize that there can be a difference between behavioral loyalty and attitudinal loyalty, with the former being influenced more by the push factors than the pull. The Consumer Value model predicts a Consumer Loyalty score, the probability of an individual's next purchase, based on three main variables:

1. The likelihood that they will consider the brand for purchase

2. The likelihood that they are innately brand loyal or price driven when buying in the category

3. A measure of brand size (simply the percentage of people who claim to have bought the brand last)

We needed to include a measure of brand size as a variable in the Consumer Value model in order to accurately predict behavioral loyalty (defined as the value weighted share of requirements for each brand for each respondent). What this effectively does is to account for the potential push factors that influence consumer choices but which are not adequately recognized by them to be accounted for by their stated brand consideration. Given that brand consideration is by its nature an attitudinal measure we assume that it already incorporates the pull influences of Double Jeopardy such as those reviewed by Rindfleisch and Inman (forthcoming). It is of note that the data for the original model was collected prior to tracking the behavior of the same respondents for a 12-week period. In other words, it is predictive of short-term future behavior. We have also observed that similar attitudinal measures lead share change over time in a variety of categories (Hollis, 1996).

Sixteen months after its introduction the Consumer Value model has now been successfully employed in over 20 categories in 20 different countries. It originated from a model designed to predict measured purchasing behavior derived from a panel, but in subsequent studies we simply aggregate the individual Consumer Loyalties multiplied by their claimed category consumption to predict market share. Typically, the Model explains over 90% of the variation in market share across brands, much better than might be achieved from a simple measure of claimed most-often usage or its equivalent. With the identification of a reliable surrogate for behavioral loyalty from a survey we can then proceed to explain the variations in that loyalty based on peoples' attitudes.

Explaining attitudinal loyalty: The brand pyramid

The attitudinal component of the BrandDynamics™ System is the Brand Pyramid shown in Figure 6. Each level is comprised of standard awareness and image measures as described in our previous paper. The average behavioral loyalty increases at each level. People who are classified as bonded are highly likely to be both attitudinally and behaviorally loyal to a brand (although their status on each is calculated separately).

The Brand Pyramid has built into it two components that relate directly to the influence of brand size. These are saliency and perceived popularity (measures that will be recognized to link directly to the attitudinal reasons for Double Jeopardy reviewed above). By specifically including these components alongside others such as relative brand appeal, whether the brand is thought to be growing in popularity, and relative perceived product performance, we can identify to what degree attitudinal loyalty appears to be driven simply by how big the brand is rather than actual perceptions of the brand that differentiate it from others. Marketers find this useful because it allows them to focus on ways of breaking the attitudinal status quo within a category.

The Pyramid itself is a useful diagnostic tool. It provides a convenient summary of a brand's attitudinal standing, and each layer can be decomposed into its component parts to reveal underlying strengths and weaknesses in the standard measures. These measures can in turn be explained by using driver analysis based on more category-specific survey measures or qualitative research allowing marketers to develop a specific action plan for their brand. However, how can we be certain that altering these awareness and image measures will result in a stronger brand overall? Can the Pyramid provide that elusive means of explaining why brands shift position in relation to the established category patterns? Is there a measure of brand strength that anticipates a strengthening or weakening of market share?

Using the pyramid to determine brand strength

Based on the results reviewed above we might expect that if such a measure of brand strength exists it will reside in the consumer attitudes and beliefs that drive the consumer's predisposition toward brands. However, brands with higher shares tend to have more presence, wider relevance, more generally acceptable performance, greater perceived advantages and higher levels of bonding. So the pyramid itself is in part just a reflection of brand size, just as is any image attribute or as trial is dependent on awareness. What we needed was a measure that would take out the brand size effect to reveal any underlying strength or weakness. For this measure to have any credibility we needed to be able to show that it did indeed relate to future marketplace performance.

Identifying deviations from category relationships within the pyramid

The clearest way of seeing if the brand has higher levels of presence or bonding for its size is to create a conversion profile. This looks at how good the conversion is to a level of the pyramid from the one below-relative to the size of the brand and the performance of other brands in the category. Essentially, it is the same analysis as reviewed above but applied to the percentage of people at a specific level of the Pyramid compared to the one below, eg, Relevance versus Presence or Bonding versus Advantage. Figure 7 shows an example from the UK automotive category.

The analysis reveals the underlying category relationship and the deviations from it. Brand A has a lower conversion to Bonding than we would expect purely from the proportion of people who reached the Advantage level. This deviation is driven by poor product-performance expectations. Brand B has a higher conversion than would be expected which is the result of stronger emotional appeal and better overall opinion. It is of note that fundamental segmenting factors, like product formulation and price point, have most effect on deviations at Relevance. The deviations at Bonding are typically based on perceptions that the brand is better or worse than its competitors in some way, eg, product performance, emotional appeal or popularity. Deviations at this level can be just as strong as those observed at lower levels of the Pyramid.

Based on the results from the initial BrandDynamics™ studies, three main types of pyramid profile were identified (see Figure 8). The profile on the left shows the brand to have relatively high Presence and Relevance but relatively low Advantage and Bonding. In contrast the middle profile shows the brand to be relatively low at Presence and Relevance but once past these hurdles to have a relatively high conversion to Performance, Advantage, and Bonding. The third profile is one where the brand is positive all the way up the pyramid-particularly at Advantage and Bonding.

We felt this pattern to be significant because brands with the left-handed profile tended to be brands which had been declining in share terms, brands in the middle tended to be new (and hence gaining share) or niche brands (often able to command a premium), and those with right-hand profile tended to be brand leaders. It made intuitive sense that brands with relative strength toward the top of the pyramid, and hence a stronger relationship with their consumers, would over time fare better in the marketplace. On the basis of this logic, and our previous experience that awareness and usage deviations were often predictive of brand success, we felt confident that the shape of the pyramid profile was a measure of brand strength. However, we still had to prove it.

Calculating brand strength

To arrive at a measure of strength we needed to find a way of quantifying the shape of the profile. The approach which ultimately produced the best correlation with marketplace performance was to multiply the conversion profile deviation at each level of the Pyramid by the average Consumer Loyalty of that brand's consumers at the respective level. Brand Strength is the sum of each of these multiplications for each brand. Given that the Consumer Loyalty score is strongest at the top of the Pyramid, deviations at Advantage and Bonding will have the most effect on a brand's Strength score. These are the levels that focus less on the fundamentals of relevance and product performance and more on emotional appeal and popularity.

Validating brand strength against share change

To prove that Brand Strength has marketplace relevance we collected value-share data for all the brands where we had conducted a BrandDynamics™ study more than a year ago. This amounts to 74 brands comprising the original calibrations from the United Kingdom and United States, plus some live studies conducted early in 1996. Clearly as we collect more data from live studies we will be able to build more data into the analysis from more categories and countries. Similarly, we will also be able to look over a longer time frame.

Figure 9 shows the correlation between the Brand Strength statistic and the change in Value Share for these brands. The change in share represents the Value Share in the year after the BrandDynamics™ analysis minus the year preceding it.

Is this a good correlation? We believe that it is given that the analysis does not take account of all the push factors that will also affect brand performance, such as distribution, promotions, competitive launches and advertising, etc. These too will have an influence on market share. The correlation above shows a clear underlying pattern that implies that a brand that has strong attitudinal loyalty is likely to grow over time.

The importance of this correlation is underlined if we contrast it with a similar analysis against brand size for the same 74 brands (brand size was defined as Value Share for the year before the BrandDynamics™ study was conducted). The result of this comparison is shown in Figure 10. Here you can see that there is a much weaker relationship. This suggests that Ehrenberg is not entirely correct – there are stronger and weaker brands, and it is possible to identify which are which, and which are likely to grow. An additional analysis combining size and strength helps to clarify this point further.

The interaction of brand size and brand strength

First we divided the brands under consideration into two groups based on size – market share above 5% versus below 5%. Then we divided each of these groups into stronger and weaker brands-strength above or below the mean strength – creating four groups:

1. bigger but weaker

2. bigger and stronger

3. smaller and weaker

4. smaller but stronger

The first finding is that very few of the smaller brands are stronger by our definition. So Ehrenberg is partially right. However, of the bigger brands there is an even split between stronger and weaker brands. (See Table 3.)

TABLE 3: BRANDS CATEGORIZED BY SIZE AND STRENGTH

If we then look at the brands in each group to see how many declined (lost more than 0.5% Value Share), stayed flat, or grew (gained more than 0.5% Value Share) a clear pattern emerges. (See Table 4.)

TABLE 4: CHANGE IN VALUE SHARE

Among the bigger brands more than half the weaker brands lost share (8 out of 15) and only 3 grew. Of the stronger bigger brands only 1 declined and more than half grew (8 out of 14). This is a very powerful finding. It shows that the brand attributes that give the bigger stronger brands greater perceived Advantage and Bonding do have value in the marketplace. Also that a lack of consumer equity for the bigger but weaker brands leaves them in a vulnerable position.

For the smaller brands very few are particularly strong. However, it is also clear that very few of the smaller brands were growing rapidly.

Is strength a guarantee of share change?

On average, bigger stronger brands did fare better than others over time; however, there were examples of weaker brands gaining share and stronger brands losing it. The following are two cases where share change did not reflect initial Brand Strength and which point to the importance of continuing to support your brand through above-the-line spending.

• Clairol Herbal Essence shampoo was one of the brands included in our initial US pilot study. Its Strength score at the time was mediocre but its Value Share grew in the year following the survey. In the intervening period Clairol had launched new advertising. The ads, which spoofed the 'When Harry Met Sally' orgasm scene, drew attention to the product's fragrance benefit in a new and memorable way. Ad awareness responded strongly. The advertising share of mind, which we have found to be indicative of change (Hollis, 1997), doubled. The status quo had been changed and the brand grew.

• A less positive example is the history of Arm and Hammer toothpaste in the United Kingdom. Of the 74 brands reviewed above Arm and Hammer was the one smaller stronger brand that declined. In 1995 this brand was identified as strong but since then received no advertising support. At the same time it was squeezed by competitive 'Bicarbonate of Soda' launches from the more established players in the market including the brand leader, Colgate. In this case, the established brands successfully moved to counteract the threat before the newcomer could solidify its position.

This confirms that Brand Strength is a measure of vulnerability and potential strength – not a guarantee of change. Many other factors will affect share over time: innovations, pricing, new advertising and changes in distribution to name a few. However, a strong brand is more likely to gain share from its own marketing actions and resist the actions of competitors and hence maintain share. A weak brand can still grow but will have to work harder to stand still, and over time be more vulnerable from the actions of other brands – not least those of the own label or price brands.

MARKETING IMPLICATIONS

The findings above have important implications for all of us involved in marketing brands, even if it is only to remind us of some important facts of life.

1. If you are responsible for an established brand, do not be too optimistic that your marketing activity will change the standing of your brand in the short term. There are many forces that work to obviate change, both market and consumer driven. Much marketing activity serves to maintain the status quo in the face of competition rather than change it. This needs to be better recognized as a positive end in itself, since, as we have seen, brands can decline if existing consumer equities are not supported.

2. Major changes in the status quo will usually result from some form of disruption. This usually takes the form of innovation of one form or another, for example, price segmentation, new creative strategy, repositioning, new product formulation, etc. Saturn's entry into the small car market was a classic example. Careful research and planning identified drivers' needs for a friendlier buying experience and a company that stood behind its product. The result was not only a successful launch but a basic change in the way the category was viewed.

3. It is easier for a new brand to disrupt a category than for an established one for three basic reasons. First, that there is a consumer 'straight jacket' based on existing perceptions of established brands. Radical changes are just as likely to disenfranchise existing buyers as bring in new ones. Second, competitors are unlikely to ignore innovations made by an established brand and will try to match them. They may be less threatened by completely new brands which take time to create a presence. Third (and linked to the first), the marketing teams of established brands (and their senior management) may simply be hesitant to deviate from the existing successful formula and experiment with new approaches.

4. Although new brands will find it easier to disrupt a category, once the initial impetus is lost the new brand is likely to be locked into the new category relationship in the same way as its predecessors were locked into the old. In other words, the launch is a one-shot opportunity to establish the brand's position in the marketplace. It is critical that it be successful. This supports the typical Procter and Gamble strategy of introducing a new product – identification of a new relevant and differentiating benefit, followed by heavy weights of advertising to establish that differentiation as a category advantage, at the same time maximizing penetration.

5. There is an important corollary to the previous point. Once disruption occurs, established brands must be prepared to respond quickly. In our examination of case-history material, established brands often decline simply because the competitive context changes, not because they make active marketing errors. A failure to match or defend against new claims can be fatal.

6. Consumer perceptions do matter but only if they create a strong predisposition toward the brand. We would identify the following broad ways in which a brand can create a strong attitudinal bond with the consumer:

• Stronger relevance to people's needs than other brands

• Better perceived product performance

• Stronger emotional appeal

• Stronger saliency

• Stronger perceived popularity – either in terms of growth or in absolute terms

• Better perceived pricing

Obviously, there are a multitude of ways in which brands can create these differentiating perceptions, and a brand might be strong on several counts. However, we would suggest it is important to identify which approach offers most leverage for your brand. This will differ from brand to brand, category to category and country to country. Strong and consistent brand communication is then necessary to turn these basic strengths to advantage.

7. Creating a strong attitudinal bond with your consumer is fundamental to your brand's long-term success. The creation of this bond is not a guarantee of growth, but a brand that fails to maintain this equity will be more vulnerable to competitive actions than otherwise. Equally, attitudinal strength alone is not enough to ensure success. Both push and pull mechanisms will affect a brand's standing over time, and the marketing team must successfully leverage both to ensure its brand's longer-term health.

OVERALL CONCLUSION

The findings reviewed here provide evidence to back up both Larry Light and Andrew Ehrenberg. Brands typically do adhere to the Double Jeopardy relationship under conditions of stability. However, it is possible for brands to create a stable and potentially long-lasting deviation from that underlying relationship. This can be the result of either structural market advantages or attitudinally based consumer equities. Survey research tools like the BrandDynamics™ System provide the means to gain insight into both the strength of a brand and what to do to make it stronger. Attitudinally based measures like Brand Strength have a positive correlation with changes in future Value Share. It is clear evidence that brands which consumers feel to have either rational, emotional, or saliency based advantages are likely to be worth more to their owners in the future than ones which are bought purely on the basis of availability or pricing. Overall, this provides continued evidence for the value of long-term brand-building activity. In other words, you can help your brand to grow up big and strong, you just need to understand how.

REFERENCES

Brown, Gordon 'Findings From Brand Attitude Grids.' In Transcript of the Market Research Society Conference, 1986.

DeKimpe, Marnick, and Dominique Hanssens, 'Empirical Generalizations About Market Evolution and Stationarity.' Marketing Science 14, 3, Part 2 (1995).

Dru, Jean-Marie. Disruption: Overturning Conventions and Shaking up the Marketplace. New York: John Wiley and Sons, 1996.

Dyson, Paul, Andy Farr. 'Understanding, Measuring, and Using Brand Equity.' Journal of Advertising Research 36, 6 (1996): 9-21.

Ehrenberg, Andrew. 'B.E. Or Not B.E.' In The eXperts Report on the 1997 43rd Annual Advertising Research Foundation Conference. New York: Advertising Research Foundation, 1997.

–. 'Toward an Integrated Theory of Consumer Behavior.' Journal of the Market Research Society 11, 4 (1969).

–. 'In Search of Holy Grails: Two Comments.' Journal of Advertising Research 37, 1 (1997): 9-12.

Hollis, Nigel, and Steve Morrisey. 'Brands in the Driver's Seat: The Saturn Case Study.' In The eXperts Report on the 1997 43rd Annual Advertising Research Foundation Conference. New York: Advertising Research Foundation, 1997.

–. 'They Said My Brand Was Popular. So What?' In Transcript Proceedings of the ARF Advertising and Brand Tracking Workshop. New York: Advertising Research Foundation, 1996.

–. 'Looking Longer-Term: How Best To Identify Longer-Term Advertising Effects.' In The eXperts report on the ARF Return on Marketing Investment Workshop. New York: Advertising Research Foundation, 1997.

Lal, Rajiv, and V. Padmanabahn. 'Competitive Response and Equilibria.' Marketing Science 14, 3, Part 2 (1995).

Laurent, Gilles, Jean-Noel Kapferer, and Francoise Roussel. 'The Underlying Structure of Brand Awareness Scores.' Marketing Science 14, 3, Part 2 (1995).

Light, Larry. 'The Trustmarketer's Road to Enduring Profitable Growth.' New York: Coalition For Brand Equity, 1993.

McPhee, W. 'Formal Theories of Mass Behavior.' New York: New York Press, 1963.

Millward Brown International. 'The Good Health Guide For Brands.' Papers from the 1996 Millward Brown International Client Conference, New York.

Reibstein, David, and Paul Farris. 'Market Share and Distribution: A Generalization, A Speculation and Some Implications.' Marketing Science 14, 3, Part 2 (1995).

Rindfleisch, Aric, and Jerry Inman. 'Explaining the Familiarity-Liking Relationship: Mere Exposure, Information Availability, or Social Desirability?' Marketing Letters 9, 1 (1998): forthcoming.

Uncles, Mark, Andrew Ehrenberg, and Kathy Hammond. 'Patterns of Buyer Behavior: Regularities, Models and Extensions.' Marketing Science 14, 3, Part 2 (1995).

http://www.warc.com

© Advertising Research Foundation