«H OW D O F I N E W I N E B R A N D S G ROW ? Byron Sharp*, Larry Lockshin*, and Justin Cohen** *Ehrenberg-Bass Institute for Marketing Science, ...»
H OW D O F I N E W I N E B R A N D S G ROW ?
Byron Sharp*, Larry Lockshin*, and Justin Cohen**
*Ehrenberg-Bass Institute for Marketing Science,
University of South Australia,
P.O. Box 2470, Adelaide SA 5000, Australia
Email: Byron.Sharo@marketingScience.info, Larry.Lockshin@MarketingScience.info
**L’UNAM - Groupe ESA - Ecole Supérieure d'Agriculture d'Angers,
55 Rue Rabelais, BP 30748, 49007, Angers Cedex 01, France Email: firstname.lastname@example.org Paper for the 6th Academy of Wine Business Research (AWBR) Conference, Bordeaux, France,
2011. Dr Hervé Remaud, chair.
Abstract Consumer behaviour is often portrayed as being so complex as to be unpredictable, and hence marketing is seen as closer to a black art than a practice informed by science. Yet, this image is based on ignorance of the decades of work by marketing scientists; work that has uncovered law-like patterns in buying behaviour. These are scientiﬁc laws in the classic sense; reoccurring patterns which therefore give the power of prediction. and explanation These empirical laws provide benchmarks for marketers, but they also give deep insight into how buyers buy, how brands compete for custom, and how marketing works.
The question that this paper investigates is whether or not such laws apply to the buying and selling of wine. There are many reasons for thinking that the market for wine might be different, however we ﬁnd that it exhibits the same empirical law-like pattern (the Double Jeopardy law) seen in other competitive markets. The implications are profound, and sweep away some widely held beliefs about the mystique of wine marketing.
The Start of a Scientiﬁc Revolution?
In this paper we report some serious science, not complex and opaque statistical modelling, but empirical scientiﬁc laws that generalise across a wide range of known conditions, and so provide trustworthy predictions. Such empirical generalisations are the building blocks of science (Bass 2001).
Science has revolutionised every discipline it has touched; now it is marketing’s turn!
- Joseph Tripodi, Chief Marketing and Commercial Ofﬁcer, Executive Vice-President, The Coca-Cola Company, Atlanta, USA (B. Sharp, 2010) Marketing practice is documented in marketing textbooks, along with advice for how to practice effective marketing. This advice has remained much the same for 50 years or so, and textbooks do not contradict one another... competition for market share is said to require creating differentiated brands that carve off sections of market share by addressing the heterogeneity in demand, thus competing brands sell to different types of customers; brands have substantially different images; and brand loyalty varies considerably and idiosyncratically depending on each brand’s positioning. This view suggests that many brands are niche brands, with small but unusually loyal customer bases; and growth is possible by becoming more tightly positioned; becoming more niche and selling more to the brand’s most loyal customers. Marketing strategy is portrayed as being largely concerned with segmentation, targeting and differentiating brands so that they may command price premiums and passionate loyalty.
It would be rather difﬁcult to ﬁnd a textbook or marketing consultant that did not offer this world view. However (B. Sharp, 2010) presents ten well-established empirical laws that contradict these widely held beliefs. Sharp states that the traditional textbook view of how brands grow is not entirely wrong, it just describes a very limited part of marketing and it fails the most basic test of scientiﬁc theories – it does not lead to the known empirical laws and in some cases directly counters the laws (i.e. it predicts different patterns). Instead, building on the ten scientiﬁc laws, Sharp argues that market share growth is largely dependent on building mental and physical availability so that brands that are easier to buy for more people on more occasions, get bought more often. Thus, the role of differentiation is vastly downplayed, while the importance of distinctive branding, and wide reaching advertising and distribution is massively up-played. Brands compete largely as if they were undifferentiated options seeking the same customers (even though they are all slightly differentiated).
He illustrates how markets work with the example of McDonalds, Pizza Hut and KFC, which although very functionally different, compete largely head-on as brands ﬁghting for attention with market share that relate to their respective successes in building mental and physical availability.
But is wine different?
Perhaps it should not be surprising that the ﬁrst scientiﬁc laws in marketing should conﬂict with theory developed in the absence of such laws. And science has a long track record of presenting fundamental discoveries that upset existing theory (e.g. Kepler’s laws of planetary motion).
Nevertheless the empirical laws have been described many times as revolutionary – AdNews melodramatically wrote that Sharp’s book “destroys just about every point of marketing convention with hard research”. The laws have been documented across hundreds of product categories, retail stores, consumer and industrial markets, different countries, and over decades. But other than cars there has been no testing in luxury, symbolic categories like wine. Also wine is a product category that is constructed of a wider range of attributes with more levels (options to choose from) than most (typical product categories are comprised of only a few hundred varieties (Kennedy, Ehrenberg, & Long, 2000)). The wine category has thousands of different offerings which is a possible argument for establishing wine as a boundary condition for the empirical laws that have largely been observed in more prosaic, less symbolic, less hedonic product categories.
The Double Jeopardy Law In this research we examine just one, probably the most famous, scientiﬁc law in marketing. It is known as ‘the Double Jeopardy Law’ and it reveals a great deal about how brands grow. It was ﬁrst discovered by a sociologist William McPhee in attitudes towards movie stars, reading of comic strips, and listening to radio DJs (McPhee, 1963). Professors Andrew Ehrenberg and Gerald Goodhardt did much testing of the generalisability of the law in regards to attitudinal and buying metrics for consumer goods brands, and television programs (‘from soap to soap operas’), such that Double Jeopardy is now considered extremely well established (Ehrenberg, Goodhardt, & Barwise, 1990).
The Double Jeopardy Law says that more popular items within a competitive set (e.g. rival brands in a product category) will be bought by many more people who will show slightly higher loyalty to the brand. Or, expressed in the negative, brands with less market share have far fewer customers and slightly lower brand loyalty. This is illustrated in the following table of Nielsen data on US shampoo
brands (table 2.4, page 21 from Sharp 2010):
Table 1: Marketing metrics for shampoo brands Double Jeopardy law - shampoo (USA 2005)
At ﬁrst glance the empirical law is surprising from a number of angles. McPhee, who was largely looking at attitudes not behaviours, thought it counter intuitive that the proportion of people who liked a particular movie star (of those who knew of the star) should depend on how many other people did not know of the movie star. Later it was realised that the Double Jeopardy Law could be explained as a statistical selection effect – less popular movie stars tended to be known by people who also knew of the more famous movie stars, when these people assigned their attitudes they split their votes (giving some to the more famous stars), meaning that less famous movie stars had to receive slightly lower liking scores. But it was still surprising that there weren’t more niche brands, with small customers bases who were highly loyal. The implication is that brands are less differentiated than expected, and sell to very similar sorts of consumers as do rival brands - this was later conﬁrmed empirically (Kennedy, et al., 2000; Romaniuk, Sharp, & Ehrenberg, 2007). It was also surprising that marketing metrics, particularly those concerning brand loyalty, were predictable - and depended simply on a brand’s market share not on its positioning or marketing strategy. Many marketing plans were, and perhaps still are, based on an assumption that dramatic increases in a brand’s brand loyalty are possible. If loyalty depends on the particular marketing strategy then we might expect large differences between brands, and no relationship with market share. Whereas the Double Jeopardy Law says that rival brands do not vary greatly in brand loyalty, and what variation there is is due to large differences in market share.
The Double Jeopardy Law has profound implications. It says that for a brand to grow its market share it must substantially increase the size of its customer base, which means reaching and inﬂuencing many light, occasional and non buyers. It also suggests that a niche strategy, of concentrating on winning more loyalty from existing heavy loyal buyers of the brand is unlikely to be successful at delivering growth, nor will targeting a select segment of the market. If these strategies were routinely successful then the Double Jeopardy Law would not exist in so many markets, and countries (from Australia to China to France to the USA).
Are there really niche wine brands and regions?
It is commonplace to talk about niche wine brands, and niche regions. But what does niche mean and how can we tell if a brand is niched? Of course, often the term is simply used to mean small but this is incorrect and quite misleading, the word small works perfectly well by itself. In marketing theory the term niche means a brand that barely, if at all, competes in the broad market against the other brands. Instead it has its own segment of buyers for whom it satisﬁes most or all of their needs, that is they are highly loyal to this brand. This shows up in Double Jeopardy statistics as a brand with low market share or penetration but higher frequency of purchase and higher share of category requirements.
In a market that conforms to the Double Jeopardy Law there are no niche brands. In both behavioural and attitudinal metrics there are simply larger and smaller brands. Very small brands do not enjoy high loyalty, they actually have lower brand loyalty than larger brands. And two brands of equal market share will have extremely similar penetration and extremely similar loyalty metrics. This pattern, the Double Jeopardy Law, makes it very easy to spot brands that are even slightly niched.
And this is our scientiﬁc approach in this paper.
In wine markets, as discussed, there are many brands, and they can be functionally very different, with different image positioning, coming from different regions and countries, with different heritage, selling at very different prices. This leads to the important empirical question, does the Double Jeopardy Law apply at all? And if it does, generally, might there not still be a number of niche brands?
We answer these questions with data concerning the upper end of the retail wine market, i.e.
above the level of discounted supermarket lines where a more limited range of brands and price points dominate. Here there are many brands, many regions, many price points, and many grape varieties. Our data set comes from the wine club of a national Australian retailer, so it describes the purchases of regular wine shoppers who are interested in wine. The data spans a three-year period from October 1, 2000 through September 30, 2003. The data consists of purchase records of a sample (n=4,768) of wine buyers, each buyer had to have made a purchase at least once in each year of the three-year period to be included (i.e. customers who left or joined during the period were excluded).
These data concern buying within a single retail chain, so it may potentially slightly overstate loyalty, if consumers tend to buy a particular brand or variety in that store and different brands in other stores.
The following table (table 2) presents performance metrics by wine brand. Because of space constraints only a representative group is included; the full analysis included many more brands but showed the same patterns. Figures are appropriately rounded for clarity.
Market share for each brand is presented in rank order, alongside the two metrics that together determine those sales – penetration (how many customers bought the brand at least once in the period) and average purchase frequency (how often those customers bought the brand). The Double Jeopardy Law can easily be seen in the table, brands with less market share have lower penetration and also slightly lower purchase frequency (i.e. lower loyalty).
A second loyalty metric is also presented; SCR refers to ‘share of category requirements’, that is of all the wine purchases that buyers of that brand made, what proportion were devoted to that brand. All the SCR metrics are below 10% which means that the buyers of each of these brands, on average, bought them only once out of every 10+ purchases. In other words, the average customer of each brand bought other brands far more often. This apparently low loyalty is quite normal. SCR has been shown to decline over longer analysis periods; which means that as buyers make more purchases their repertoire size increases. Also given the huge number of brands in the wine market we would expect consumers to have quite large repertoires (and hence each brand gains a low SCR).
Not reported here is the proportion of each brand’s customers who were 100% loyal over the period, meaning that they bought no other brand (at least within that retail chain). These scores were typically less than 1%, that is 100%-loyalty was extremely rare – hardly anyone buys just one brand of wine. Again these scores followed the Double Jeopardy law.
Smaller wine brands have fewer buyers and also slightly lower brand loyalty.