Promotional Effectiveness

(1)

Promotional Effectiveness

A study of the effectiveness of promotional activities of

coloration products in the Dutch beauty care market.

- Public Version -

Menno Buikema

(2)

Promotional effectiveness

A study of the effectiveness of promotional activities of

coloration products in the Dutch beauty care market.

July 2010, Amsterdam

Author: Menno Buikema

University of Groningen, Faculty of Economics and Business, Marketing department. Msc. BA Marketing Management & Msc. BA Marketing Research

L‟Oréal Nederland B.V., Consumer Division, Category Management

E-mail: mennobuikema@hotmail.com

Studentnumber: s1361813

1st Supervisor RuG: Drs. Gert Haanstra 2nd Supervisor RuG: Dr. Sonja Gensler Supervisor L‟Oréal: Drs. Fokke Visscher

(3)

Abstract

This study examines the effectiveness of promotional activities of coloration products in the Dutch beauty care market. A multiplicative model which is estimated with logistical regression is proposed. With the use of this model and several statistical tests, the effects of advertising spending and price promotional activity are

(4)

Preface

This thesis is written to complete the Master of Business administration Marketing Research and Marketing Management at the University of Groningen. I would not have been able to write this thesis and complete this master without the help of several people, therefore I would like to thank them and show my gratitude here. First of all I would like to thank my parents for supporting me emotionally and financially during my study. They made it possible for me to enjoy the benefits of studying for several years, which is a gift which I will benefit from the rest of my life. Without their support I would not have reached the point of writing this thesis, and I would probably be doing a simple job instead of the interesting job I have now.

Regarding the thesis itself I owe a big thanks to my supervisors Gert Haanstra and Sonja Gensler. I would like to thank their coaching, guidance, critical remarks, and inspiring conversations. Also I would like to thank L‟Oréal Netherlands for giving me the opportunity to do an interesting internship and providing me with the data I needed. Especially I would like to thank Fokke Visscher from L‟Oréal, for his supervision during my internship and his feedback on this thesis, and for being a nice colleague.

Amsterdam, July 2010

(5)

1 Introduction

Promotions have an important influence on the market of fast moving consumer goods. For most of the producers of consumer packed goods, around 70% of their marketing budget is allocated for price promotions (Srinivasan et al. 2002). Price promotions can be beneficiary for producers, retailers, and also consumers. Therefore, they have become highly popular since the 70‟s. Several causes can be named for this increase in popularity. The fierce competition among retailers and among producers, caused them to promote more to survive in the market. The increased power of retailers in the channel forced producers to offer promotions if the retailers requested them. Furthermore, consumers increasingly make impulse buying decisions when shopping at a retailer, which makes it increasingly important to influence the consumers while they are in a store.

In the Dutch beauty care industry, price promotions are a much used tool as well, especially by large brands like L‟Oréal. In highly competitive markets, like the coloration market, promotions are held frequently. They often feature deep price cuts, fierce advertising and feature support.

Because promotion spending and its importance for retailers and producers have increased, it has been a subject which is of continuous high interest among academics (Dawes 2004). Moreover, because of the increase of available data and the increase of knowledge of marketing models, more research is conducted by academics and practitioners in this field.

Background

It is generally accepted in marketing literature that price promotions cause an increase of the short term sales volume (Pauwels et al., 2002). However there are several studies (Blatberg and Neslin 1990) that describe negative effects of price promotions on consumer behavior. Price promotions increase consumer price sensitivity and decrease brand equity in the long term. However, other researchers find that price promotions can stimulate consumers to consume more of a product, and consume the product faster (Ailawadi and Neslin 19998). Morover, consumers can be trained to buy more of a product at the same time, which can be beneficial to the producers to prevent customers to switch to another brand, and benifical to the retailers to prevent customers to switch to another store (Mela et al. 1998). However, this can also be a disadvantage for the retailer as it results in less visits of the shop which means decrease of shop traffic (Bell et al. 1999). Price promotions can have a positive effect on sales in the promoted period and the weeks following the price promotion, however, this effect is not long lasting (Pauwels et al. 2002).

Academic and managerial relevance

(8)

on effectiveness of promotions is focused on generalizable findings, which results in conclusions that are hard to use as input for individual retailers or producers in specific markets with products in specific categories. Therefore, specific knowledge is missing in the academic literature which can be used by marketing managers to improve their promotion strategy. This thesis provides more specific information in this field for the Dutch beauty care market in general, and more specific for the coloration market.

(9)

2 Problem definition

As described in the previous chapter the importance of the effectiveness of price promotions is increasing because their frequency is increasing, more money is spent on them, and the need for effectiveness is higher because of increasing competition. The problem that is addressed in this research is that there is a lack of information in academic literature about the effectiveness of promotions in specific markets and categories like the coloration market in the Dutch beauty care market. As most of the academic literature is focused on findings that are as much generalizable as possible, it often lacks in depth analysis of one market or one category. This thesis will provide more specific knowledge regarding the effectiveness of promotions in the specific category of coloration products in the Dutch coloration market. This will help marketing managers to execute their promotion strategies in a more effective way, which in the long term will result in increased sales with lower costs, which means higher profits.

Research objective

The goal of this research is to develop a model that measures the effect of price promotions and advertising spending on sales, to explain, and optimize the allocation of these resources.

Main research question

What is the effectiveness of price promotions and advertising spending, for coloration products in the Dutch beauty care market, and how can this effectiveness be explained and optimized?

Sub questions

1. How can advertising and promotions be defined?

2. How can the Dutch beauty care market and the coloration market be defined?

3. What are aspects of advertising and promotions that are relevant for the effect on sales, and in what way do they relate to each other?

4. What factors influence the relationship between advertising and promotions and sales, and how can this be explained?

5. What other factors besides advertising and promotions have an effect on sales volume of coloration products in the Dutch beauty care market?

Structure of the thesis

(10)

3 Theoretical framework

To be able to find answers to the research questions, a thorough understanding of the problem is needed. Furthermore definitions of the variables in the research question are needed. These definitions and understanding of the variables and their relationships is sought for in academic literature and presented in this chapter. The different viewpoints in literature are compared and critically discussed, to come to a clear overview of the problems at hand.

3.1 Advertising

This chapter starts with a definition of advertising which is followed by an overview of the literature on the effect of advertising on sales. First, to find the relation between advertising and sales in the Dutch beauty market, a more general explanation is sought to find factors that can be of influence on the Dutch beauty care market.

3.1.1 Advertising defined

Advertising is defined by Percy and Elliot (2005) as an indirect way of turning a potential customer towards the advertised product or service by providing information that is designed to cause a favorable impression, what we will call a positive brand attitude (Weilbacher, 2001). According to Keller (2008) advertising is a powerful way of creating a strong, favorable and unique brand association and positive feelings and judgments.

3.1.2 How advertising affects sales

Dekimpe and Hanssens (1995) have identified a framework of six effects through which advertising can influence sales based on other literature. The impact of these effects is discussed separately below with a focus on the relation between advertising and sales.

Consensus exists in the marketing field that advertising often has a considerable immediate positive impact on sales, which is called contemporaneous effect.

Numerous studies have argued that the effect of advertising in one period may be carried over (at least partially) into future periods, which is referred to as carry-over effects ( see, e.g., Givon and Horsky 1990). Consumers are supposed to remember past advertising messages and create “goodwill” towards the brand that only gradually deteriorates because of forgetting.

Givon and Horsky (1990) argue that the dynamic impact of advertising on sales can also work indirectly through

purchase reinforcements: a given outlay may create a new customer who will not only make an initial purchase

but also repurchase in the future.

Bass (1969) warned that advertising spending may be influenced by current and past sales, and should not be treated as exogenous. This effect is known as the feedback effect, as it gives feedback about the sales in the past. This is certainly the case when percentage-of-sales budgeting rules are applied.

Traditional single-equation models treat advertising as exogenous and do not model the dependence of current on previous expenditure levels. Empirical evidence contradicts this “independence” assumption: published time-series models often find significant autoregressive components in a firm‟s spending pattern which results in

firm-specific decision rules (Hanssens 1980). Here again, a chain reaction may occur that affects the total long-run

(11)

Competitive activities may change advertising‟s effectiveness drastically. For example, even though the instantaneous sales response may be positive, its long-run effect could be zero because of competitive reactions.

3.1.3 Different forms of advertising and their effects on sales

There are many different forms of advertising. The most important forms are television and radio commercials, print advertising, outdoor advertising, and internet advertising. According to Benedixen (1993), there is no difference in the effectiveness of different media forms. The creative devices and content of an advertisement determine whether or not that advertisement will be effective, but not the magnitude of the effect. The magnitude of the effect of an advertisement will is dictated by the extent to which the medium is used. The greater the extent of usage of the medium, the less the effectiveness of that medium. However, Dekimpe and Hanssens (1995), do find differences in the effectiveness between different media forms.

Television Television is a very powerful advertising medium, due to the fact that it is an audiovisual

experience. A good TV commercial is like a short movie which gives the consumer a certain look and feel, which gives the consumer an experience that is impossible to match by another advertising medium. Because most channels are watched by very many people the reach can be enormous. Two features of a TV commercial are important: firstly, a commercial can show intangible assets of a brand like a certain feeling; secondly it is a very good way to explain customer benefits of the product. These two strengths can create a positive attitude towards the brand if the commercial is a persuasive commercial (Keller, 2008). Disadvantages of television advertising are that they are very expensive to make and to air on moments when many people watch. Furthermore it is hard to make a commercial that draws the attention of a consumer, due to the overkill of advertising on most television channels.

Television and radio commercials do not have a significant instantaneous effect on sales, but result in a much larger long run effect. For developing sales in the long run, the image oriented TV/radio spending is most effective, with an absolute (net) persistence level of about 75 cents per extra dollar spent on that medium. However, building long-term sales trough advertising is costly, it needs high margins to make it profitable (Dekimpe and Hanssens 1995). As image is very important, if not the most important thing in the beauty care market, television is a very useful medium to increase sales. Because margins on most beauty care products are high, using television advertising can increase profitability if used well.

Print advertising Print advertising can be defined as all paper advertising forms like in magazines newspapers

(12)

Another advantage of print advertisements is that, depending on which medium the advertisement is placed in, a very specific target group can be reached. A disadvantage of print advertising is the passive character of it which makes it hard to attract the attention of a consumer with it. Much used media for coloration products and beauty care products in general are: glossy‟s, gossip magazines, glamour magazines and many other specifically female oriented magazines.

Even though print advertising has a significant instantaneous impact on sales, no meaningful long term impact is observed. For achieving short-term promotional goals such as reducing store inventories, print advertising is more effective than the other forms of advertising.(Dekimpe and Hanssens 1995). This instantaneous effect makes it very suitable for boosting the sales of a product when it is first introduced or when it is part of price promotion.

Outdoor advertising Examples of outdoor advertising are bill boards or other advertisements in public places

like train stations or sports venues. By means of outdoor advertising consumers can be confronted with products during the whole day (Keller, 2008). The main advantage of outdoor advertising is that it can reach a broad target group in a cost-effective manner. A disadvantage is just like with print advertisements that it can be hard to draw the attention of the consumer, because of the passive character of the medium, and also because of overexposure of the consumers.

3.2 Price promotions

To give an answer to the question what the influence of promotions are on sales in the Dutch beauty care market, in the first place a definition of promotions is needed. In the second place a more general overview of what the influence of promotions is on sales as found in literature is useful. By defining the factors that determine the effect of promotions on sales, the more specific effect of promotions on sales in the Dutch beauty care market can be further researched.

3.2.1 Price promotions defined

Many different definitions of price promotions exist in literature. According to Blattberg and Neslin (1990) price promotions are temporary price discounts offered to a customer. They define four characteristics of promotions. The first characteristic is that promotions are action focused: the consumer must take action to get the promotion. The second is that promotions are marketing events: promotions require a relationship between the manufacturer and its customers. The third is that promotions are developed to have a direct impact on behavior: the intended effect is not in the long run like with advertising; instead a direct change in behavior of the consumer is intended. The fourth is that a promotion is developed to influence consumers or marketing intermediaries: this influence by the manufacturer can be direct or through an extra party in between, which normally is a retailer.

Kotler (1988) defines a sales promotion somewhat different: “Sales promotions consist of a diverse

collection of incentive tools, mostly short-term, designed to stimulate quicker and/ or greater purchase of a particular product by consumers or the trade”.

(13)

of an increase of the value of a product can be that a consumer receives more of a product for the same price or a premium or gift is included in the product.

3.2.2 The rise of price promotions

Price promotions have become increasingly popular among marketing practitioners, especially in the Fast Moving Consumer Goods Industry. From 1990 to 2000, the marketing budget spent in the US has increased from 60 to 70% (Aliliwadi et al. 2001). This rise is combined with an increase of tailor made promotions. Tailor made promotions are promotions that are developed by the manufacturer in close cooperation with a specific retailer. These tailor made promotions enable retailers to diversify themselves from the other retailers, which can ultimately result in store switching by the consumers. This type of promotions also benefits the manufacturer as the manufacturer has high influence on how the promotion is executed. The rise of sales promotions has several explanations in literature:

- The increase of the amount of products and brands sold. The resources which are available for promotions among the manufacturers and retailers have to be spread over more and more products and brands.

- Brands have a hard time to diversify themselves from their competitors.

- Retailers are becoming increasingly powerful, they demand many and deep promotions from the manufacturers (Leeflang, 1994).

3.2.3 Different types of price promotions

(14)

Figure 3-1 Different types of sales promotions (Blattberg and Neslin 1990).

3.2.4 General effects of price promotions and their decomposition

A general finding in literature is that retail price reductions cause a significant short-term sales spike (a.o. Blattberg et al. 2001). However, there is a threshold of the effect of these price reductions: discounts below 10 percent often generate sales levels that differ little from baseline sales. There is also a saturation effect: discounts above 25 percent often provide minimal sales increases relative to sales obtained at a 25 percent discount (Van Heerde et al. 2001).

The effects of price promotions on sales which cause these sales spikes can be decomposed in different ways and by different definitions. By decomposing the effects of the price promotions, it becomes visible what the source of the gains in sales is that is caused by the promotion. The decomposition also provides insight in for whom these gains are beneficial. Much research has been done concerning the cause of the increase of sales as a result of promotions. Most of this research divides the effect between the main cause, brand switching, the second cause, purchase acceleration and the third cause, volume growth. The results of these studies show great differences, this is probably due to differences in the researched product categories, which has a great effect on the results (Blattberg and Wisniewski, 1990).

According to Blattberg and Neslin (1990) and Narasimhan et al. (1996) the decomposition of the effect of price promotions is fivefold: brand switching, repeat purchasing, purchase acceleration, category expansion, and store switching.

Brand switching - Blattberg and Neslin (1990)

Brand switching means that customers that normally buy products of another brand switch to the promoted brand. Brand switching exists in two forms: aggressive and defensive. An aggressive switch is when a customer that normally buys another brand but now makes an actual switch of brand. A defensive switch is when a customer buys the same non promoted product instead of a different brand that is on promotion. For a producer it is very useful to know if promoted products are bought because of a defensive or an aggressive switch as it shows if it is attracting new customers or it is just rewarding existing customers. Brand switching can occur in an asymmetrical way, which means that customers do switch from one brand to another in case of a promotion however this does not happen or on a lower level when the situation is reversed (Blattberg et al. 1989).

(15)

Repeat purchasing - Blattberg and Neslin (1990)

Repeat purchasing refers to the probability that a consumer will buy the promoted brand again after the promotion. This probability can change due to a promotion in two ways. Firstly it can increase because every time a consumer buys a certain brand that consumer learns more about the brand and forms a habit of buying the product. Because the promotion was the incentive to make the first purchase of this brand the habit could not have been formed without the promotion being held. This effect is called the purchase effect. Secondly the effect can also be opposite to the effect of the purchase effect. While the purchase effect is increasing the probability of repeat purchasing, the second effect which is called the promotion usage effect can decrease the probability of repeat purchasing. This effect occurs when the consumer promotion results in a negative influence on the perception of the consumer due to the promotion which causes a negative attitude towards the product. These both effects can occur separately or simultaneously, which can result in an increase, decrease or unchanged probability of repeat purchasing. However, Ehrenberg et al (1994) state that price promotions do not have an influence on the probability of repeat purchasing.

Purchase acceleration - Blattberg and Neslin (1990)

Purchase acceleration occurs when consumers change the quantities they buy of a product compared to their normal buying behavior, or the time when they buy it as a reaction to a promotion. When consumers buy more of a product, they often buy less of that product in the future. This increase of buying is called forward buying if it concerns retailers, whereas stockpiling is used when for the same effect regarding consumers. Purchase can result in a change of the moment of purchase without increasing the total sales, this effect is called sales displacement.

Category expansion - Blattberg and Neslin (1990)

Category expansion is an increase in category sales due to increase of primary demand of the category. This increase does not result in a decrease in sales of other brands or other time periods or other stores. However, in this case a promotion increases sales by means of creating new opportunities or occasions to use and purchase the product, which results in an increase of the consumption rate. The consumption rate and purchase timing accelerates yet not to make up for this purchase by purchasing later or purchasing less next time.

Store switching - Narasimhan et al. (1996)

Promotions can induce consumers to switch from the shop store where they normally shop at another store. The switch can be for one time only, however it can also be the beginning of a new habit of shopping frequently at a different store. One of the reasons for retailers to have promotions is to create consumer traffic. This means that consumers visit the store for a particular promotion but also buy other products, which can lead to a large revenue growth for the retailer.

Other views on the decomposition of the effect of promotions

(16)

Another perspective on the decomposition of the effect of promotions is proposed by Van Heerde et al. (2002), which is measured by the unit sales. The unit sales effect of a promotion for a brand can be decomposed into different effects: one attributable to other brands, another attributable to stockpiling, and a third attributable to category expansion; on average, each source accounts for about one third of the unit sales increase; however, the nature of the decomposition depends on the magnitude of a price discount and on the promotion signal. The category expansion effect in a store or chain can be decomposed into a store-switching effect and a within-store effect attributable to other categories.

3.2.5 Why price promotions are offered

Heerde (1999) listed several economic theories from other research which explain why promotions are offered to consumers which are relevant for the beauty care market and the coloration market specificly.

Demand uncertainty (Lazear 1986)

Under demand uncertainty it is more profitable to have a relatively high price in a first period and a lower price in a second period rather than having a fixed price. This theory is in contrast with how most new beauty care products are introduced in the Dutch market. The most common strategy is to introduce the product with a price promotion in the first weeks. It does happen that products which are on the decreasing side of the life cycle curve decrease in price, so this strategy is partly applied, however the differences between the high and low price is generally not very large.

Inventory cost switching (Blattberg et al. 1981)

Retailers use promotions to shift their holding costs to consumers. A low-holding-cost consumer may forward buy or stockpile due to promotion and hold inventory instead of the retailer. The size of coloration products packages is relatively large compared to other beauty care products and because retailers need to have many SKU‟s in stock due to the many color versions. This makes it more likely that retailers use promotions to realize inventory cost switching for this product.

Differential information (Varian 1980)

Consumers have differential information on prices. Some consumers may not find the lowest price in the market because the search cost is too high. Consumers can be divided two groups:A set of uninformed consumers who shop randomly and a set of informed consumers who go to the store that offers the lowest price. Therefore the optimal strategy for retailers is to offer a low price occasionally so as to attract the informed consumers. This is a very much used strategy in the Dutch beauty care market. The drug store chains frequently offer price promotions that are primarily announced through door-to-door promotional folders. The difference between the well informed and uninformed consumers is between those who read these folders and those who don‟t.

Price discrimination (Narasimhan 1984; Jeuland et al. 1985).

(17)

above, the common way to announce promotions in by drug stores in The Netherlands is through door-to-door promotional folders. These folders often include coupons which enable price discrimination.

Prisoner’s dilemma (Blattberg and Neslin 1990)

The concept is that if firm A cuts price, firm B must match the price or else lose profits. While it is optimal for firms collectively to price at a higher level, it is more profitable for each firm individually to cut price if the other does not match. Thus, the firms end up at a lower price than if they colluded.

Firm B

No promotion Promotion

Firm A No Promotion Promotion

High, High Lowest, Highest

Highest, Lowest Low, Low

Table 3-1 Prisoners dilemma Promotions

Table 3-1 represents an overview of the prisoners dilemma applied to promotions. It shows the options for two companies and the results for the profits. Axelrod (1980) has written several articles in which the strategies of these games are analyzed. The result is a “tit-for-tat” strategy in which a retailer follows the moves the opponent is making.

3.2.6 Long term versus short term effects of promotions on sales

According to Nijs et al. (2001) promotions generally do not result in permanent changes in demand. In their paper they investigated the category-demand effects of consumer price promotions. They found out that short-term effects of price promotions are high but the long-short-term effects are weak. After an average period of 10 weeks, there is no impact in 98% of the 560 researched cases.

In the long term even a negative impact on the sales volume of promoted brands can be caused by price promotion because of lowered reference prices. (Ehrenberg et al., Pauwels et al.)

3.2.7 Price promotion effectiveness

The effect of a promotion is depending on many different factors, which are not always easy to measure and to separate from other factors. The effect of price promotions can be measured in the long or in the short run; the differences are discussed in paragraph 3.2.6. The effectiveness of a promotion can be measured by looking at the price elasticity of the demand, which is a reaction of the consumer in the amount of demand following a price cut. Factors that are of influence on this price elasticity are discussed in paragraph 3.2.7.

(18)

quantity bought on deal. Because a regular price reduction is long term, there is no need for the consumer to forward buy to take advantage of the price reduction. Therefore, the regular price elasticity is lower.

Other explanations also exist. One is that because of a threshold effect, the shape of the demand curve follows an S-shape. If the offered price reduction is very small, which is often the case with regular price reductions, there is very little sales impact. A reason for this little impact is can be that consumers do not notice the price difference, and therefore do not react. Another explanation is that in-store promotional cues such as “specials” may serve as Pavlovian conditioned stimuli that induce the consumer to “salivate” in anticipation of a good deal (Blatberg and Neslin 1989).

Promotional elasticity plays an important role in planning and evaluating sales promotions. For example, promotional elasticity is a key factor in determining the manufacturer's trade-promotion policy and, in turn the retailer's decision of whether to pass these promotions through to consumers (Blattberg et al 1990; Walters 1989). It is generally accepted that own price elasticities are negative. Furthermore, price discount elasticities models are often greater than | − 2| (Wittink et al. 1988). The research in the field of elasticity of promotions has taken two approaches (Bolton 1989). One approach considers the relationship between own price elasticity and one explanatory variable in a single category. The second approach considers the relationship between own price elasticities and one explanatory variable across multiple product categories. The second approach makes it easier to generalize the findings for other products and markets. The first approach results in more easy to implement conclusions for a company, and also the amount of data needed is smaller. Because of these two reasons the choice is made in this research to use the second approach.

3.2.7.1 Effects of promotion frequency and time since last promotion on price elasticity

Evidence regarding the dynamic price sensitivity effect is mixed (Kopalle et al. 1999). According to Blattberg et al. (1995) increased promotions reduce the discount spike. They find that the frequency of sales promotions influence the reference price. Consumers are therefore becoming less willing to pay as much as they did for a certain product if it is promoted very often. In contrast to this, Zenor et al. (1998) find that increased promotions amplify the discount spike. On the other hand, Bolton (1998) finds no effect. Boulding et al. (1994) finds the effect varies by brand. Foekens et al. (1999) find that the higher the frequency of a promotion and the deeper the promotion, the lower (toward zero) the price discount elasticity. Mela et al. (1997) find that a high frequency of promotions has a positive effect on the short term, however in the long term this effect disappears because of negative side effects. These negative effects can occur when a brand is over-promoted. If a product is promoted heavily (meaning discounted deeply an promoted frequently), the consumers reference price of the product decreases. The consumer will then buy less of the product at regular price but also will react less when a deal is offered, as the offered price is has become the regular reference price for the consumer (Blattberg et al., 1995). Furthermore, increased frequency of promotions will change the expectation of consumers. Because consumers learn that there is a certain frequency of promotions, it will cause them to wait for the next promotion and stockpile (Bolton 1989). As these conclusions are somewhat contradictory it is hard to predict what the influence is of an increased frequency of promotions in the coloration market on the price elasticity. As the dataset which is used is less than three years in length, the long term results will probably be impossible to measure.

H1a: Brands with a high promotion frequency are less price elastic than brands with a lower promotion

(19)

H1b: The higher the number of promotions in the period before a promotion the less price elastic a current price

promotion is.

In the various extensions of the SCAN*PRO model proposed by Heerde et al. (2002), the time since last promotion is included as an independent variable based on the work of Raju (1992) who found that the recency of promotion‟s effectiveness on sales.

H1c: The time since the last promotion has a positive effect on the price elasticity of the promotion that follows.

3.2.7.2 Effects of promotion depth on price elasticity

Only little is known in academic literature about the influence of the most recent promotion of the own brand on the elasticity of a current promotion. Foekens et al. (1999) find that the deeper the most recent price discount, the lower (toward zero), the price discount elasticity of the promotion that follows.

H2a: The larger the price discount of the most recent promotion, the lower the price elasticity of the promotion

that follows.

As mentioned before, discounts below 10 percent often generate sales levels that differ little from baseline sales. There is also a saturation effect: discounts above 25 percent often provide minimal sales increase relative to sales obtained at a 25 percent discount (Van Heerde et al. 2001).

H2b: The depth of a promotion has a non linear effect on the price elasticity of the promotion, with a threshold

effect below 10% and a saturation effect above 25%.

3.2.7.3 Effects of life cycle stage on price elasticity

Another aspect of a product that is of influence is the stage of the product in the product life cycle. Kotler (1988) found that elasticity increases over the life cycle and found several reasons for this. Firstly, consumers are likely to be better informed about products as the products mature. This increased knowledge about the products, especially their availability, prices, and discounts, makes consumers more price conscious. Second, consumers in the early life cycle (early adopters) are likely to be less price sensitive than later entrants because of their focus on novelty and not economy (Nagle 1987, p. 137; Rogers 1983). Third, because competition is more intense in the mature stage, consumers will be better able to shop around for a good price (Kotler 1988, Sethuraman and Tellis, 2000)

H3: Products that are in an early life cycle stage are likely to be more price elastic than products that are in a

later life cycle stage.

3.2.7.4 Effects of market share on price elasticity

(20)

function, where “flatness” seems to reflect consumers‟ preferences. These high market shares should tend to be own price inelastic (Ghosh et al. 1983). This also holds for promotional elasticity.

H4: The larger the market share of a brand the smaller the price elasticity of the price promotions of that brand.

3.2.7.5 The effect of displays and features on price elasticity

In most price promotion research display is used as a variable as it is of large influence to sales. Blattberg et al. (1995) for instance state that the effect of displays on item sales is known by most practitioners – it is somewhat obvious. They also explain that there can be an interaction effect between display and feature advertising. Displayed promotions in high priced categories are less effective than in low priced categories. (Narasimham, 1996) Display- and feature multipliers show similar average magnitudes in parametric models; in the absence of a price cut sales often doubles with display or feature (Wittink et al. (1988)). Unfortunately there was no data available about display placements available in this research. Thererfore no hypothesis can be tested for the effect of displays on price elasticity and a possible interaction effect of display and feature.

A feature is an advertisement which is placed in newspapers, magazines or special circulars (Blattberg and Neslinn 1990). Only the advertisements in special circulars are counted as features in this research, as advertisements in newspapers and magazines are put in the category of advertisements. These special circulars are used very frequently in the drug store channel. Retailer A sends a circular weekly and one biweekly, Retailer C and Retailer B both have a biweekly circular. These drug store chains almost exclusively have feature placements combined with price promotions. Therefore it is hard to separate the effect of the price promotion and the feature spending. However, as the feature spending is dependent on the size of the feature, the effect of the size of a feature on the price elasticity of a promotion can be measured. The placement of a feature is expected to have a positive effect on sales and on the price elasticity of promotions which is also depending on the size and place of the advertisement, which is measured by the spending on the feature.

However, if a feature is supported with a discount, it can result in delayed purchases, because customers tend to postpone their purchases in anticipation of the upcoming promotion (Heerde et al. 2002). This results in an expected decrease of sales in the period between the publication of the circular and the start of the promotion. In case of the drug store circulars this is normally not more than a couple of days, therefore the pre promotion dip is expected not to be very large.

H5a: Feature advertising spending increases the elasticity of promotions.

H5b: Feature advertising leads to a pre promotion dip in sales due to delayed purchases.

3.2.7.6 Retail channel

(21)

store chain, logical reasoning suggests that there is also a difference in elasticity of promotions between chains for all brands together. Therefore elasticity of promotion is expected to be different per chain, with an expected higher elasticity for high priced brands at price oriented retail chains. This results in the following hypothesis:

H6a: The price elasticity of promotions is significantly different for every drug store chain.

H6b: High priced brands are more effective in their promotions at price oriented retail channels than at more

service oriented channels.

3.2.7.7 Lagged effect

As sales of previous periods can be of high influence to the current sales due to stockpiling, a lagged effect variable should be included in the analysis. A lag of the own sales and of the sales of other products should be included.

H7: Price promotions have a negative effect on the sales of the promoted product in the week after the promotion.

3.2.7.8 Control variables Weighted distribution

Weighted distribution shows the percentage of turnover that is made in the product group, within a specific market, that is realized by the selling shops of that product. This is a factor that is generally considered to be olarge influence on the sales and price elasticity. Therefore, to prevent a bias it is included in the model. However, it as it is not part of the research, no further analysis regarding this variable is included in the research.

Seasonality

(22)

4 The model

In this chapter the research phase of building the model is described. This model will lead to an explanation for the effectiveness of acvertising and promotions in the further chapters.

As acknowledged by Leeflang et al. (2000) it is important that the structure of marketing models meets severeral quality criteria which are definede by Little (1970):

Simplicity: models are representations of the real world. To keep a model simple it should not have too many variables and only include variables that really matter. The models should be easy to understand. Because of this criterion the amount of variables has been kept small and the most simple functional form that fitted has been used.

Evolutionary: The first step in evolutionary model building is to develop a formalized or conceptual model which reflects the model builder‟s view on the studied marketing phenomenon. After the estimation procedure the model should be reviewed to see if the assumptions made in the conceptual model are still valid. In this research this is done in the validation chapter. Over time the model builder will experience the shortcomings of the model. Therefore it should be easy to extend it, while keeping the first criterion in mind to keep the model simple. During the model building process several functional forms, variables, and relations have been checked and sometimes have been removed or added to the model. Also in the conclusion recommendations are made for further evolution of the model.

Completeness: for a decision model to be useful it should include all the essential elements of the studied phenomenon. This criterion easily conflicts with the simplicity criterion.

Adaptive: When the market and or the market behavior changes, a model should be able to accommodate these changes easily. An example could be that it should be possible to ad another brand or chain to the model. Robustness: A model is robust if it is hard to get false outcomes from it. Robustness is achieved through the specification of a structure that constraints outcomes to a meaningful range of values. This presumes that the model builder has enough market knowledge to judge the meaningfulness of the results. The robustness of the model is assessed in paragraph 4.3.2

To build the marketing model to find an answer to the research questions the three steps as proposed by Leeflang et al. (2000) will be used. These three steps are:

1. Specification

2. Parameterization or Estimation 3. Validation, Verification or Evaluation

(23)

4.1 Specification

Accoding to Leeflang et al. (2001) the specification of a model consists of two substeps:

1. Specifying the variables to be included in the model, and making a distinction between those to be explained (the dependent variables), and those providing the explanation (the independent variables).

2. Specifying the functional relationship between the variables, for instance linear/non-linear, additive/multiplicative, immediate and/or lagged etc.

4.1.1 Selecting the independent variables

In this paragraph, the variables in the model and their relation will be determined. Some of the variables that have been described in the theoretical framework and are generally considered to be of influence will be used. Additionally, some extra variables will be introduced.

Price

The price of the products is the consumer price per unit sold. As a coloration product is always one for single treatment of the hair, the volume is always the same. There are no „extra volume for the same price‟ promotions, and no differences per product. Therefore the prices can be compared very well. To be able to compute a relative price change in case of a promotion or smaller price changes, a regular price is computed. This regular price is the price of the product in the periods where there is no price promotion and in case there is a price promotion, the price is changed to a running average of the non promotion price of the previous five and upcoming five periods. By dividing the actual price by the regular price, a price index is computed, which shows the depth of the promotion. This price index is divided in two variables, which are the price index with price promotion and the price index without price promotion. This way the effects of normal price changes can be separated from the effects of promotional activities. A discount is considered a price promotion if the discount is larger than 10%. In addition to this the prices of other brands are of influence on the volume sold. Therefore cross price variables are included in the model, which represent the price indexes with support and the price indexes without support of the competing brands. Furthermore, lagged effects of the price variables are of influence and are included for the previous period of all the variables.

Print and Television Advertising

(24)

Feature

Just like for display placements, feature advertising has an effect on sales according to Blattberg et al. (1995). The effect is somewhat obvious and has been found by many other researchers. Feature advertising is therefore a dependent variable in this model. The variable is measured by the weekly value of feature advertising for the specific product for the specific chain. The value of these features is measured by the research company PCN as a theoretical value. The value is depending on for which chain the folder is, where the feature is placed, the position in the folder, and the size of the feature. The actual prize paid can be very different from this estimated value as it can be part of a package deal, however the actual prizes are unknown. The data is in Euros divided by 1000 for the convenience of using smaller numbers in the analysis.

Distribution

The weighed distribution of some of the products has large fluctuations at some drugstores, however for some other products and drugstores, the distribution is almost 100% all the time. This means that the influence of distribution is expected to be different per chain and product. The weighted distribution is an index from 0 to 100.

Brands in this research

As stated before the data set which is used for this research consists of 10 brands which are part of 4 umbrella brands and the private labels. Due to the fact that 9 of the brands are part of 4 umbrella brands, early testing showed that these brands have large multicollinearity issues. This is statistically tested by checking the Variation Inflation Factor for all the variables. Many of them appeared to have values high above the critical value of 10. This is caused by the fact that these brands have most of their promotions at the same time. So almost all the prices of the brand that are part of a particular umbrella brand, have their prices synchronized almost all the time. Therefore the decision is made to proceed with only one brand per umbrella brand. The brands that have been selected are Brand 1, Brand 3, Brand 4, and Brand 5.

Aggregation level

This research is based on store chain level data which means that all the sales of a certain product in all the stores of a particular chain are aggregated to one figure. Non aggregated data would have been better to estimate the effects of the promotions, however this is unfortunately not available for this research. Most research in the field of promotion effectiveness is done with household level panel data because its more detailed and therefore has more possibility for analysis of elasticities. However this data is very expensive and also not available for this research. As all the chains that are in the dataset have a national marketing strategy that is the same for all the stores of the chain.

(25)

and sometimes in a one or three week period, this aggregation level seems appropriate. Daily scanner data would have been somewhat better, however this is not available. A disadvantage of the weekly data is that the amount of working days in which the stores have been opened are not accounted for. Some weeks can include banking holidays or extra opening hours before banking holidays or extra opening hours on Sunday, when the shops are normally closed.

4.1.2 Functional form

In order to keep the model as simple as possible, to meet the simplicity criterion, the first functional form that is tested is the additive functional form. First an additive model has

been estimated with the use of linear regression. However, this did not appear to fit the data. An analysis of the residuals showed that an additive model leads to heteroscedasticity issues and to a non-normal distribution of the residuals. This is visualized in Figure 4-1 which shows that the residuals are not in a straight line in the P-P plot, which means that they are not following a normal distribution. Furthermore, the residuals are not independent from the dependent variable. This is visible in Figure 4-2 which is a scatter plot with the residuals plotted against the dependent variable. There is a very distinct pattern in these residuals which is caused by the correlation between both variables, which shows that the additive functional form is not suited for this model.

To resolve these issues the original model is changed to a model with a multiplicative functional form. After estimating this model it appeared to have a much better fit. Therefore the choice is made to proceed with the multiplicative functional form for the model.

The model is used to estimate the elasticity of promotions, and it this is done for all the chains at the

same time. Because the model is estimated for all the chains at the same time the amount of data points is bigger, which makes it possible to only use the weeks in the data set that had a promotion and model the elasticity of these promotions.

The dataset that is used contains only the weeks which had a promotion, which is defined as a price reduction of 10% or more. For this model all data of the different brands and chains is pooled. This means that the dependent variable is Sales in units*1000 per brand

i

in drug chain

j

in period

t

. To distinguish the difference between the chains and brands, dummy variables are included as independent variables. The model has a multiplicative functional form, it is presented in Equation 1. The multiplicative model is transferred to an additive model in equations 2 and 3 by taking its natural logarithm, to be able to run a linear regression. The main parameter of

Figure 4-1 P-P plot

(26)

interest is the B for Price, which represents the price elasticity. The other parameters of interests are the other Betas which indicate the influence of that independent variable on the dependent variable Sales.

ij ij ij ij ij ij ij ijt ijt I J ijt ijt ijt ijt ijt ijt ijt

PI

TSLP

F

TV

NPPP

PILP

D

S

1 2 3 4 5 6 7

(

      







ijt ijt ijt ijt ijt ijt

PI

S

U

PI



8ij



9ij



10ij

₎



2 1 1   

Equation 1: The proposed model

Natural logarithm transformation

)

ln(

)

(ln

)

ln(

)

ln(

(

)

ln(

1 2 3 4 1 ij ij ij ij ijt ijt ijt ijt I t ijt ijt

PI

TSLP

F

TV

S



















 

This leads to:

ijt ij ijt ij ijt ij ijt ij I t ijt ijt

PI

TSLP

F

TV

S

)

(

ln

ln(

₁ ₂ ₃ ₄ 1











 ijt

S

= Sales in units*1000 per brand

i

in drug chain

j

in period

t

. (LNVolume1000)



= Constant

ij



= Brand Elasticity, Dummy for Brand 3, Brand 4, and Brand 5, Brand 1 is the base. (DummyBrand 3, DummyBrand 4, and Dummy Brand 5)

ij

S

=Store elasticity for brand

i

in drug chain

j

Dummy for chains Retailer B or Retailer C, Retailer A is the base (DummyRetailer B, DummyRetailer C)

ijt

PI

= Price index per brand

i

in drug chain

j

in period

t

. (LNPrice)

ijt

F

= Feature spending per sub brand

i

in drug chain

j

in period

t

. (LNFeature)

it

TV

= TV Advertising spending per brand

i

in period

t

. (LNTV)

x ijt

PI

/ = Price index per brand

i

in drug chain

j

in period

t



or



x

.

ijt

TSLP

= Time since last promotion per brand

i

in drug chain

j

in period

t

. (LNTimeSinceLastPromModified)

ijt

NPPP

= Number of Promotions in previous period. (LNAmountPromoLastPeriod)

ijt

PILP

= Price index of last promotion per brand

i

in drug chain

j

in period

t

.

1 8

7 6

5

ln



ln



ln



ln





_ij

NPPP

_ijt



_ij

PILP

_ijt



_ij

D

_ijt



_ij

PI

_ijt

)

ln(

)

ln(

)

ln(

)

ln(

)

ln(

5 8 9 1 1 7 6_ij _ij _ij _ij ij ijt ijt ijt ijt ijt

PILP

D

PI

NPPP













_ 



_ 



Equation 2: Step 1 of transformation of The model

Equation 3: Step 2 of transformation of The model

(27)

ij

10 1



= Elasticity of the variable it is connected to per sub brand

i

and / or drug chain

j

ijt

U

= Error Term

4.2 Parameterization or Estimation

As the Dutch beauty care market is the object of study, data of this market is used for the parameterization of the model. The estimation is done by using the dataset of coloration products which is introduced at the beginning of this chapter by ussing linear regression in SPSS. This way the parameteres are determined, which are used in the next step of the model building process.

4.2.1 The Dutch beauty care market

The object of study in this research is the Dutch beauty care market. The definition of this market in this paper is: all fast moving consumer goods that are intended to enhance and maintain the physical appearance of the human body that are available for consumer use through the retail channel. This market is subdivided in several submarkets, LOréal is active on most of them. L‟Oréal is active on the following submarkets: Hair care, Styling, Coloration, Men Care, Body, Sun care, Face Care, Make up, Fragrances. Submarkets that L‟Oréal is not active on are: Bath & Shower, Deodorant, and Soap. Figure 4-3 Submarkets of the Dutch beauty care market

shows the annual sales figures in some of the submarkets and their growth. Coloration is part of the Hair Care submarket, which is the second largest submarket. This research focuses on the coloration market, as only data of this market is used. However, many of the brands that are object of study are active in many of the other submarkets. Most of these submarkets have large similarities with the coloration market, so valuable information for these other submarkets can also be gained from this research.

4.2.2 The Dutch coloration market

The coloration market is defined as the market for do it yourself coloration products to dye hair for both men and woman. Coloration products are generally sold in a package which contains one coloration treatment. The coloration market in the Netherlands is dominated by a couple of big companies that have a large brand portfolio with one or more umbrella brands and many brands or subbrands. In this research both the words brand and sub brand are used for the sub brands, the producers (i.e. Schwarzkopf) are also referred to as umbrella brands.

2005 Growth 2006 Growth 2007 Growth 2008 Growth

Bath and Shower Decorative cosmetics Deodorant Fragrances Hair Care Skin Care Mouth Care Shaving

(28)

Producers and brands of coloration products

The most important producers are L‟Oréal, Schwarzkopf/Henkel, KAO brands / Guhl, the Private labels and Chefaro. A short description of these producers is given below, a further analysis can be found in paragraph 3.1.1 and 3.1.2.

L’Oréal In 1909 Eugène Schueller founded the company that later changed its name to L‟Oréal and therefore

it celebrated its centenary in 2009. It all started as a research driven company, when the chemist Schueller began to design, produce and sell hair dye to Parisian hairdressers. In 1935 Schueller developed Ambre Solaire sun oil after getting a sun burn on a sailing trip. From 1957 to 1983 L‟Oréal expanded internationally under chairman François Dalle, and a large expansion of the portfolio was realized. In the twelve years after this, from 1988 to 2000, L‟Oréal became the number one in the worldwide beauty industry under the management of Sir Lindsay Owen-Jones. In the 21st century Owen-Jones remained in charge together with CEO Jean-Paul Agon. In the early days of L‟Oréal, Schueller stated that company mission was: “research and innovation in the interest of the beauty”. Nowadays the company mission is, as stated on the website: “At L‟Oréal, we believe that everyone aspires to beauty. Our mission is to help men and women around the world to realize that aspiration, and express their individual personalities to the full. This is what gives meaning and value to our business, and to the working lives of our employees. We are proud of our work.”

L‟Oréal has always been a pioneer in the beauty industry with its advertising. Already in 1931, Eugène Schueller started developing promotional events and inventing new advertising strategies. One of his ideas was creating a giant billboard to promote his hair lotion O‟Cap by draping a sheet over the face of a Parisian building. In 1932, when radio commercials were still primitive and only spoken, Schueller was the first to air a radio commercial that was sung, which was the invention of the “Jingle”. According to Schueller there were two types of advertising: publicité d’attaque, designed to raise interest, and publicité de rendement, designed to maximize sales.

L‟Oréal is divided in 5 divisions of which only the consumer division is relevant for this research as it is narrowed down to the consumer market. All L‟Oreal‟s mass brands are in this division and they have a combined market share of 52%. These brands are: L‟Oréal Paris, Garnier, Maybelline New York, and Parfumeurs Créateurs. L‟Oreal‟s coloration brands are: Excellence Preference, Nutrisse (Garnier), Couleur Exerte, and Casting Créme Gloss.

Henkel (Schwarzkopf) Henkel is a large international company originated from Düsseldorf, Germany. Henkel

is active in more than 125 countries and is active in three strategic divisions: Washing and cleaning products, Cosmetics and Adhesive Technologies.

Within the cosmetics division the largest brand is the umbrella brand Schwarzkopf, which includes several coloration products. Schwarzkopf has more small coloration brands compared to L‟Oréal, which has fewer brands with a larger market share.

KAO Brands (Guhl) KAO brands is a Japanese company with its headquarters situated in Tokyo. Its activities

(29)

Drug 93% Perfumery

3% Food 4%

Distribution channel Coloration market Netherlands 2007, 2008, 2009 - wk 1-36 Retailer A 47% Retailer C 13% Retailer B 20% Other 20%

Market shares in Turnover Coloration Drug Market Netherlands 2007 wk 1

-2009 36

Care Business. In 2009 the turnover in the European and American consumer market has decreased by 16,6%. KAO Brands‟ haircare, styling and coloration brand is called Guhl and is the third largest brand in the Dutch coloration market.

Private Labels The largest drug stores in the Netherlands all have a private label for coloration products.

However it remains unknown what the exact market share of these brands is (as the largest drug store, Retailer A, is not willing to share its sales figures for their private label products). The market share is estimated by Reed business to be around 10,4% over the year 2008.The prices of these private label products are much lower than the prices for products of the A brands and are typical “me too products”. The private labels are typical followers in the market regarding product innovations of the A brands.

Other producers Besides the large producers, a couple of very small other producers exist for niche markets.

One of them is Chefaro, which is a Dutch pharmaceutical that produces the coloration product “Just for men” which had a market share of 3,1% in 2008.

4.2.3 Distribution channels in the Dutch

coloration market

Coloration products for use at home by consumers are almost exclusively sold at drug stores in the Netherlands, 93% of the market is serviced through this channel. The remaining part is 3% at Perfume stores, and 4% through the food channel (see Figure 4-4). In this study only the drug market is researched as this is the vast majority and therefore the most relevant. Furthermore the choice is made to focus only on the three biggest drug store chains in this market. The rest of the market consists of only very small players, which are not very relevant.

4.2.4 Drug store chains in this research

The chains that are included in this researched are Retailer A, Retailer C and Retailer B. They have a combined market share of 80% of the total coloration market in the drug store channel. Thus So 20% of the coloration products in the drug market are sold at other small chains or independent drug stores, which have a very small impact on the total market.

Retailer A Retailer A is by far the biggest chain in

the Netherlands, it has a 47% market share (See Figure 4-5) in turnover of the total coloration market in

Figure 4-4 Distribution channels NL

(30)

drug channel in the Netherlands. It is part of AS Watson, which also owns the smaller Drug store chain Retailer D. Retailer A increased its amount of shops from 739 shops in week 1 of 2006 to 775 in week 36 of 2009. Retailer A has a low price / low service positioning compared to Retailer B and Retailer C. Retailer A has a high frequency of promotions which are often very deep. Most Retailer A shops have a very wide assortment with also a lot of non-drug store products, for example toys, books, clothes, food and drinks, computer articles etc.

Retailer C Retailer C is perceived as a very high service quality / high priced drug store chain. Retailer C has

decreased its amount of shops during the research timeframe from 431 in week 1 of 2006 to 388 in week 36 of 2009. Retailer C has a 13% market share (See Figure 4-5) of the Coloration market in the drugstore channel in the Netherlands.

Retailer B Retailer B has a 20% market share in turnover of the Dutch coloration market in the drugstore

channel (See Figure 4-5). Retailer B is part of the large international corporation Ahold has 509 shops. It is the second largest chain in the Netherlands when ranked by amount of shops and turnover. In 2009 Retailer B was voted best drug store of the Netherlands by Dutch consumers. Retailer B is known for its high service and quality but with a slightly higher price.

4.2.5 The dataset

The dataset which is used in this research to execute the empirical analysis contains an extensive amount of variables which are combined from a variety of different sources. In this paragraph the source of the information is pointed out, a further explanation of the variables that are derived from this data and which are used for the actual analysis is given in paragraph 4.1.1. The data is all weekly data from week 1 of 2007 to week 36 of 2009, so in total 140 weeks.

Sales, price, market share, distribution, amount of shops

The sales figures, the distribution, the price, market share and amount of shops are scanner data which is bought from ACNielsen by L‟Oréal. ACNielsen collects this information by registering all products that are scanned in retail outlets. The scanner data is weekly and is aggregated per drug store chain in the Netherlands.

Feature spending

Promotional Effectiveness

Promotional Effectiveness

A study of the effectiveness of promotional activities of

coloration products in the Dutch beauty care market.

- Public Version -

Menno Buikema

Promotional effectiveness

A study of the effectiveness of promotional activities of

coloration products in the Dutch beauty care market.

Abstract

Preface

Table of contents

1 Introduction

2 Problem definition

3 Theoretical framework

3.1 Advertising

3.1.1

Advertising defined

3.1.2

How advertising affects sales

3.1.3

Different forms of advertising and their effects on sales

3.2 Price promotions

3.2.1

Price promotions defined

3.2.2

The rise of price promotions

3.2.3

Different types of price promotions

3.2.4

General effects of price promotions and their decomposition

3.2.5

Why price promotions are offered

3.2.6

Long term versus short term effects of promotions on sales

3.2.7

Price promotion effectiveness

4 The model

4.1 Specification

4.1.1

Selecting the independent variables

4.1.2

Functional form

i

j

t

PI

TSLP

F

TV

NPPP

PILP

D

S

(







PI

PI

S

U

PI







)



)

ln(

)

(ln

)

ln(

)

ln(

(

)

ln(

PI

₎