Segmentation for Private Labels and National Brands – an examination of ‘Within-Demographic Market Share’ Rui Hua Huang and John Dawes, Ehrenberg-Bass Institute for Marketing Science, University of South Australia Abstract The study examines the extent of brand segmentation, for private labels and national brands, in four UK grocery categories. It uses a straightforward but effective method – ‘within- demographic market share’. This approach brings together the concepts of segmentation and brand performance. We find that both private labels and national brands exhibit quite different levels of market share within particular demographic groups. Specifically, cheaper private labels and low priced national brands are comparatively more popular in larger households and in lower social class households. The more expensive private label and national brands have comparatively higher market shares among higher social class households. That said, it is noteworthy that expensive brands still retain some share among low socio-economic households and cheap brands still sell into high socio-economic group households. Generalising these findings to more categories will help retailers and manufacturers to understand brand strengths and weaknesses and to formulate targeting and positioning strategy. Introduction and Background Manufacturers of packaged goods brands have always been interested in understanding whether their markets exhibit brand-level segmentation. This interest extends to the arena of private labels, which are continuing to threaten manufacturers’ brands. What is known about brand segmentation – and are private labels segmented in any meaningful way? The logic of segmentation is that marketing will ‘work better’ by identifying groups with different needs and specifically catering to chosen target segments. However, as Wright and Esslemont (1994) point out this is a logical belief, not an empirically validated fact. Another obvious area of interest for marketers is brand performance measures. This paper examines market segmentation in a way that ties it directly to brand performance measures. Extensive research has been conducted on the topic of segmentation in packaged goods generally, and some other work has been done specifically examining store brands and manufacturer brands. For example, Cunningham, Hardy and Imperia (1982) noted that demographics vary among the users of national and store brands in terms of education level, age, and the attitude toward price and quality. Ailawadi, Neslin and Gedenk (2001) found that private label usage segments are associated with certain psychographic and demographic traits. Kamakura and Russell (1989) found that propensity to buy private label products (and cheaper national brands) was related to consumer’s self-stated price-sensitivity. While these studies have yielded interesting results, it is fair to say the emphasis has in many cases been more about technical exposition with less emphasis on the development of generalisability across markets (for an exception see Fennel et al., 2003). Interestingly, some prior work on brand segmentation has found that there is not a lot of segmentation among directly competing brands (Hammond, Ehrenberg and Goodhardt, 1996; Kennedy and Ehrenberg, 2001). The conclusion from this work is that substitutable brands “usually compete in a largely unsegmented mass market” because their users’ profiles tend to be very similar in terms of demographics and attitudes (Kennedy and Ehrenberg, 2001). The methodology in these two pieces of research is distinctive for its scope – analysis across a large number of categories. It was also appealingly straightforward - tabulation and comparison of the user profiles for each brand. There is arguably a methodological advance that can be made to that body of work. For the buyers to be counted as part of the brand’s users’ profile, they only had to buy that brand once in a time period (say, a year). This approach therefore does not consider the weight of purchases made by buyers – how much or how often people are buying the brand. Perhaps differential brand appeal among demographic groups would be more transparent if the number of buyers and weight of purchase were considered together. Effectively this means examining the brand’s market share among specific groups of buyers. As marketers are keenly interested in market share as a performance metric, breaking market share results down by demographic sub-groups seems an appealing way to advance both the segmentation and brand performance literature. Hence, this study is designed to identify segmentation – for both private labels and national brands, using this proposed approach. It uses a similarly straightforward method to identify market segments, namely ‘within-demographic market share’. This method involves simply tabulating the market share for each brand, within each demographic group. The advantage of this method is that it captures several brand performance measures – market share, penetration, and purchase frequency simultaneously. The results will be particularly useful to retailers and manufacturers to understand whether brands have differential appeal to specific, identifiable groups of buyers. Method The study looks at this ‘within-demographic market share’ for two demographic groupings – household size and social class. Both the number of people and the income level in households directly influence purchase incidence or buying power (affordability). ‘Within- demographic market share’ represents the brand’s share of the total purchases of the category made by consumers in that demographic group. For example, Nescafe instant coffee might have 29% market share overall, but what is its share among say, people in social class AB? The data analysed here is from the TNS Superpanel in the UK to whom we are indebted. The panel comprises approximately 15,000 consumers. The study uses four CPG markets - ground coffee, instant standard coffee, sweet/semi-sweet biscuits, and RTE (ready-to-eat) cereals. Data are for a 12-month period. The largest market share brands are selected for examination. There are many small brands with market share below 1%, thus we aggregate several of them to make a ‘super brand’. These are denoted as PL01, PL02, NB01, and so on. PL means private label and NB means national brand. The aggregation is carried out based on price level – lower, equal, or higher than the average price. The numerical ordering for the ‘super brands’ is in order of average price level, for example, PL01 (or NB01) is cheaper than PL02 (or NB02), which is additionally cheaper than PL03 (or NB03). Results & Discussion Table 1. Within-Demographic Market Shares for Ground Coffee (2002) Private Labels PLs National Brands NBs PL01 PL02 PL03 PL04 PL05 Average NB01 NB02 NB03 NB04 Average Average Price (£/kg) 5 6 7 8 9 6 7 9 11 Aggregated price level Š5 5~6 6~7 7~8 9 Š6 6~7 8~9 11 Overall Market Share (%) 15 5 9 6 5 8 5 16 7 11 10 Household Size No Children 15 5 8 7 5 8 6 15 7 11 10 1 Child HHs 14 6 10 4 4 8 4 19 8 12 11 2 Child HHs 15 6 8 5 7 8 7 15 6 11 10 3+ Child HHs 9 9 16 7 8 10 4 17 6 8 9 * % Deviation Max to Min 67% 80% 100% 75% 100% 75% 27% 33% 50% # Abs. Deviation Max to Min 6 4 8 3 4 3 4 2 4 Social Class Class AB 11 3 9 9 5 7 4 12 8 16 10 Class C1 13 5 9 6 7 8 5 16 8 11 10 Class C2 16 5 12 6 4 9 6 17 6 10 10 Class D 21 7 8 3 4 9 8 19 5 6 10 Class E 16 5 7 6 6 8 6 16 7 11 10 * % Deviation Max to Min 91% 133% 71% 200% 75% 100% 58% 60% 167% # Abs. Deviation Max to Min 10 4 5 6 3 4 7 3 10 * % Deviation Max to Min = (Maximum share – Minimum share)/Minimum share; # Abs. Deviation Max to Min = Maximum share – Minimum share; - Brands are ordered by average price in ascending order from left to right in private label and national brand sub-markets. - Maximum and minimum shares are highlighted in bold. Table 1 displays the within-demographic market shares for ground coffee. We report how much variation there is between the brand’s lowest share in one demographic grouping compared to its highest share in another demographic grouping. For example PL03 has 8% share in no-child households and 16% share in 3+child households. In proportional terms this is a 100% deviation within the household size demographic grouping. This “% Deviation Max to Min” shows that certain brands over-perform in some sub-groups and under-perform in others. Overall, most private labels and cheap national brands tend to have comparatively higher shares in the households with bigger size, such as PL02, PL03 and NB01, or in lower social class, such as PL01, PL02, NB01 and NB02. Table 2. Within-Demographic Market Shares for Instant Standard Coffee (2000) Private Labels PLs National Brands NBs PL01 PL02 PL03 Average NB01 Nescafe NB02 Average Average Price (£/kg) 6 12 18 14 16 22 Aggregated price level 5~10 12 15~19 Š15 - 20 Overall Market Share (%) 7 4 3 5 20 29 24 24 Household Size No Children 7 4 3 5 19 28 26 24 1 Child HHs 8 3 3 5 21 30 22 24 2 Child HHs 7 4 3 5 22 30 22 25 3+ Child HHs 11 3 2 5 20 31 19 23 * % Deviation Max to Min 57% 33% 50% 16% 11% 37% # Abs. Deviation Max to Min 4 1 1 3 3 7 Social Class Class AB 3 3 5 4 13 21 39 24 Class C1 5 3 4 4 18 29 28 25 Class C2 7 4 3 5 20 32 22 25 Class D 10 5 2 6 22 29 19 23 Class E 11 6 2 6 23 28 20 24 * % Deviation Max to Min 267% 100% 150% 77% 52% 105% # Abs. Deviation Max to Min 8 3 3 10 11 19 * % Deviation Max to Min = (Maximum share – Minimum share)/Minimum share; # Abs. Deviation Max to Min = Maximum share – Minimum share; - Brands are ordered by average price in ascending order from left to right in private label and national brand sub-markets. - Maximum and minimum shares are highlighted in bold. In the instant coffee category (Table 2), the cheap private label PL01 and national brands NB01 and Nescafe01 particularly appeal to bigger size households, while the expensive store brand PL03 has higher share among smaller households. Lower priced brands tend to have relative higher within-demographic shares in lower social classes. For example, PL01, PL02, Nescafe Original, and NB01 all have 50% more share in lower social classes than in higher social classes. Indeed the proportional deviation of 267% to PL01 indicates that this brand performs three times better in the lower social class households than in higher social class households. Table 3. Within-Demographic Market Shares for the Sweet/Semi-sweet Biscuits (2002) Private Labels PLs National Brands NBs PL01 PL02 PL03 Average McVities NB01 NB02 NB03 Average Average Price (£/kg) 1 2 3 1 2 3 6 Aggregated price level Š1.5 1.5~2.2 2.6 - 1.2~2.4 2.7~4.4 4.8 Overall Market Share (%) 23 18 4 15 5 11 16 5 9 Household Size No Children 19 19 5 14 6 13 15 3 9 1 Child HHs 24 18 3 15 4 10 17 6 9 2 Child HHs 28 16 3 16 3 9 17 7 9 3+ Child HHs 33 13 2 16 3 8 16 6 8 * % Deviation Max to Min 74% 46% 150% 100% 63% 13% 133% # Abs. Deviation Max to Min 14 6 3 3 5 2 4 Social Class Class AB 18 19 5 14 5 11 16 6 10 Class C1 20 19 5 15 5 11 16 6 10 Class C2 26 18 3 16 5 10 16 5 9 Class D 25 16 3 15 5 11 15 4 9 Class E 22 17 4 14 5 13 15 3 9 * % Deviation Max to Min 44% 19% 67% - 30% 7% 100% # Abs. Deviation Max to Min 8 3 2 - 3 1 3 * % Deviation Max to Min = (Maximum share – Minimum share)/Minimum share; # Abs. Deviation Max to Min = Maximum share – Minimum share; - Brands are ordered by average price in ascending order from left to right in private label and national brand sub-markets. - Maximum and minimum shares are highlighted in bold. Table 3 presents the within-demographic market shares for the biscuits category. Overall, the biscuits market is fragmented with many small brands. Prices also do not vary much from each other (averagely priced around £1-3/kg except NB03 with the highest price level). Here there is not much segmentation evident. The cheap private label brand PL01 enjoys comparatively more market share in larger households. McVities is more popular in ‘no- children’ households. Implications and Directions for future research Overall, the analysis of ‘within-demographic market share’ shows that cheaper private labels and cheaper national brands are comparatively more popular among households with more people, or those in a lower social class. More expensive brands appeal comparatively more to the buyers in a higher social class. The deviations among within-demographic market shares suggest that some price-based segmentation does exist in grocery markets – across private labels and manufacturers’ brands. This study is consistent with an examination of an Australian grocery category by Dawes (2006), who found markedly different brand performance levels across various demographic groups. It perhaps indicates somewhat more segmentation than suggested by Hammond, Ehrenberg and Goodhardt (1996) and Kennedy and Ehrenberg (2001). This could be due to the fact that the present study simultaneously incorporates the two variables that comprise market share – penetration and purchase weight. However, it is important to also note that cheap private labels still apparently sell into high social-class households and that expensive brands still sell into low social-class households and large households. So while there is some segmentation, brands at different ends of the price spectrum are not ‘cut off’ from certain parts of the market. Further work could also examine the extent to which temporary discounting by manufacturer brand extends their shares into private label prone demographics. Results of this kind help the marketer to understand the composition of their overall market share. A brand may enjoy X% market share overall, but it does better in sub-group Y and somewhat more poorly in sub-group Z. Such results can be the basis for further investigation - why do we do so well in demographic group Y? The results also have some implications for manufacturers and retailers, particularly if they eventually generalise across many categories. If a manufacturer has certain brands shown to under-perform among certain sub-groups, this identifies an opportunity for adding to the product line with less cannibalisation effect (Lomax and McWilliam, 2001; Mason and Milne, 1994). The findings may also indicate some scope for targeting specific groups. For example, suppose there is a strategy of promoting a manufacturer brand such that it takes some market share away from a private label brand. Can the promotion be targeted to the group among whom the private label brand is particularly popular? The approach used here may identify the target group demographic characteristics. For manufacturers it may be also advantageous to know if their brand has particular appeal to specific demographic sub-groups. Take the example of a smallish brand being considered for de-listing by a retailer. Perhaps the manufacturer can point out that the brand, while small overall, is particularly popular among a specific consumer sub-group – say high-income earners or heavy buyers. This would be a good argument not to delist this small brand, because arguably the retailer does not want to upset such customers. As a final note, we re-iterate that although the results show some segmentation, it also shows that higher priced or lower priced grocery brands are not ‘cut off’ from certain parts of the market. It also highlights that mass appeal is still needed to be a big brand. References Ailawadi, K. L., Neslin, S. A. and Gedenk, K. 2001. Pursuing the value-conscious consumer: Store brands versus national brand promotions. Journal of Marketing 65 (January), 71-89. Cunningham, I., Hardy, A. and Imperia, G. 1982. Generic brands versus national brands and store brands. Journal of Advertising Research 22, 25-32. Dawes, J. 2006. Interpretation of brand penetration figures that are reported by sub-groups. Journal of Targeting, Measurement and Analysis for Marketing 14 (2), 173-183. Fennell, G., Allenby, G. M., Yang, S. and Edwards, Y. 2003. The effectiveness of demographic and psychographic variables for explaining brand and product category use. Quantitative Marketing and Economics 1, 223-244. Hammond, K., Ehrenberg, A. S. C. and Goodhardt, G. J. 1996. Market segmentation for competitive brands. European Journal of Marketing 30 (12), 39-49. Kamakura, W. A. and Russell, G. J. 1989. A probabilistic choice model for market segmentation and elasticity structure. Journal of Marketing Research 26 (November), 379- 390. Kennedy, R. and Ehrenberg, A. 2001. There is no brand segmentation. Marketing Insights, Marketing Research 13(1), 4-7. Lomax, W. and McWilliam, G. 2001. Consumer response to line extensions: Trial and cannibalisation effects. Journal of Marketing Management 17, 391-406. Mason, C. H. and Milne, G. R. 1994. An approach for identifying cannibalization within product line extentions and multi-brand strategies. Journal of Business Research 31, 163-170. Wright, M. and Esslemont, D. 1994. The logical limitations of target marketing. Marketing Bulletin 5, 133-120.