How Brands Grow

by Byron Sharp

  31 Fortunately for marketers, most targeting attempts fail because media are far less targeted than they claim (e.g. see Sharp et al 2009). The danger is that in the future truly targeted media strategies may become available, then there is a real potential for marketers to do damage to their brands through following textbook targeting recommendations.

  32 However, Gucci has many more buyers, far greater sales revenue and lots more stores (see Chapter 12).

  33 Researchers newly recruited to the Ehrenberg-Bass Institute, and marketing executives from our corporate sponsors, regularly report that this finding is very surprising. The conference paper by Kennedy, Ehrenberg & Long (2000) won two awards for the implications of this discovery to marketing practice.

  34 Of course, the actual number of customers shared with Coca-Cola depends on the size of the brand and on how many customers they have to share.

  35 If they are complementary products (e.g. taco shells and refried beans) then we'd expect high levels of sharing. But complementary products are easy to identify prior to any customer-sharing analysis. Brands within a product/service category are nearly always rivals not complementary. They share customers because they compete with one another as alternatives.

  36 Also sometimes called a brand-switching table, but the term switching is clumsy, as it implies defection from one brand and uptake of a new (to the buyer) brand, which is an exaggeration. Just because I bought Fanta this time when I more often buy Coke doesn't mean anything has changed about my Coca-Cola buying; I just occasionally buy Fanta. Duplication of purchase tables reflect how people hold repertoires of brands, i.e. divided polygamous loyalty.

  37 Some brands might successfully target a particular usage or occasion rather than type of person, for example, ice-creams to eat at the movies. This should be expected to produce demographic differences in customer bases, and deviations from the duplication of purchase law. The fact that it usually does not result in such deviations suggests that this targeting usually doesn’t work.

  38 Consumers often behave as if they haven't read the marketing plans for the brands they buy.

  39 We saw in Chapter 5 that even brand variants (SKUs) largely sell to similar customer bases.

  40 All these brands have benefited from the growth of the sports wear category; this growth was due to sports wear becoming street wear. Stellar success in marketing is often due to good luck and clever ability to ‘ride the wave’.

  41 It is often said that this experiment is based on the Pepsi challenge. However, the famous Pepsi challenge consists of two unlabelled cups, one containing Pepsi and one Coke, which differs from this experiment where both cups contain the same cola. With both cups containing the same cola, respondents obviously found it hard to detect a difference and so were undoubtably more easily swayed by the label. I repeat that it is amazing that in the 21st Century such shoddy research studies are still being done and being accepted by respected journals.

  42 I use the term 'favourites' with some caution as it implies that buyers have strong attitudinal references, which often simply isn't the case. I'm using this term to refer to the fact that we have brands that we favour (i.e our repeat purchasing is biased (non-random) towards these brands).

  43 Remember if a brand has 100% share then all its buyers must be 100% loyal. This loyalty metric (percentage of 100% loyal buyers in the customer base) rises with market share, like all other measures of loyalty.

  44 This is an extreme example because of the extraordinary market share of Heinz, which translates to large differences in metrics between it and other brands. Normally the natural monopoly pattern is mild, particularly if the analyst is only looking at the top dozen brands. There are other categories, like television programs, where there are incredible differences in market share, and so the natural monopoly pattern is quite marked. There are some television programs that are watched by hardly anyone compared to popular shows, and these tiny programs have a viewing base substantially skewed to the heaviest of television viewers (i.e. people who watch almost everything). In 2013 my September column in Admap discusses this and implications for other media.

  45 There are more exotic stories of passionate loyalty but these seldom come from observing buyers, and never from observing large number of buyers repeatedly buying. Instead, these myths originate from interrogating small samples of biased customers, or personal introspection.

  46 There is a long (too long!) history of marketing writers debating what ‘true loyalty’ is. This is a perfect example of what the twentieth century's most famous philosopher of science, Sir Karl Popper, called essentialism: seeking to define the essence of an abstract theoretical concept (Esslemont & Wright, 1994). We can forever debate the meaning of abstract nouns (like, What is true love? What is marketing?), but as these questions are simply about the definitions we currently might decide to use there is no logical way of ever resolving them. To suggest that one approach captures the true meaning of brand loyalty, while another (by implication) does not, is bad philosophy, and bad science. Contrary to popular belief the purpose of science is not to say what things are but rather to say things about things: how they behave, how they relate to other things. Physicists can tell you a great deal about the properties and behaviour of things like gravity and mass while there is no rigid definition of what these things ‘truly’ are. Hence, this book is about what we can say about real world loyalty-type behaviour, both verbal behaviour (expressed attitudes) and overt behaviour (buying).

  “Never let yourself be goaded into taking seriously problems about words and their meanings. What must be taken seriously are questions of fact, and assertions about facts; theories and hypotheses; the problems they solve; and the problems they raise” (Popper 1976).

  47 Saatchi and Saatchi's Kevin Roberts says he came up with his Lovemarks idea very late one evening after a couple of bottles of Bordeaux red wine. He claims to have later done research to 'prove' the idea! I’ve yet been unable to find this research.

  48 This is not to say that emotions play an unimportant role in decision-making and viewing of advertising. The vital role of emotion in decision-making and processing of information is now well established. But it is a huge unrealistic leap to say that this means consumers are, or must be, passionately attached to brands. Most emotions are not particularly heated – indeed neuroscientists say that most emotional responses are so slight that we fail to even feel them. And brain scans suggest that our brains react to objects in different ways to how they react to people. Love is a many-splendoured thing, but not a big part of brand buying.

  49 This law-like pattern has now been reported and replicated many times in the academic scientific literature, including several PhD theses. These cover many product categories, B2B and consumer services, retail stores, cultural events, TV programs, government payments, and tourism destinations. The law-like pattern has also been shown for different question types, and in the US, Europe, Asia and Australia; for example, Castleberry et al. 1994; Dall'Olmo Riley et al. 1997; Sharp et al. 2002; Sharp & Winchester 2002; Rungie et al. 2005; Dolnicar and Rossiter 2008.

  50 Anne Sharp (2002) showed that the repeat rate is a bit higher than expected if university students are the respondents (i.e. probably due to experiment artificiality). Dolnicar and Rossiter (2008) were unpleasantly surprised by the discovery of these low repeat rates and sought to dismiss the phenomenon as a methodological artifact. They attempted a variety of method variations on surveys of university student respondents and reported a minuscule increase in stability of response (repeat-rate) above usually observed.

  51 Amazingly some consumer behaviour textbooks still today, in spite of decades

  of counter evidence, say that attitudes (positive or negative evaluations about something) are enduring and situation a-specific. If we define attitudes like this then we hold very few attitudes indeed!

  52 My young daughter has many favourites, many most favourite toys, many most favourite sweets, a number of best frien
ds. We adults know not to make such illogical statements, or at least we try. However, I expect that my daughter’s innocent language illustrates well our human tendency to over-estimate our commitment.

  53 According to Morgan Spurlock, director of the movie “Super Size Me”, Don Gorske stores the boxes from all the Big Macs he has eaten in his attic. Wikipedia also reports that Don proposed to his wife in his favourite McDonald's car park. See http://en.wikipedia.org/wiki/Don_Gorske.

  54 If the Ford car that you'd just chosen and paid for (but not insured) had been stolen, but a kind millionaire, hearing your story, sent you a new similar model Chrysler – would you mind in any way? A few (ungrateful) people might wish that they'd been sent a Ford (there are all types in this world), but I suspect not many.

  55 The Associated Press 2008.

  56 The answers to these two questions are: 1) Largely no, this isn't the main effect of advertising; 2) No.

  57 These perceptual cues are developed with little prior thought about how these cues should be interpreted in relation to one another; for example, what does it mean if a brand scores well on being 'a good business partner' but scores less well on 'understands our needs'? Consequently, a great degree of effort goes into untestable post-hoc explanations of what a subtle shift in brand image might mean.

  58 Generally image scores reflect the law of proto-typicality, in that attributes that describe the product category well (e.g. 'lends money' for banking) score higher than less prototypical attributes (Romaniuk & Sharp 2000). Here the attributes are abstract, so this pattern is more difficult to discern. Abstract attributes like 'trust' have been shown to reflect past usage (e.g. see Romaniuk & Bogomolova 2005).

  59 There is an interesting story of accidental scientific discovery behind this publication. When Dr Jenni Romaniuk received a set of ‘Brand Asset Valuator’ perceptual data, kindly provided by Y&R, she noticed that the same attributes were used in every product category including some that wouldn’t be relevant (e.g. ‘healthy’ for automobiles). To save work for her assistant in preparing the data for analysis she asked her to put aside any image attribute that failed to be mentioned by more than 10% of a brand’s users. When she looked at the list of rejected attributes it reminded her of Aaker’s (1997) brand personality measurement bank. She checked and yes the overlap was astonishing. This discovery was the inspiration for her report for Institute sponsors and subsequent journal article (Romaniuk & Ehrenberg 2012) showing that people don’t consider brands to have human personality traits. For instance very few people (around 3%) in Britain consider their condiment brand to be 'charming'.

  60 See the results of Tucker's experiment described in Chapter 7. Also chapter 2 of Sharp 2013.

  61 The Dirichlet even correctly predicts the slight asymmetry in competition between large and small brands observed in empirical work on price changes (Kamakura & Russell 1989; Scriven & Ehrenberg 2004); in doing so it explains this phenomenon as simply a statistical selection effect, not due to differences in perceived differentiation. A small brand is more affected by actions of a large brand than a large brand is affected by actions of a small brand. For example, it is likely that wherever the small brand is stocked so is the big brand, but not vice-versa. So if the big brand runs a promotion it will affect the small brand in all the locations that the small brand is sold, whereas the small brand's promotions will affect the big brand only in those places where it shares distribution with the small brand.

  62 It must be noted that within these expensive/luxury sub-categories the rival brands compete as if there is little differentiation.

  63 This concurs with research that shows that perceptual surveys employing either rating scales, rankings or pick-any all produce similar findings (Barnard & Ehrenberg 1990; Driesener & Romaniuk 2006).

  64 When advertising agency Young & Rubicam undertook this survey Apple computers generally ran MacOS 9 and did not yet have the easy ability to run Windows. Other computer brands generally ran Windows 95. Apple was marketing coloured iMacs while competitors' computers were beige or, very occasionally, black.

  65 Apple's legendary loyalty also appears to be an exaggeration; see Chapter 7.

  66 Interestingly, out of pizza, burgers and chicken, chicken is generally the most popular. Yet McDonald's is the largest company – its sales are more than double Pizza Hut's and KFC's sales combined. In most markets McDonald's also has a very substantial mental and physical availability advantage over these two rivals (see Chapter 12).

  67 The Ehrenberg-Bass Institute has conducted R&D into how to perform these measures. And how to present them on a Distinctive Asset Grid (Romaniuk & Hartnett 2010). This has now been adopted by a number of market research agencies sometimes without acknowledging their source (how impolite!), more importantly mistakes are made. We’d rather everyone got it right, if you have questions please ask the Institute.

  68 The lack of apparent sales responsiveness to advertising has been observed by most astute marketers for decades. It is also well documented in academic research on changes in advertising weight (e.g. Hu et al. 2009; Lodish et al. 1995) and advertising elasticities (e.g. Tellis 2009). The implication that advertising has almost no sales effect is a common misconception arising from this research.

  69 Advertising with very strong 'recency' effects – this is in categories where buyers show very little interest until they actively start searching; for example, mortgages, insurance, furniture, travel agents, computers, perhaps cars, and certainly Persian rugs. In these categories, when consumers are not ready to buy they screen out much advertising; but, shortly before they do buy, they become dramatically more receptive to advertising. Thus advertising in these categories has a more immediate effect. An equally good way of describing this is to say that it has less of a longer term effect (little effect on buyers not ready to buy yet). However, even in these categories there is occasionally advertising that does break through and reach people who are not yet ready to buy. So it would be wrong to judge the sales effectiveness of this advertising purely according to how much sales increased that week – its sales effects are spread out way into the future when the exposed buyers finally buy from the category. Such advertising can be particularly sales effective, yet it may be no better than average at affecting a week's sales figures. It's the sort of advertising that is entertaining enough to grab people's attention who are not yet 'in the market'; for example, Apple's “I'm a Mac, I'm a PC” and HSBC's “different perspectives” advertisements.

  70 Of course, other people have been hit by competitor advertising and other influences, so this powerful advertising can still not show up in aggregate sales shifts.

  71 Imagine your advertising effectiveness was unchanged but you could concentrate it in time, by arranging for your advertising to only hit people that were going to buy from the category that week. So the ad’s effect on their purchasing probability (ie nudging them towards your brand) would show up immediately. The advertising would look so much more effective, great ROI. Looks can be deceiving, it only looks great because the effect is concentrated not spread out over time. Today there are people promoting this sort of behavioural targeting and claiming that it produces higher ROI/elasticity but this is likely to be just due to this concentration effect. Having your sales effect spread out over a month or a quarter makes no difference in overall effectiveness.

  72 Also the effect is easy to see because consumers' change in buying propensity is quite large, because we are spending quite a lot per consumer, whereas advertising spends very little per consumer, so its effect is a gentler nudge.

  73 Indeed some memories created by advertising last forever. Their salience can fade, but they are always there and able to be refreshed. You'll probably never forget that the Golden Arches mean McDonald's, that Google is a search engine and so on. This is why advertising after a long hiatus can show dramatic effects as it refreshes old memories.

  74 Some academic writers, particularly in the US, think the word ‘persuas
ion’ means sales effectiveness. This is confusing and leads to the statement that “only persuasive advertising can generate sales” becoming tautological.

  75 Authors such as Reichheld and Sasser (1990) had promised huge increases in profit from small improvements in loyalty (as discussed in Chapter 3).

  76 Documenting such loss aversion and its logical implications was a big part of Kahneman and Tversky’s work that resulted in a Nobel Prize for economics for Kahneman (Tversky was not alive and therefore could not be awarded).

  77 The other sort of unusual market share a brand could have is unusually low loyalty and high penetration. This is occasionally seen for brands that many people buy but only on particular occasions (e.g. chocolate eggs at Easter).

  78 Chapter 4 shows the skewed pattern of buying rates (negative binomial distribution). The rate of category buying is also typically very skewed with lighter than average buyers being the norm (this follows a gamma distribution).