How Brands Grow
Page 6
Figure 4.2: Percentage of UK cola buyers purchasing Pepsi x times, 2005
Source: Kantar Worldpanel.
Is the UK cola market strange? What does the US look like? Figure 4.3 shows the purchasing pattern (this time at household level) of Coke and Pepsi in the US in 2012. The UK and US buying patterns are astonishingly similar because this pattern generalises around the world, over time, across product categories, and for all the different market research providers. This pattern is nothing to do with soft-drink, it concerns all markets.
This time the population base is all shoppers not just cola buyers so the penetrations are lower (more zero buyers). Clearly both of these large brands have plenty of room to recruit consumers.
Figure 4.3: Percentage of US households purchasing Coke & Pepsi x times, 2012
Source: Nielsen 2012.
So brands have many light buyers who buy the brand very occasionally. They buy the brand infrequently because they don't buy from the category very often, and they buy a number of different brands. For example, it's very common for a high-volume, packaged goods category to be bought on average less than 10 times a year by buyers.21 Brands in such categories are only bought by their consumers three or four times a year on average. For such a brand their most typical buyers buy them barely more than once a year.
Marketers easily forget how rarely their buyers buy their brand. They are often surprised how low their average purchase frequency metric is (and they mistakenly conclude that this means there is plenty of scope to easily increase this metric). Few appreciate that this average is still much heavier than their typical buyer – and that the calculation of the average doesn't include all those who happened to not purchase the brand at all during the period. The great mass of typical buyers only very occasionally buy a brand.
This shows up in services markets too, where a large proportion of customers buy most of their purchases from another brand. For example, in personal banking, almost half a bank's customers will list another bank as their main financial institution.
At the other end of the buying spectrum there are a few buyers who buy a category often – and a few of these buy a category very often. These buyers are important because they deliver a lot of sales volume in spite of their small numbers. For example, the 4% or so of Coca-Cola's buyers who purchased Coke once a week or more (52+ times that year) delivered a bit more than 20% of its sales volume that year.
Fortunately, these ultra-heavy buyers are comparatively easy to market to, because the category and the brand are, comparatively speaking, much more important to them than to the typical buyer. These heavy buyers get many opportunities to see point-of-sale material and packaging changes (and they are presumably very good at learning about regular promotions). These buyers are also far more receptive to the brand's advertising: they notice it more and they find it easier to process and remember.
At the very extreme, the buyer who drinks Coca-Cola morning, noon, and night has a very well ingrained habit. They are locked in their ways and may even be a little bit addicted to the product category22. One could argue there is little need to market to these buyers at all, especially considering that they are comparatively non-responsive to advertising (i.e. their behaviour tends not to change in response to advertising; their buying neither increases nor decreases). They will keep buying in large volumes until one day something momentous happens and they downgrade, or drop the brand, or quit the category (or die), all of which is usually out of your control.
At the other end of the buying spectrum are the typical, very infrequent buyers. These people – who are most of your consumers – are a marketing challenge because it is hard to justify spending money on them individually (direct mail is usually out of the question). And yet collectively they are important for sales volume and offer great potential for growth. An implication of the skewed distribution of buying rates is that to maintain sales a brand needs to reach out to these masses of buyers. For two reasons:
1. there are so many of them
2. they buy so infrequently and could easily forget about you.
Pareto's law (but not as you know it)
You may be wondering if this distribution of buying frequencies has something to do with marketing's best-known law: 80% of sales come from the heaviest 20% of brand buyers. Yes it does, it underpins Pareto's '80/20' law. However, it is important to know that the '80/20 law' is a misleading simplification. The ratio is rarely as extreme as 80/20.
The share of purchases undertaken by the heaviest 20% of a brand's buyers (the 'Pareto share') reflects the polarisation in buying rates between the heaviest and lightest buyers. In our Coca-Cola example, there are plenty of cola buyers who buy themselves a Coca-Cola only a few times a year, and a smaller group who buy much more regularly (once a week or daily or more). This results in a very typical Pareto share (closer to 50% than 80%).
In contrast, there are categories where buyers are more homogeneous. For example, most people fill their car with petrol once a week. There is a tiny group of people who don't own a car, yet who occasionally buy fuel when they hire a car – but only a few of these people show up in a period of analysis. There are some other people who drive a lot, and fill up twice, maybe three times a week. But the vast majority of people buy petrol once a week (and even the heavy buyers don't buy significantly more than this), so buying rates aren't polarised and the Pareto share is not very extreme.
The Pareto share metric depends on the time period of study (Schmittlein, Cooper & Morrison 1993). In a very short time period all the people who have bought the brand will have bought at a very similar rate; for example, in a week almost everyone who has bought has done so just once, perhaps a few have bought twice, in which case the heaviest 20% of buyers will be responsible for about 20% of sales (say 25/20). As time goes by the heaviest buyers reveal their true colours and show how much more frequently they buy the brand. Also, larger numbers of very light buyers enter the analysis window by buying the brand just once. This increases the polarisation between heavy and light buyers – giving a more extreme Pareto share metric.
Our research (Sharp & Romaniuk 2007) of many dozens of brands, across product categories, shows that over a three-month period a 'fast-moving consumer good' brand will typically have a Pareto share of only 35%. Over a year this metric will have risen to over 50%, usually not far over 50% and rarely anything like the proverbial 80%.
Another way of putting this is that the lightest half of buyers deliver only 20% of volume, leaving the heavy half to deliver the other 80%23. Which can be summed up by Professor Gerald Goodhardt's 20:30:50 law, which states that the 20% heaviest buyers account for 50% of purchases (proved true in Sharp & Romaniuk 2007), the 50% lightest buyers account for 20% of purchases, and so the middle 30% of buyers account for 30% of purchases. In short, 20:30:50 buyers accounting for 50:30:20 purchases.
Table 4.1 shows that brands within a category have a similar Pareto share. This doesn't vary a great deal between categories. Table 4.2 contains 2007 Nielsen BrandScan data reporting the average Pareto shares for brands in a selection of US product categories.
Table 4.1: Pareto share for brands in the body spray and deodorant category
Brand Market share (%) Percentage volume accounted for by heaviest 20% of buyers
3 months 12 months
Sure 16 42 53
Lynx 14 41 53
Impulse 8 45 55
Rightguard 7 39 51
Dove 6 36 48
Tesco 4 43 53
Asda 3 42 54
Adidas 3 35 45
Gillette Series 3 39 50
Nivea 2 36 46
Natrel Plus 1 35 47
Mum 1 35 46
Average
39 50
Brands within a category have a similar pareto share (Sharp & Romaniuk 2007)
Data Source: Kantar Worldpanel.
Table 4.2: Pareto shares for brands in a selection of US product categories
Product Brand average
volume accounted for by its top 20% of consumers (%)
Soft drinks 64
Yoghurt - Refrigerated 60
Cat food - Dry 56
Dog food - Dry 54
Breakfast cereals 54
Yellow Fats 53
Canned Soup 53
Fabric softeners - Liquid 51
Detergents - Light Duty 49
Deodorants - Aerosol 48
Creme rinses and conditioners 47
Automatic dishwasher compounds 45
Deodorants - Roll on 44
Fabric softeners - Dry 43
Shampoo 42
Fabric Softeners - Aerosol 35
Average 50
Source: Sharp & Romaniuk 2007, data courtesy of The Nielsen Company.
If 80% of your customers delivered only 20% of your annual sales, it would be tempting to ignore them. But when these light buyers deliver around half your sales, do you still want to ignore them? Marketing's Pareto law is important, but the ratio isn't 80/20 and the traditional implications are incorrect.
Normally the Pareto law is used to justify a strategy that concentrates on the brand's heaviest buyers (e.g. Koch 1999). The (misplaced) logic being that these buyers are worth more, and marketers can justify spending more per buyer. But making them the dominant focus of marketing activity is not wise. Ignoring light and non-buyers of a brand is no recipe for growth (as we saw in the Chapters 2 and 3). Besides the logic is flawed, what matters is not so much the weight of a buyer as their propensity to change in reaction to marketing. And how many of these customers there are; what they are worth in total?
Buyers aren't always what they seem
The logic of targeting heavy buyers is undermined further by the fact that the future sales potential of individuals is different than their current buying suggests. This is true even when you have perfectly reliable sales data on your individual buyers, and even when there is no real change in their behaviour. Non-buyers and light buyers are heavier buyers than you think, and heavy buyers are lighter. This is neatly illustrated in a two-year analysis of a leading brand of tomato sauce in the US using IRI and Nielsen panel data (Anschuetz 2002). Although the brand was stationary (not growing or losing sales volume), 14% of its sales came from households that did not buy it at all in the year before; in other words, from people who the brand's marketing team would have considered were non-buyers of the brand. While the small group (9%) of heavy buying households delivered 34% of volume, this was less than the 43% they delivered in Year 1. Table 4.4 shows that over time the heavier buyers get lighter, and the non-buyers and light buyers get heavier.
The way the lightest buyers became heavier and the heaviest buyers became lighter is a 'regression to the mean' phenomenon. This law (which we'll call 'the law of buyer moderation')24 applies to all brands and can be precisely predicted from the known distribution of buying frequencies.
Table 4.4: Sales volumes from different buyer groups one year later
Buyer group % of total sample Buying frequency
in year 1 Representing brand sales volume
Year 1 Year 2
Non-buyers 44 0 0% 14%
Light buyers 22 1 14% 16%
Moderate buyers 25 2 - 4 43% 36%
Heavy buyers 9 5+ 43% 34%
Total 100
100 100
Source: Anschuetz, 2002; US IRI panel data.
In fact, few of these buyers are actually changing, which can seem rather mysterious, and is widely misunderstood. Most people don't know about the law of buyer moderation. This law undermines marketers' strategies to target heavy buyers of a brand and ignore light buyers (and the great deal of sales potential that they offer).
The law of buyer moderation occurs because of variation in the timing of individuals' purchasing. Some years buyers purchase the brand once, other years they buy it twice as much – this isn't real change; there is simply a (predictable) degree of wobble even around a stable ongoing buying rate. This wobble means that some of the households that were called 'non-buyers of the brand' weren't really – they just hadn't bought in the particular base year and so were misclassified as non-buyers. Just as some buyers were misclassified as light buyers when they really were heavier. Other customers were misclassified as heavy buyers because they bought the brand a bit more often than usual in the year the analyst chose to classify them (perhaps relatives came to visit so one year they had to buy extra). This phenomenon is much worse over short periods (a month or a quarter) than over a year, though it still occurs in annual data, and even for a big brand, as Table 4.4 above shows.
Remember people wobble/vary in both the amount they buy from the category and how they distribute their buying within their repertoire – due to thousands of things like the weather, getting caught in traffic, having a party to go to, having someone to stay, taking a different path through the supermarket and so on. It’s easy for a person to go from buying a brand twice in one year to four times in the next, and the other way round, without any change in that person’s loyalties (see Sharp et al 2012). For many brands random wobble like this can move a consumer from being a light buyer (bottom 80%) to being a heavy buyer (top 20%) because most buyers buy less than twice a year – see Figures 4.1, 4.2, 4.3.
So we have three key facts about marketing's Pareto law:
1.It is law-like and applies across brands and categories.
2.It's not as severe as 80/20.
3.The analytical time period affects Pareto metrics and messes up attempts to target based on customer value. Put simply, next period your heaviest 20% of customers won't be so heavy, the light buyers will be heavier, and some of the non-buyers will buy. This is the law of buyer moderation.
However, there is another law-like pattern that, in Professor Andrew Ehrenberg's (2004) words, “should have been the end of the marketing pipe-dream of just recruiting heavy-buying buyers”. Let's look at what happens to a brand's distribution of buying rates when the brand grows its market share.
Changes occur across heavy and light segments
Look again at Figures 4.1 and 4.2, which show the buying frequencies of Coke and Pepsi buyers.
The shapes of the distributions are very similar – going from Coke to Pepsi it looks as if the frequencies have simply slid to the left (i.e. everyone buying a little less frequently). In fact, both these distributions have the same mathematical properties; this type of distribution is called a negative binomial distribution (NBD). NBD seems to describe the purchase frequencies of all brands, and has done so for decades (it was discovered in 1959 by Andrew Ehrenberg). The NBD is typically a very skewed distribution (i.e. there are many more buyers who are lighter than the average buyer). Hence, there is a 60/20 Pareto concentration of sales volume.
This large proportion of very light buyers shows up in brand performance metrics (as we saw in Chapters 2 and 3). Remember that brands with larger market share typically have substantially higher penetration figures than smaller brands (the double jeopardy law). But in contrast, the purchase frequency scores of the larger brands (how often their buyers buy their brand) are only marginally higher. And when brands grow or decline there is a lot of change in their category penetration (the size of their buyer base) and little change in their purchase frequency. Now we can see the reason why the penetration metric shows so much change. It's due to the NBD's skewed distribution of buying rates. When a brand grows it recruits a lot of light buyers who become heavy enough to buy at least once during the period of analysis; so now they show up in the penetration metric. Some heavier buyers also buy a bit more frequently, but remember there aren’t many of these people so overall the average rate of buying doesn't change much (see how little the mean average differs in the Coke and Pepsi charts). Therefore, the NBD explains the double jeopardy law.
One of the great insights is that all brands, irrespective of their size, face an NBD of heavy to light buyers. When brands grow, or decline in market share, they simply move from one weight of this distribution to
another. Put another way, changes in sales come from buying propensities changing across the entire market – from heavy to light and non-buyers of the brand. Every buyer group (or weight) changes25. When marketing is successful in delivering more sales and market share, it does so by giving the brand more heavy buyers, more medium buyers and a lot more light buyers. This means that for maintenance or growth, a brand's marketing has to somehow, at least over time, reach all the buyers in a category.
Even if a brand gained 100% market share, which would mean it had 100% penetration and perfect loyalty, it would still have many light buyers because category buying also shows a skewed distribution. For example, Figure 4.4 shows the distribution of buying for the toothpaste category - one where we would hope there was rather regular buying. Yet is seems rather rare for a UK household to buy toothpaste more than six times a year. Most bought three times or less, indeed 11% did not buy toothpaste that year.
Figure 4.4: Toothpaste category buying distribution
Source: data kindly provided by Kantar, UK, 2007.
Of course, for an individual toothpaste brand these figures are even lighter, less frequent, because a brand’s uses also buy other brands. See Figure 4.5 which shows the buying distribution for Colgate in the same market in the same year. Now 86% of shoppers bought three times or less, with 59% not buying Colgate that year.
Figure 4.5: Colgate buyer distribution
Source: data kindly provided by Kantar, UK, 2007.
Sales growth for Colgate, or any brand, won't come from relentlessly targeting a particular segment of a brand's or the category’s buyers. Yet this targeting fantasy continues to appear in marketing plans and underpins the use of loyalty ladders and other market research products (e.g. the Conversion Model, ‘super consumers’). This mis-guided fantasy is harming marketing effectiveness.