Background

Brand extensions (products which are sold under the same brand name across multiple categories) are a common way for companies to capitalise on the mental availability they have built.  However, a brand may be mentally available to a consumer in one category but not another. 

It’s widely thought that brand extensions encourage the purchase of new products in the second category, particularly for extensions where the product ‘fits’ with the original. However little is known about whether this is actually the case.

Prior research has shown that line extensions (Lomax & McWilliam 2001), sub-brands (Dawes, 2017), brand extensions (Mundt, 2011) and private label extensions (Nenycz-Thiel & Romaniuk 2018) are more likely to be purchased by brand buyers than non-brand buyers. While these studies give some insight into the level of customers purchasing multiple products from a brand, they also present limitations such as small sample sizes, small number of categories analysed, or category pairings analysed being within the same category, which inhibit their generalisability for the purchase of brand extensions. We conduct a large scale study to overcome these limitations. 

Method and data

Our study uses a one-year Nielsen US panel dataset (2014), with purchases from approximately 60,000 households. We investigate 99 brand extensions from 30 category pairings (a pairing constituting of two categories, such as shampoo and conditioner). These categories consist of food and non-food grocery, dairy, deli, and healthcare products.

We use the Duplication of Purchase (DoP) analysis as a benchmark for measuring cross-category purchasing. This analysis is generally used to look at the extent to which the buyers of any brand in a category also purchase all other brands in that the same category. Further information on Duplication of Purchase can be found in Report 51 (Sharp, 2009). We modify the application of this technique for the purpose of this research to measure the percentage of brand buyers in a given category X, who also purchase the same brand, as well as other brands, in a second category Y. This gives us a benchmark for the sharing of cross-category buying, against which we compare whether the level of cross-category purchasing for a brand and its extension is higher (or lower) than for two unrelated brands. We used 1 as a benchmark for the expected level of cross-category sharing for two unrelated brands. Scores above this indicate a brand extension effect. 

Results

Our results show that, across a one-year period, purchasing a brand in one category makes consumers, on average, 2.4 times more likely to purchase the brand extension in a second category, than consumers who purchase a different brand in the original category. Of the 30 pairings analysed, 25 (83%) achieve higher levels of cross-category sharing than what is expected based on that of two unrelated brands. 

While the average level of cross-category sharing is 2.4 times what is expected, there is considerable variation within these scores. The lowest level is 0.8 for baking soda and deodorant, meaning purchasing a particular brand of baking soda corresponds with the consumer being 20% less likely to purchase the same brand in the deodorant category, compared to someone who purchased a different brand of baking soda. The highest level of cross-category sharing is for shampoo and conditioner, where the purchase of a brand of shampoo corresponds with an almost nine times increase in likelihood to purchase the conditioner extension compared to those who purchase a different brand of shampoo. Figure 1 illustrates the level of cross-category sharing for all pairings analysed. 

Figure 1: Level of cross-category purchasing for category pairs

There is also much variation at a brand level. Of all brands analysed, 13% achieved cross-category sharing of equal to or less than one, one being what is expected for two unrelated brand names. Six percent of brands analysed achieved cross-category sharing greater than 10 displaying a very large brand name effect. The majority of brands achieved cross-category sharing 2 – 5 times what is expected,  similar to the average score at a category level.

Table 1: Level of cross-category purchasing for brands

Complementary and substitutable categories

Complementarity and substitutability are said to influence the level of cross-category purchasing that brand extensions achieve. Complementarity is where two products can be jointly consumed to fulfil a need, such as shampoo and conditioner both being used to wash hair. Substitutability is where two products can be used alternatively to fulfil a need, such as fresh meat and packaged meat. 

Our results show that complementarity is the biggest driver of variation of cross-category sharing achieved. Extensions in complementary and substitutable categories both achieved higher levels of cross-category sharing than expected for two different brands. However, extensions in complementary categories achieve higher levels of cross-category sharing than extensions in substitutable categories; three times what is expected for complementary versus two for substitutability. 

Similarity between categories

Similarity, the extent to which the product features/attributes of one product are similar to that of another product, is also said to have a positive impact on the level of cross-category sharing extensions achieve. We conducted a consumer survey to understand the perceived level of similarity between product categories. From this, we were able to classify category pairings as having high, moderate or low levels of similarity to each other based on the similarity of product features are between pairings.

In this research, extensions released in product categories with low levels of similarity achieved the lowest levels of cross-category purchasing (1.6 times what is expected). Extensions in product categories with moderate and high levels of similarity achieved the same level of cross-category purchasing (2.8 times). While this level of cross-category purchasing is higher than what is expected for two different brands, it would be expected that the higher the level of similarity between product categories, the higher the level of cross-category sharing. Further, the difference between cross-category purchasing of low, moderate and high levels of similarity is not statistically significant. Extensions in categories with moderate and high levels of similarity both achieve lower levels of cross-category sharing than extensions in complementary categories. Therefore, while all conditions explored achieved cross-category sharing at levels higher than that for two unrelated brands, extensions in complementary categories achieve the highest levels of cross-category sharing. 

Implications

Our results are in the expected direction; those consumers who buy a brand in one category are more likely to buy it in another.  Which is comforting news to those contemplating launching a brand extension.

Though, there is considerable variation in this effect between categories. Even complementary categories, where the effect tended to be strongest, did not always show excess sharing. Perhaps in these instances this is due to brand effects. That in some cases a brand has been able to translate its mental availability across categories, whereas other brands have built quite separate mental availability in different categories.  

We can certainly see a number of obvious real-world examples, for example Husqvarna is a major brand of sewing machines, which also features on chainsaws, lawn mowers, leaf blowers, and even motorcycles. Yamaha is another motorcycle brand but which also manufactures pianos (where it in fact started). Perhaps there are many consumers who notice (for example) the brand Heinz in the mayonnaise category but not in ketchup, while for many others it is the other way around. There is nothing to say that there is anything wrong with this situation, but the brand owners would need to understand that Heinz operates essentially as a different brand in each category, just as Walkers crisps and Walkers shortbread are completely different brands; though in this case with different owners.

Our discovery reinforces the value of a brand and its mental availability (further information about mental availability can be found in Report 16 (Romaniuk & Sharp 2004)). However, cross-over effects for a new launch brand extension can not be taken for granted, which is be supported by the prior research showing that brand extensions do not enjoy higher success rates. It is important companies provide marketing and advertising support to brand extensions to ensure they do not fail.