The most effective marketing tools for premium brands and value brands.

The most effective marketing tools for premium brands

and value brands.

Measurement of the effectiveness of marketing tools and interactions in

short term and long term for premium brands and value brands

In the FMCG industry

Master Thesis

By

Thomas Boers

The most effective marketing tools for premium brands

and value brands.

Measurement of the effectiveness of marketing tools and interactions in

short term and long term for premium brands and value brands

In the FMCG industry

Master Thesis

Msc Marketing Intelligence

University of Groningen

Faculty of Economics and Business

June 18, 2018

Management Summary

The effectiveness of the marketing tools in the FMCG industry is a very well researched topic. The findings estimated the effectiveness of marketing tools in short-term and long-term. Furthermore, differences between different types of brands in marketing tools effectiveness, such as store and national brands showed or low-tier and high-tier brands, provided some new valuable insight for brand managers. In this study, the effects of marketing tools are estimated in short-term and long-term and possible interactions between each of the marketing tools. In addition, with these already existed insights and the importance of the pricing strategy of the brands nowadays, this study will provide insight on the differences for the higher priced premium brands and lower priced value brands. In the different product categories brands have their own pricing strategy and the consumer have their behavior regarding to the reaction to the marketing tools. For a large FMCG firm as P&G, which has a higher priced brand and lower priced brand in the laundry detergent market, it is crucial to know the effectiveness of the different tools. In addition, a firm as Nestlé is mainly focused on producing premium brands, knowing the differences between marketing tools is important to keep the image of a premium brand. Spending budget on ineffective tools will be waste of money and influence the demand for the product. Therefore, this study will provide insights in the effectiveness of different marketing tools in short-term and long-term, possible interaction of tools and differences between premium and value brands.

In order to find these effects, data is used from the FMCG industry. Scanner data is gathered from the three major supermarkets chains over 450 supermarkets. The dataset comprises 8 product categories, including 40 brands in which the marketing tools and volume sales are measured for 208 weeks, from 1994 till 1988. The dataset contains the marketing tools as features, display, advertisement, price and volume sales. With an error-correction model short-term and long-term effects of the marketing tools are estimated over the 40 brands separately. First, these 40 models are combined in one overall model with the

weighted mean of parameters. Second, a median split for the price is used to divide the brands within each category in premium and value brands, the parameters estimates are gathered with the weighted mean.

of features or display. Consistent with the overall outcomes, value brand will benefit on short-term from the combination of both. Contradictory, based on the total effect, the marketing tools for value brands should be used individual. For premium brands features is found to be an effective marketing tool, the individual effect and the combination with a price promotion. In addition, some differences are found between the premium brands and value brands in food and beverages. Where the value brands in food are more short-term focused, both features and display and the interaction with price. The premium brands in food are more long-term

focused with interactions of advertisement with the other marketing tools.

Preface

Two years ago, I started with my pre-master Marketing program. In this year I got exposed to statistics and marketing. After that year I chose the Marketing intelligence track to further develop skills in a more data driven marketing world. In addition, my future interest is

focused on the FMCG-sector, food and drink in particular. Therefore, this thesis helped me to further develop insight knowledge about this sector. Working with a real-time dataset from the FMCG sector and my interest for the topic were a great motivation during the process of writing this thesis. This thesis forms the end of my time as a student. I look back on two great years as a student of the Faculty of Economic and Business.

I want to thank my supervisor Maarten Gijsenberg for the time, feedback and guidance he provided. Furthermore, I would like to thank my fellow students with whom I exchanged feedback. Moreover, I would like to thank my family and friends for the support during the entire academic curriculum.

Thomas Boers

Abstract

Consumers in the FMCG industry are exposed all kind of marketing tools from a wide variety of brands. These marketing tools are individually or combined to even increase the attention and purchase probability of the consumers. Large FMCG firm offer a wide variety of brands which all have their own strategy regarding to the marketing tools. Which tool should be used to achieve the highest impact in short term and which tool is on the longer term more interesting. This study investigates the effectiveness of different marketing tools as features, display, advertising and price. Furthermore, the individual effectiveness and possible interaction are measured in the short-term and long term. In addition, a distinction is made between higher priced premium brand and the lower priced value brands. This study uses scanner data weekly gathered from the three major supermarkets from 1994 until 1998. Results show that price is the most effective tool in short term and long term. Furthermore, premium brands should use in-store promotion and price promotion separately, while value brands benefit from combining these marketing tools. These findings, together with managerial implication and limitation and future research opportunities are discussed.

Table of Content

1. Introduction 8

2. Literature review 10

2.1 Theoretical background of marketing tools ... 10

2.2 ST effect of marketing tools ... 12

2.3 LT effect of marketing tools ... 13

2.4 Synergy effect ... 16

2.5 Premium brand and value brand. ... 18

2.6 Conceptual model ... 21

3. Data description 22 4. Methodology 25 4.1 Model choice ... 25

4.2 Model specification ... 25

4.3 Insights of premium and value brands ... 27

5. Results 28 5.1 Model quality ... 28 5.2 Overall model... 29 5.3 Premium brands ... 31 5.4 Value brands ... 32 5.5 Additional insights ... 34

6. Discussion and conclusion 36 6.1 Hypotheses testing ... 36

6.2 Discussion ... 37

6.3 Managerial implications ... 41

6.4 Limitations and future research ... 43

Appendix A 45

Appendix B 46

Appendix C 47

Appendix D 48

1. Introduction

In the Fast-Moving Consumer Good (FMCG) industry companies spend billions of dollars on their advertising budgets and pricing strategy of the products. In 2014 Nestle, one of the biggest FMCG companies in the world, spent 2.75 billion dollars only in the US (Nestle, 2014). Procter & Gamble spend last year 7.6 billion dollars worldwide on TV, print, radio, internet and in-store campaigns (Procter & Gamble, 2017). Than the British-Dutch FMCG giant Unilever, who spend in 2014 over 7,7 billion dollars on their marketing campaigns for their product (Unilever, 2014). All these companies offer a broad portfolio of brands in the FMCG sector. Each year the companies face a complex decision in allocating the budgets to the different brands, each brand has its own behavior regarding to the marketing tools in supermarkets. Firms can use in-store displays, feature advertisements, price reduction and out-of-store advertisement. Key elements for a brand manager in budget estimation are which tools or combination of tools is effective for his brand. Which tool is waste of money and which tool is a good investment in short-term (ST) and future period/long-term (LT).

Brand managers can divide a proportion of their marketing budget to traditional out-of-store promotions and in-out-of-store promotions. The in-out-of-store promotions as features and display are more focused on the ST effect that drives consideration and choice, present at point-of-purchase (Chandon et al. 2009; Steenkamp and Gielens 2003; Zhang 2006). Out-of-store promotion, advertising, found to focus on countering the adverse effect of forgetting, creating brand equity focusing on the LT such as brand loyalty (Keller 2013; Naik, Mantrala, and Sawyer 1998; Sethuraman, Tellis, and Briesch 2011). Additionally, with the advertising tools, pricing is the other important tool a brand can use in influencing the ST and LT sales. This research will provide new insight in the ST and LT effect of the in-store promotions, out-of-store promotion and price. In addition to the individual effects, possible interaction between the different marketing tools will be studied.

purchase, consumers from national brands focus more on the image and perceived value of buying (Manzur et al. 2011; Sethuraman 2002). This study will provide insight to see if premium and value brands have similar behavior regarding to the effectiveness of the marketing tools.

In general, for a brand manager it is interesting to know which marketing tools should be used and which should be combined. More important, where should a brand manager from a premium focus on or how should a value brand compete to the premium brands. As mentioned, the brand portfolio of large FMCG companies is extensive, for a company as P&G it is essential to know the difference between premium brand and value brand. As example, P&G has two laundry detergents in two price ranges, the higher priced brand Ariel and lower priced brand Lenor. Using the same strategy for both brand could be ineffective for one of the brands and would influence the demand for the product. In addition, a company as Nestlé is aimed at producing premium brands and it is worthful to know how they can compete against the value brands of other companies. This study will estimate the ST and LT effect of the individual marketing tools and which tools are the best investment for premium brand and which for value brands. More interesting, which tools should be combined to create a higher ST and LT effect. In this research, the following main research questions will be addressed:

- Which marketing tool is most effective in ST and which is most effective in the LT - Which marketing tools should be combined to create synergy?

- To what extent are the findings different between national and store brands?

To answer these questions, data is used of four years of weekly data which was available from week 29 of 1994 until week 28 of 1998, within these four years 208 observations are made. The dataset included 40 brands over 8 categories. The data is collected through scanner data from the Dutch market from the three largest supermarket chains. All numbers are projections based on a sample of 350 stores. The study will employ a time-series model that measures the ST and LT effect of the different marketing tools, possible interactions between marketing tools and differences among premium and value brands.

interpreted. Than the practical and scientific relevance of the findings is mentioned, the research questions will be answered and is the research critical reflected on shortcomings and other issues in the limitations.

2. Literature review

This study focusses on the effectiveness of marketing tools in ST and LT in the FMCG industry. In addition to the overall insight, the effectiveness of marketing tools for value and premium brands is estimated. In this section, the academic literature is reviewed relevant to these subjects. With this existing literature hypotheses are derived and a graphical illustration is given of these hypotheses in a conceptual framework. First the academic background is provided of the different marketing tools. Than the influences in ST and LT is discussed. Thirdly, the possible synergies between the different marketing tools in ST and LT is provided. Furthermore, the differences among value and premium brands is discussed. Lastly, the hypotheses are combined in the graphical illustration, the conceptual framework.

2.1 Theoretical background of marketing tools

Marketing is still an art and the marketing manager must creatively marshal all his marketing capabilities to advance the ST and LT interests of product. These activities are all related to the firm such as pricing, branding, distribution, promotions or servicing. Furthermore, this interest of the product is influenced by external market forces like competition, economy or the government (Borden 1964). A firm who can influence the demand for its product by advertising will, in order to maximize profit, choose the right advertisement budget and price such that the increase in gross revenue resulting from one dollar increase in advertising expenditure or price change is equal to the ordinary elasticity of demand for the product (Dorfman and Steiner 1954). In addition, more recent studies suggest that both advertising and pricing are two of the most important elements in marketing (Bijmolt, Van Heerde, and Pieters 2005; Sethuraman, Tellis, and Briesch 2011).

intention (Mitchell and Olson 1981; Sallam and Wahid 2012). Sallam and Wahid (2012) provided understanding of consumer attitude and purchase intention by the use of endorser credibility. Focusing more on the attractiveness of expertise to enhance the consumers’ attitude toward advertising that affect consumer’s attitude to the brand and subsequently form consumer’s purchase intention. This attitude represents customers’ internal feeling of liking or disliking the advertisement (Mitchell and Olson 1981). According to Petty, Cacioppo, & Schumann (1983), changing the customers’ attitude emphasize two distinct routes. One called the central route, diligent consideration of information a customer think is central. The comprehensive, learning, and retention of product relevant information. Second, the peripheral route, is more attitude based. The customer has not personally considered the pros and cons, object is associated with positive or negative cues (Petty, Cacioppo, and Schumann 1983). So, the effect of advertisements is the customers’ attitude towards an advertisement which affects the purchase intention. In this study the advertisement is differentiated in features, display, and mass advertising as separate variables. Mass advertisement is used to reach a broad spectrum, features and displays can differ across the retailers.

Features and display are called in-store promotions, these promotions can differ across retailers. A feature is a special in-store and outside-store attention for the brand. The attention is either in the store flier or in an advertisement from the retailer in a local door-to-door newspaper or magazine. A display is a special inside-store attention for a brand, a temporary shelf on one of the aisles, or change in the brand’s regular shelf (Leeflang et al. 2015). Feature advertising and the in-store displays build new product awareness and influence the trial decision, both tools are present at the point of purchase (Papatla and Krishnamurthi 1996). According to Nordfält (2011) , the in-store displays are used as attention-capturing component, which increase trial decision or purchase intention. The study estimated that once a product is positioned in the shelf, one in a hundredth consumer realizes that they need to buy one. If, on the other hand, the store positioned the product in the middle of the racetrack with a display every eight customer realizes a need and buy the product. The increase of shelf-space and display size are both drivers of attention and increase brand sales (Chandon et al. 2009). Two third of the salience of the shelf is due to in-store marketing, one-third is due to out-of-store promotion (van der Lans, Pieters, and Wedel 2008).

advertisement, such as TV, radio and print advertising is useful to reach a broad spectrum of consumers for creating brand equity and usage imagery (Keller 2013). According to Keller, brand equity is explained by the customer-based brand equity (CBBE). CBBE is defined as the differential effect that brand knowledge has on consumer response to the marketing of that brand. CBBE occurs when the consumer has a high level of awareness and familiarity with the brand and holds some strong, favorable, and unique brand associations in memory. So, mass advertisement is mainly focused on creating brand awareness and preferred brand image in order to create brand equity.

Additionally, a firm can use pricing by influencing the demand for their product (Dorfman and Steiner 1954). This relationship is been market by the price sensitive of the customer (Bijmolt, Van Heerde, and Pieters 2005; Tellis 1988). According to Tellis (1988), the term “price sensitivity” is a latent construct to the extent to which customers vary their purchases of a product as its price changes. This price sensitivity coefficient is estimated as an elasticity. This price elasticity is significantly negative, depending the complexity of consumers’ information processing and competitive effort to reduce price sensitivity with advertising and product differentiation. Bijmolt and colleagues (2005) offered new empirical generalizations on the determines of price elasticity, based on their meta-analysis over 1851 price elasticity studies, the average elasticity is set on -2,62.

2.2 ST effect of marketing tools

The ST effect is the impact a tool will have in the current period, the effect on sales in the period of occurrence. The effect is measured by the ST elasticity, the percentage change in sales due to a 1% change in one of the marketing tools. (Köhler et al. 2017; Mela, Gupta, and Lehmann 1997).

case the difference between features and display was significant. The advertising elasticity are smaller on the ST compared to the LT, advertisement is mainly focused on creating brand equity instead of the immediate effect (Van Heerde et al. 2013; Keller 2013).

H1a: Feature has a positive effect on ST sales. H1b: Display has a positive effect on ST sales.

H1c: Feature has a higher impact on ST sales compared to displays. H1d: Advertising has a positive effect on the ST sales.

The price elasticities is found to have a higher elasticity in the ST than the LT, this price elasticity is general estimated as negative (Van Heerde et al. 2013). When a firm uses promotional prices, the magnitude is larger than the actual price elasticity (Bijmolt, Van Heerde, and Pieters 2005). Price is a strong tactic in generating sales, discounting plays a largely tactical role by generating strong bumps in sales in ST (Ataman, Van Heerde, and Mela 2010). The “bump” during a price promotion is to most extent due to brand switching (Gupta 1988). As the relative price rises, the probability of switching will increase simultaneously (Carpenter and Lehmann 1985). The increasing probability of switching behavior is due to the increase of price sensitivity of the consumers (Sloot, Verhoef, and Franses 2005). In the case study of Leeflang and colleagues (2015), the price elasticity has the highest impact of all marketing-tools in ST.

H1e: Price has a negative effect on the ST sales.

H1f: Price has the largest impact of all the marketing tools in ST.

2.3 LT effect of marketing tools

on two or more successive purchasing decisions of a consumer regarding to a given product, or the effect of an advertisement that influence the buying behavior of a consumer beyond the period of appearance (Palda 1965). This behavior can be explained by delayed-response effects and customer-holdover effects. The delayed-response effect arises from the delay between the marketing expenditure and the sales, three explanations are made. Firstly, the time between the money spend by the management or preparing an add and its appearance, the execution delay. Subsequently the noting delay, the time between a magazine being published and read. Finally, the time a customer receives the stimulus and a purchase being made, the purchasing delay. Customer-holdover effect occurs because of the retention of customers, sometimes they make repeated purchases. This effect occurs due to new customer who are attracted by the stimulus and make a repeated purchase or the increase of the average quantity purchased (Leeflang et al. 2015).

Consumers make the bulk of their final decisions in the store. Two third of the salience these final decisions are influenced by the in-store promotions, feature and display. Both tools are designed to increase the number and nature of spontaneous buying decisions (Keller 2013), or so called trial decisions (Steenkamp and Gielens 2003). The in-store promotions are available at the point-of-purchase to drive consideration ,choice and creating attention (Chandon et al. 2009; DelVecchio, Henard, and Freling 2006; Steenkamp and Gielens 2003). The new customers attracted to the brand may attribute their purchase to the promotion and not the merits of the brand per se, as a result, may not repeat their purchase when the promotion is withdrawn (Keller 2013). Therefore, the expected LT effect of both marketing tools will be lower than the ST.

H2a: Feature has less effect in LT compared to ST. H2b: Display has less effect in LT compared to ST

advertisement is TV advertisement. The TV-advertisement has a positive effect in the ST, but will even double in the LT (Hu, Lodish, and Krieger 2007). In line with this study, Sethuraman and colleagues (2011) found that the LT advertisement elasticity is higher than the ST elasticity, the estimate is found two times higher in LT. So, the expected LT effect of advertisement will be higher than the ST elasticity (Van Heerde et al. 2013; Hu, Lodish, and Krieger 2007; Sethuraman, Tellis, and Briesch 2011).

H2c: Advertising will have a higher effect in the LT, compared to the ST

In the long run, price promotions will have a negative effect on consumer’s reference price and quality perceptions. Promotional effects are short-lived and is followed by a post promotion dip, Therefore the expected price elasticity is less negative for LT compared to the ST (Bijmolt, Van Heerde, and Pieters 2005; van Van Heerde, Leeflang, and Wittink 2000; Pauwels, Hanssens, and Siddarth 2002). Price promotions have a large immediate impact on the consumers’ brand choice, but their total impact on brand choice is relatively low (Pauwels, Hanssens, and Siddarth 2002). The price promotion will have a significant effect on the repeat purchase if this price promotion will be retracted. The price promotion negatively affect the perceived quality of the product, since the benefits that were gain through price promotion incentives are limited and do not secure or confidence the consumer that a brand should fulfill the expected utility (Villarejo-Ramos and Sánchez-Franco 2005). According to Keller (2013), these promotions will affect brand equity, it will erode the perceived value. Consumers wait until the brand is on discount or on special to buy it, creating brand association as “discount” or “don’t pay the full price”. When the price promotion is retracted the higher price will negatively influence the sales. The observed effect of the customer after retraction of a price promotion can be explained by the self-perception theory (Dodson, Tybout, and Sternthal 1978). When the price promotion is retracted the higher price will negatively influence the sales. This theory is interpreted as that the individuals’ own examination of behavior and circumstances in which that behavior occurs as determination of their attitude towards an object (Bem 1972). Dodson and colleagues suggest that individuals who make purchases on price promotion, will be uncertain whether their behavior is attributable to liking the purchased product or desire to take advantage of the promotion.

Macé and Neslin (2004), the reasons for the dip after a promotion are potential indicators of consumer stockpiling and deceleration. A consumer who increase their inventories above the normal level of purchasing within that category or purchasing greater than normal quantities, represent stockpiling behavior (Neslin, Henderson, and Quelch 1985). Deceleration is the willingness of consumer to deplete their inventories below normal levels by waiting for an anticipated promotion (Mela, Jedidi, and Bowman 1998). So, The generated bumps in sales due to a promotion have an adverse effect on the LT elasticity and the expected price elasticity is less negative for LT (Ataman, Van Heerde, and Mela 2010; Pauwels, Hanssens, and Siddarth 2002).

H2d: The price elasticity is less negative in the LT period, compared to the ST period.

2.4 Synergy effect

In addition to the individual elasticities of the marketing tools, this study focuses on possible interaction between the variables. These interactions effects or the synergy effect, implies that the joint effect of the marketing variables is higher than the sum of the individual effects. The combined impact of multimedia activities such as television, print, radio, internet, direct response, sales promotions, and public relations can be much greater than the total sum of the individual effects. This synergy between multimedia activities is driven by the fact that the added value of one activity is caused by the presence of another activity (Naik and Raman 2003). The negative side of integration has to be addressed as well, negative synergy or dysfunctionality. Negative synergy indicated that the tools perform better independently and not should be integrated with each other (Pickton and Broderick 2005).

consideration and choice at the point-of-purchase (Chandon et al. 2009). Therefore, the interaction between both in-store promotion is ST focused. Whereas the out-of-store promotion, advertisement, is focused on creating on creating brand equity, LT focused (Keller 2013). Both in-store and out-of-store promotions are expected to strengthen the individual effects by an interaction in the ST and LT.

H3a: Feature together with advertising will have a positive synergy effect in ST. H3b: Display together with advertising will have a positive synergy effect in ST. H3c: Feature together with advertising will have a positive synergy effect in LT. H3d: Display together with advertising will have a positive synergy effect in LT. H3e: Feature and display will have a positive synergy effect in ST.

H3f: Feature and display will have a less positive effect in LT compared to ST.

(2005) suggest that these integrations emphasize on ST results and reduce brand loyalty. The expectation will be that the negative synergy effect will be less negative in the LT.

H3g: Features and price will have a negative synergy effect in ST.

H3h: Features and price will have a less negative effect in LT compared to ST H3i: Display and price will have a negative synergy effect in ST.

H3j: Display and price will have a less negative effect in LT compared to ST

Advertising is mainly focused on creating brand equity and brand imagery. De advertisement campaigns build awareness, conveying product information, and countering the adverse effect of forgetting. The created CBBE by advertisement will make consumers less price sensitive and turn into intense and active brand loyalty (Keller 2013). Contradictory, Kaul and Wittink (1995) suggest that consumer that are exposed to advertisements which contain price information, their price sensitivity increases. A more recent study confirmed that the interaction of a price promotion together with advertisement makes consumer more price sensitive, the effect on sales is higher. If the message is consistent, the advertisement increases the attention to the promotion which creates the benefit of interaction (Pickton and Broderick 2005). Advertisement has a LT effects estimated higher than the ST, while for price the LT will be less negative compared to the ST (Bijmolt, Van Heerde, and Pieters 2005; Sethuraman, Tellis, and Briesch 2011).

H3k: Advertisement and price will have negative synergy effect in ST

H3l: Advertisement and price will have a less negative synergy effect in LT compared to the ST.

2.5 Premium brand and value brand.

differences among brands in a product category. The differences can be related to attributes or benefits of the product, service itself, and they may be related to more intangible image considerations. Branding is all about endowing products and services with the power of brand equity, which empathize how important the brand is in performing marketing strategies. As mentioned, Keller introduced customer-based brand equity (CBBE). He defined CBBE as the differential effect that brand knowledge has on the way a consumer responds to the marketing activities of certain brand. To build CBBE, a marketer should establish a positive brand image and create brand awareness. Brand awareness consists of brand recognition and brand recall. The more a consumer “experiences” the brand, such as seeing it, hearing it, or thinking about it, the more likely he or she is to register the brand. The experience of a brand is created by different activities as name, symbol slogan, advertising and promotion, publicity and outdoor advertising, these activities all increase familiarity and awareness of the brand. Brand image is created, after creating certain level of brand awareness, through linking strong, favorable, and unique association to the brand in memory on the consumer. These associations may be either brand attributes or benefits. Brand attributes are more focused on the features that characterize a product, while brand benefits are the personal value and meaning that consumer attach to the brand. The differential response from consumer to different brands leads to CBBE, brands need to make sure that not only their brand is favorably but also unique and not shared with other brands. These unique associations will make consumer purchase most favorable and unique associated brand (Keller 2013).

perceived value. A strong brand can benefit from this price premium through image (Keller 2013). Therefore, this study focuses on the pricing strategy of the different brands. Consumers will pay a price premium for higher priced products even when the quality of the higher priced brand and the lower priced brand is the same, the so called image premium (Sethuraman 2002). The brand associations are product-related or performance related versus non-product related or imagery-related attributes. One association of these non-product related attributes is price. The pricing strategy can dictate how consumer categorize the price of the brand; as low, medium or high. Nowadays, firms place greater importance on consumers perception in their pricing strategy (Keller 2013). In this study, the data will be split based on their price, a median split will divide the brands in the lower priced brands called value brands, and the high priced brand called premium brands (Gijsenberg 2017; Van Heerde et al. 2013).

The differences between value brands and premium brands is not addressed in earlier studies. However, previous studies estimated differences among brands such as the lower priced store brands and the higher priced national brands (Keller 2013; Lemon and Nowlis 2002; Manzur et al. 2011; Sethuraman 2002; Steiner 2004). Differences are found in behavior of customers from low-tier brands and high-tier brands (Kaul and Wittink 1995) Chandon and colleagues (2009) found differences between consumer response to low-market-share brands and high-market share brands. Whereas brand equity influence the willingness to buy and being loyal to certain brand, differences are found between high-equity brands and low equity brands (Sloot, Verhoef, and Franses 2005).

customer attempt to maximize the quality/price ratio of their purchase. When a higher priced brand as national brand has a promotion, the reaction will be higher than the promotion of lower priced store brands. Due to the higher permanent price of national brands consumers will react more to the promotion. The effect of the promotion on brand loyalty is for national brand lower compared to the store brands (Manzur et al. 2011).

Furthermore, differences are estimated between multiple brands in responses to the in-store promotions and out-of-in-store promotions. The image of the premium brands as national brands can be maintained or increased by marketing activities that enhance perception of brand equity, the advertisement tool in this study (Sethuraman 2002). These national brands benefit from the effect of marketing tools separately, while store brands benefit from a synergy effect (Lemon and Nowlis 2002). In addition, the higher priced national brands mostly use product innovation and advertising to attack store brand instead of using price reductions (Keller 2013). In line with these finding, the interaction of price with each of the in-store promotions are found to be higher for low-tier brands compared to the high-tier brands (Kaul and Wittink 1995). Where the higher priced brands or high tier brands more focuses on creating brand image, lower priced brands or low-tier brands are more focused on the in-store promotion on creating attention (Keller 2013; Manzur et al. 2011; Sethuraman 2002).

Concluding, these studies suggest that there are differences in marketing tools reactions among customers of different type of brands. Contradictory, there is no clear scientific evidence for value brands and premium brands. As a result, no hypothesis will be formulated for this research. However, the effects of premium brands and value brands will be tested. All the previous specified hypothesis will be tested for premium and value brands.

2.6 Conceptual model

3. Data description

In order to answer the research questions, data is used of four years of weekly data which was available from week 29 of 1994 until week 28 of 1998, within these four years 208 observations are made. The data is collected through scanner data from three major supermarket chains in the Netherlands. All numbers are projections based on a sample of 350 stores. This study used five different measures: volume sales, feature only, display only, advertising and price. In total 40 brands are included over eight different categories, table 1 provides an overview of the chosen categories, the cumulative market share, average market share and the standard deviation of the market share of the chosen brands over the category based on volume sales.

regarding to the standard deviation. Sparkling Water shows higher differences in market share for the brands, the standard deviation in market share is the highest all categories with 0,29. The Cola and Chips category show slightly similar behavior in deviation of the market share, while their average market share differs strongly. The similarity in standard deviation is also found in the categories of Orange Soft Drinks and Cottage Cream. In addition, the categories of Beer and Yoghurt show slightly similarity in the standard deviation of the market share.

Table 1

Overview of included categories and brands

Category Number of brands Cumulative market share Average market share Standard deviation of the market

share Example brands Cola: 6 80,62% 13,44% 0,14 Coca-Cola, Pepsi

Beer: 8 83,57% 10,45% 0,09 Heineken, Grolsch

Orange Soft drink: 6 55,97% 9,33% 0,07 Fanta, Sisi

Sparkling water: 3 90,47% 30,16% 0,29 Spa, Sourcy

Yoghurt: 5 48,67% 9,73% 0,08 Melkunie, MonaVifit

Cottage Cream: 7 71,48% 10,21% 0,07 Danone, Almhof

Chips: 3 73,24% 24,41% 0,13 Croky, Smith

Sweets: 3 31,49% 10,50% 0,04 Mentos Fruit, Fruitella All the brands that are included in the analysis are national brands. One of the requirements which were assigned for including the brands is that all brands were available in market for the full period. Beside the require of selling sufficiently enough, an advertisement (mass advertisement, feature, and displays) threshold is set. The threshold for the other marketing tools is set on at least ten percent of the weeks. Both thresholds, sales and advertisement, were set to obtain reliable estimates. In table 2 an overview is given of the advertisement spending’s and the price in which a unit is sold in the supermarket, divided both brands in premium and value brands.

Table 2 General Descriptive

Category Premium Value Category Premium Value

Cola: 3 Brands 3 Brands Yoghurt: 2 Brands 3 Brands

Price € 1,51 € 1,33 Price € 3,41 € 1,92 Adv. 187307,93 35660 Adv. 4834 1567 Feat. 5,9303 2,5863 Feat. 0,6884 1,8987 Disp. 9,2511 2,7119 Disp. 0,2809 0,1423

Beer: 5 Brands 3 Brands Cottage Cream: 3 Brands 4 Brands

Price € 2,74 € 2,35 Price € 9,27 € 8,20 Adv. 174.434 176.859 Adv. 2556 4019 Feat. 2,433 2,4851 Feat. 1,0036 2,3461 Disp. 4,8919 1,8839 Disp. 0,2492 0,3755

Orange Soft Drinks: 3 Brands 2 Brands Chips: 1 Brand 2 Brands

Price € 1,51 € 1,31 Price € 9,25 € 9,49 Adv. 51120 11170,5 Adv. 45700 17386 Feat. 2,943 2,817 Feat. 5,9943 3,315 Disp. 3,7946 1,8945 Disp. 11,051 7,375

Sparkling Water: 2 Brands 1 Brand Sweets: 2 Brands 1 Brand

Price € 0,90 € 0,58 Price € 16,08 € 13,60 Adv. 28729 15019 Adv. 5275,2 27020 Feat. 1,343 1,0831 Feat. 1,0668 0,9206 Disp. 1,1283 0,7074 Disp. 3,8 2,712 of brand sales in stores where a product is featured or displayed, divided by the sum of category sales in all stores. Finally, prices and advertising expenditures are adjusted for inflation using Consumer Price Index data from the CBS.

4. Methodology

In this section, we discuss the methods that are used in order to answer the research questions. Given the research objectives of this study, multiple challenges arise in modelling. First, the ST and LT effect of each of the marketing tools are estimated. Second, the interaction between the marketing tools is included, in ST and LT. Finally, the differences among premium brands and value brands is estimated.

4.1 Model choice

To assess the impact of the marketing tools on the sales in ST and LT, an error correction model is specified (Fok et al. 2006; Gijsenberg 2014; Van Heerde, Helsen, and Dekimpe 2007). Before specifying the model, this study tested for stationarity to test for certain trends in the time-series data. The tests used for stationarity are the tests of Levin, Lin, and Chu (2002) and Im, Pesaran, and Shin (2003). Both test assume that the series are cross-sectional independent. In both of the test the H0 is rejected, all of the five series are significantly stationary. Now the five series are significantly stationary, the LT parameters can not only be interpreted as the permanent effect of permanent changes, but also as the cumulative effects of temporary changes (Gijsenberg 2014).

4.2 Model specification

Where

∆ The first difference operator 𝑆𝑎𝑙𝑒𝑠₍₎ Volume sales of brand b in week t

𝛼 Intercept

𝐹𝑒𝑎𝑡₍₎ Features of brand b in week t 𝐷𝑖𝑠𝑝₍₎ Displays of brand b in week 1 𝑃𝑟𝑖𝑐𝑒() Price of brand b in week t

𝐴𝑑𝑣₍₎ Mass advertisement of brand b in week t

𝐶𝑜𝑚𝑝𝑃𝑟𝑖𝑐𝑒₍₎ Average price other brands in premium or value brands part, control variable 𝐶𝑜𝑚𝑝𝐴𝑑𝑣₍₎ Average advertisement (mass, features and display) other brands in premium or

value brands part, control variable π₍ Adjustment effect for brand b 𝑄₍₎ Dummy for seasonality

The variables are specified as natural logarithms, with this transformation the estimates can be interpreted as elasticities. The first part, 𝛽_/(01 _{until 𝛽}

=(01, are the ST elasticities of the marketing tools and the 𝛽_@(01 _{until 𝛽}

/F(01 are the ST elasticities of the interaction between tools. The ST variables are estimated as the first-differencing operator. According to Fok et al. (2006), the first-differencing operator of the ST variable is calculated as ∆𝑋() = 𝑋()− 𝑋()M/.

The LT effect of marketing are estimated with 𝛽_//(P1 _{until 𝛽}

/=(P1 , the LT elasticities of the interaction are estimated with 𝛽_/@(P1 _{until 𝛽}

/E(P1 . This LT impact reflects the cumulative effect of a one-period shock of one of the marketing tools and the interactions (Van Heerde et al. 2013). This cumulative or LT effect is the relationship between the natural logarithm of lagged sales and the sum of all the natural logarithm of the lagged marketing tools and interactions between brackets (Fok et al. 2006). The LT effect of the marketing tools is estimated with the adjustment effect π multiplied with the all the lagged variables between square brackets. To estimate the individual LT parameters the lagged estimates of the marketing variables are divided by – π, e.g. Z[[\]^

According to Little (1970), a model has four different criteria for an appropriate model, one of these criteria is completeness. This criterion is a relative concept, all important variables should be included in the model. In a market competition has also influence on the sales of a brand. When competition matters, this variables should be included in the model (Leeflang et al. 2015). For unbiased testing of the effects from brand, this model controls for competitive effects (Nijs et al. 2001). The effect of competitors is in most studies addressed as competitive reactions (Leeflang and Wittink 1996; Nijs et al. 2001; Steenkamp et al. 2005). The competitor price variable is a market-share-weighted average price and the competitor advertisement is the total advertisement spending of all competing brands merged in one variable. The competitor variables are included as ST, 𝛿_/01 _{and 𝛿}

601. For controlling for the LT effect 𝛿:P1 and 𝛿=P1 are included.

In order to control for differences due to possible seasonality, dummy variables of the quarters are included in the model. It is likely that the sales peaks of beverages will be within the summer period and the sales will lower in the winter period. For the four quarter, three dummy variables are included: 𝛿@, 𝛿B, and 𝛿C.

4.3 Insights of premium and value brands

5. Results

This section gives answers to the research questions and examines if the stated hypotheses are significant. First, the quality of the model is discussed. Second, the overall model is estimated, in which all the 40 brands are included. Finally, the model for premium and value brands are estimated separately.

5.1 Model quality

Before discussing the results of the error correction model, the quality of the model is examined. The overall performance of the model is judged by the R-square. In table 3 an overview is provided with the different categories and the quality of the overall model, premium model and value model.

Table 3 Quality of the models

Category Premium Value Category Premium Value

Cola: 3 Brands 3 Brands Yoghurt: 2 Brands 3 Brands

R-square 0,7279 0,8205 R-square 0,4379 0,4290

Beer: 5 Brands 3 Brands Cottage Cream: 3 Brands 4 Brands

R-square 0,3968 0,4973 R-square 0,5259 0,7094

Orange Soft Drinks: 4 Brands 2 Brands Chips: 1 Brand 2 Brands

R-square 0,6179 0,8022 R-square 0,7515 0,8004

Sparkling Water: 2 Brands 1 Brand Sweets: 2 Brands 1 Brand

R-square 0,4052 0,4486 R-square 0,7420 0,6161

R-square

Overall model: 0,5979

Premium brands: 0,5633

Value brands: 0,6379

strong, value brands models are performing better than the premium brands. For Yoghurt’s premium brands model preforming slightly better than value brand with the model quality.

The performance of the overall model is classified as good with a R-square of 0,5979, since Gijsenberg (2014) classified a model with an average R-square of 0,395 as good. Similarly, the models of both premium brands and value brand can be classified as good, respectively 0,5633 and 0,6379.

5.2 Overall model

In the overall model, the parameter meter estimates are the weighted average of all the 40 brands. For the significance level the added Z method is used over all the 40 brands. In line with Bijmolt, Van Heerde, and Pieters (2005), the price elasticity is significantly negative and close to the price elasticity of -2,62 from the meta-analysis (𝛽_{aaaaa= -2,4359, P < 0,01). The in-/}01 store promotions have a significantly positive impact on the volume sales, with an increase in volume sales by displays (𝛽_{aaaaa= 0,0222, P < 0,05) and even stronger impact by features (𝛽:}01

=01 aaaaa= 0,0309, P < 0,01). For the out-of-store promotion, advertisement, no significant effect is found in the ST.

This study included multiple interaction in the model to test for possible synergies. The effect of an in-store displays becomes slightly stronger with an out-of-store promotion as advertisement as implied by the positive synergy effect (𝛽aaaaa= 0,0003, P < 0,05). Whereas, the _@01 other interaction with advertisement found to be insignificant. Contradictory, the interaction of displays and price is found significant, both elasticity decrease with certain interaction (𝛽_{aaaaa= -E}01 0,0408, P < 0,05). With a price promotion, display would strengthen this ST effect. Consistent with the interaction by price and displays, the use of feature with a price promotion strengthen this promotion. The elasticity of features and price will decrease by interacting of both tools (𝛽_{aaaaa= -0,0540, P < 0,01). /F}01

Furthermore, this study is focused on the LT effect of the different marketing tools and the effect of interactions in the LT. Consistent with the expectation, this study found that the effect of price elasticity in LT is negative and less strong than the ST elasticity (𝛽aaaa= -1,5769, _/6P1

P < 0,01). For the in-store promotion only the elasticity of features is found to have a LT effect.

was found to be insignificant. Contradictory, multiple interactions found to significantly affect the volume sales. The interaction of displays and advertisement will increase in the LT, the interaction is higher compared to the ST interaction (𝛽aaaa= 0,0008, P < 0,05). Contradictory, _/@P1 other interaction with both tools are found to be insignificant. The interaction between price and the in-store promotion with features is found to be negative in the LT, this interaction has decreased compared to the ST interaction (𝛽aaaa= -0,1739, P < 0,01). _6FP1

Table 4

Overall across-brand parameter estimates

Weighted

Beta's Z-scores P-Value

5.3 Premium brands

The 8 categories are divided individually with a median split within the specific category based on the price of the average price of the brands. The higher priced brands, price higher than the median within the category, are called premium brands. To summarize the results of the 21

Table 5

Overall premium brands parameter estimates

Weighted

Beta's Z-scores P-value

premium brands, the weighted mean of parameters is used to summarize the parameter estimates and the Rosenthal’s added Z method for the significance levels across the premium brands. Table 5 provides an overview with all the parameters estimates of the premium brands.

In line with expectation, the ST price elasticity is negative and lower than the overall model (𝛽_{aaaaa= -1,9316, P < 0,01). The individual effect of features for premium brands was found /}01 to be significantly positive (𝛽_{aaaaa= 0,0433, P <0,01). These elasticities of features and price will :}01 decrease if the in-store promotion tool will interact with price, while feature will strengthen a price promotion (𝛽aaaaa= -0,0837, P < 0,01). Contradictory, displays and advertisement have no _/F01 effect on the ST sales. An interaction between both in-store promotions, display and features, will strengthen both individual elasticities with a positive synergy effect in ST (𝛽_{aaaaa= 0,0007, P D}01 < 0,05). In addition, the model found a marginally significant interaction between out-of-store advertisement and in-store display, this interaction is positive and slightly increases both elasticities in the ST (𝛽aaaaa= 0,0003, P = 0,1193). _@01

In line with the overall model, the LT feature elasticity is found to have a stronger effect than the ST elasticity (𝛽aaaa= 0,1467, P < 0,01). Whereas the LT elasticity of feature will strongly _/:P1 decrease by the interaction with price (𝛽aaaa= -0,2790, P < 0,01). Contradictory, price, _6FP1 advertisement and displays are found to be insignificant in the LT. However, the interaction between advertisement and display is found to be stronger in the LT compared to the interaction effect in the ST (𝛽aaaa= 0,0014, P < 0,05). _/@P1

5.4 Value brands

With the median split, 19 brands are divided as the lower priced brand, these are used in the parameters estimation of value brands. Table 5 provides an overview with the parameter estimates of the value brands.

display negatively influence both elasticities (𝛽aaaaa= -0,0004, P < 0,10). Additionally, the price _D01 will have a negative impact on both instore promotion, the interaction with display or features with price will decrease both elasticities (respectively 𝛽aaaaa= -0,0633, P < 0,01 and 𝛽E01 aaaaa= -0,0333, /F01

P < 0,05). Contradictory, the positive effect of display will increase by combining with

out-of-store promotions, the interaction with advertisement will increase the display elasticity (𝛽aaaaa= _@01 0,0003, P < 0,05). The combination of advertisement with the other two marketing tools is found to be insignificant on the ST.

Table 6

Overall value brands parameter estimates

Weighted

Beta's Z-scores P-value

Inconsistent with the premium brands, the price elasticity of value brands has an impact in the LT (𝛽aaaa= -2,6101, P < 0,01). The rest of the individual effect are found to be insignificant. In _/6P1 addition to the insignificance individual effects, only one interaction is found to be significant in the LT. The interaction between display and features will even increase in the LT, the interaction will negatively influence the volume sales in the LT (𝛽aaaa= -0,0018, P < 0,05). _/DP1

5.5 Additional insights

Adjoining the insights for premium brands and value brands, this study provided additional insights for premium brands in food, premium brands in beverages, value brand in food and value brands in beverages. Table 1 provides an overview with the different food and beverages categories, both have four different categories. In appendix A until D all the overviews with the parameter estimates of the four different models are shown.

In appendix A an overview is provided for the value brands in food. For these 10 brands, a price elasticity is found which is slightly higher than the overall value brand model (𝛽aaaaa= -_/01 2,9674, P < 0,01). For the Value brands in food both in-store promotion features and displays are found to have significantly positive effect on the volume sales, where displays have a stronger ST effect than features (respectively 𝛽aaaaa= 0,0540, P < 0,05 and 𝛽_:01 aaaaa= 0,0751, P < 0,05). ₌01 Whereas, the interaction between both in-store promotion is found to be insignificant. In line with the overall value brands model, the interaction with price and one of the in-store promotions features or displays will decrease the effect of both (respectively 𝛽aaaaa= -0,0543, P _E01 < 0,01 and 𝛽aaaaa= -0,0352, P < 0,05). In the LT, the price elasticity will decrease compared to _/F01 the ST elasticity (𝛽aaaa= -2,5577, P < 0,01). The elasticities of both in-store promotion will _/6P1 increase in their effectiveness in the LT compared to ST, where the effect of features is found to be higher than the displays (respectively 𝛽aaaa= 0,3173, P < 0,01 and 𝛽_/:P1

//P1

aaaa= 0,2246, P < 0,1). The negative interaction founded with one of the in-store promotions and price in ST is even more negative in LT. Features and price found to have a more negative interaction than display and price (respectively 𝛽aaaa= -0,1968, P < 0,01 and 𝛽_6FP1

/EP1

For the value brands in beverages the Appendix B gives all the parameters estimates with the p-values. In comparison to the value brands in food, the beverages were found to have a slightly lower price elasticity (𝛽_{aaaaa= -2,6356, P < 0,01). The other marketing tools are found to be /}01 insignificant in the ST. The combination of the in-store promotions with advertisement is found to have a positive interaction, features was found to have a slightly higher interaction effect with advertisement than display (respectively 𝛽aaaaa= -0,0005, P < 0,01 and 𝛽_B01

@01

aaaaa= 0,0004, P < 0,05). Than in the ST two interactions were found significantly negative, the displays will negatively interact with features and with price which is the strongest negative interaction (respectively 𝛽_{aaaaa= -0,0009, P < 0,05 and 𝛽D}01

E01

aaaaa= -0,0762, P < 0,05). Inconsistent with the previous models, the LT elasticity of price will be even more negative than the ST elasticity (𝛽aaaa= -2,6533, P < 0,01). Contradictory to the value brands in food, features is found to have a _/6P1 insignificant LT effect. Furthermore, a negative LT effect of display is found, on the LT display will negatively affect volume sales (𝛽aaaa= -0,0566, P < 0,10). This negative effect on volumes _//P1 sales of display will be even higher if displays will interact with features (𝛽aaaa= -0,0023, P < _/DP1 0,05). Furthermore, advertisement and feature were found to have a slightly positive interaction (𝛽aaaa= 0,0003, P < 0,05). _/BP1

Appendix C reports the parameters estimates of the premium brands in beverages. This study found a smaller elasticity in the ST for price compared to the overall premium brands model (𝛽aaaaa= -1,6726, P < 0,01). For these brands no LT effect of price is found. Furthermore, _/01 the in-store promotion of features was found to have significantly positive effect on the volume sales in the ST (𝛽_{aaaaa= 0,0341, P < 0,01). For display and advertisement none individual effect :}01 was found in ST and LT. Whereas the founded ST effect of features will even increase by the interacting with displays (𝛽_{aaaaa= 0,0009, P < 0,05), the interaction with price will decrease both D}01 elasticities (𝛽aaaaa= -0,0826, P < 0,01). In ST no significant interaction is found with _/F01 advertisement. Whereas in the LT the combination of an in-store promotion as displays and features with the out-of-store promotion advertisement is found significantly positive (respectively 𝛽aaaa= 0,0020, P < 0,01 and 𝛽_/@P1

/BP1

Finally, the premium brands in food will be discussed. The overview of the parameter estimates is provided in the Appendix D. The ST elasticity of price for premium brands in food is found to be higher than the premium brands in beverages respectively (𝛽_{aaaaa= -2,5489, P < 0,01). The /}01 ST elasticity of features is found to be higher for premium brands in food than the elasticity of premium brands in beverages (𝛽_{aaaaa= 0,1817, P < 0,05). The other in-store promotion displays, :}01 found to have a slightly lower positive elasticity on the ST (𝛽_{aaaaa= 0,1620, P < 0,10). Than in the =}01 ST this study found two negative interaction with features, advertisement and price both negatively interact with features (respectively 𝛽aaaa= 0,0004, P < 0,10 and 𝛽_BP1

/FP1

aaaa= -0,0866, P < 0,10). In the LT elasticities price was found to decrease in impact compared to the ST elasticity (𝛽aaaa= -1,7985, P < 0,05), this elasticity will increase in impact when the price is interacting _/6P1 with features (𝛽aaaa= -0,2416, P < 0,10). Furthermore, the rest of the tools and interactions are _/BP1 found to be insignificant. For these types of brand none LT effects of the marketing tools are found, except for price.

6. Discussion and conclusion

In this last section the outcomes of the models are discussed. First, an overview is provided with the stated hypotheses and their result. Second, a summary is provided from the outcomes in order to answer the research questions in the discussion. Furthermore, the results from this research provides several useful managerial implications for marketing. Lastly, this study has some limitations and these need to be acknowledged.

6.1 Hypotheses testing

Table 7

Overview of the hypotheses tested

Hypotheses Overall Premium

brands

Value brands H1a Feature has a positive effect on ST sales. Supported Supported Supported

H1b: Display has a positive effect on ST sales. Supported Rejected Supported

H1c: Feature has a higher impact on ST sales compared to

displays.

Supported Rejected Rejected

H1d: Advertising has a positive effect on the ST sales. Rejected Rejected Rejected

H1e: Price has a negative effect on the ST sales. Supported Supported Supported

H1f: Price has the largest impact of all the marketing tools in ST. Supported Supported Supported

H2a: Feature has less effect in LT compared to ST. Rejected Rejected Rejected

H2b: Display has less effect in LT compared to ST Rejected Rejected Rejected

H2c: Advertising will have a higher effect in the LT, compared to

the ST

Rejected Rejected Rejected

H2d: The price elasticity is less negative in the LT period,

compared to the ST period

Supported Supported Supported

H3a: Feature together with advertising will have a positive

synergy effect in ST.

Rejected Rejected Rejected

H3b: Display together with advertising will have a positive

synergy effect in ST.

Supported Supported Supported

H3c: Feature together with advertising will have a positive

synergy effect in LT.

Rejected Rejected Rejected

H3d: Display together with advertising will have a positive

synergy effect in LT.

Supported Rejected Rejected

H3e: Feature and display will have a positive synergy effect in

ST.

Rejected Supported Rejected

H3f: Feature and display will have a less positive effect in LT

compared to ST.

Rejected Rejected Supported

H3g: Features and price will have a negative synergy effect in ST. Supported Supported Supported H3h: Features and price will have a less negative effect in LT

compared to ST.

Rejected Rejected Rejected

H3i: Display and price will have a negative synergy effect in ST. Supported Rejected Supported

H3j: Display and price will have a less negative effect in LT

compared to ST.

Rejected Rejected Rejected

H3k: Advertisement and price will have negative synergy effect

in ST.

Rejected Rejected Rejected

H3l: Advertisement and price will have a less negative synergy

effect in LT compared to the ST.

Rejected Rejected Rejected

6.2 Discussion

brands were measured for 208 weeks at the end of the 90’s. Subsequently, these previous and new insights are discussed in this section. The ST effects, LT effect, synergy effects, and differences among premium brands and value brands will structure the discussion.

In line with earlier studies, the price elasticity is found to be negative, the elasticity of this study is slightly similar to the meta-analysis. In addition, the LT effect of price is found to be less negative compared to the ST. (Ataman, Van Heerde, and Mela 2010; Bijmolt, Van Heerde, and Pieters 2005). When a brand has a price promotion this will positively influence the sales in ST. This effect or the small “bump”, the promotion plays a largely tactical role by generating ST sales which is in line with previous studies (Ataman, Van Heerde, and Mela 2010; Van Heerde et al. 2013). These ST effect of price promotions will even increase by using one of the in-store promotions. A price promotion offer with display or features will have a higher impact on sales (Lemon and Nowlis 2002; Zhang 2006), where the combination with features has a larger impact than the combination with displays. In addition, the interaction between feature and price is found to be more negative in the LT, this will have negative effect on sales when the promotion is retracted. The elasticity is negative, when the promotion is retracted the price rises and sales will decrease. This is similar for the negative LT of the price elasticity. Than for displays no effect or interaction with features and price is found in the LT. These missing effects could be explained by the fact that display and the promotions are only available at the point-of-purchase (Chandon et al. 2009; Keller 2013). In addition, this behavior can be explained by the self-perception and post promotion dip. Consumers increased their inventories or wait for the next promotion. Furthermore, consumer who make a purchase on price promotion are uncertain whether their behavior is due to liking the product or desire to take advantage of the promotion (Dodson, Tybout, and Sternthal 1978; Keller 2013; Macé and Neslin 2004).

Furthermore, the effect of display will increase by the use of advertisement (Pickton and Broderick 2005; Zhang 2006). The interaction between advertisement and display will increase in the LT, which is in line with the expectations. In contrast with the expectation, the advertisement has not an individual effect in ST and LT. Partly in line with these outcome, Chandon and colleagues (2009) suggested that the out-of-store promotion influences visual attention much less than the in-store promotions. Two third of the salience is due to in-store promotion, while one-third is due to out-of-store-promotions (van der Lans, Pieters, and Wedel 2008). In addition, Keller (2013) suggested that advertisement is focused on creating brand equity and brand imagery, instead of the immediate effect. Furthermore, the missing advertisement effect can be explained by the frequency of advertising exposure. Consumer get confronted with the advertisement such as TV, radio, newspaper on a daily basis, or multiple times a day. The increase in frequency of advertisement exposure increases the probability of brand choice, however at a decreasing rate. Therefore, at certain time the more exposure will have less effect to the brand choice of the customers (Tellis 2009).

Furthermore, the study estimated differences among premium brands and value brands. In line with Steiner (2004), this study estimated that consumers of value brand are more price sensitive than consumer of premium brands, the sensitivity is the lowest for premium brands in food. The difference between brands can be explained by the switching behavior from consumers of value brands, consumers are more sensitive to price changes. This finding is in line with the behavior of low-tier brand or store brands consumer and high-tier or national brand consumers (Sloot, Verhoef, and Franses 2005; Steiner 2004). The LT effect of price is only negatively affecting the sales of value brand, slightly different from the ST effect. In line with these outcomes, Manzur and colleagues (2011) estimated that the effect of a price promotion on brand loyalty is lower for higher priced national brands compared to the lower priced store brands. Furthermore, are the consumer for higher priced brands as national brand more loyal to the brand compared to the lower priced brands as store brands. For premium brands in beverages the LT effect is found, this effect is lower compared to the value brands in beverages and food.

the effect is more than two times higher than value brands. In addition, a positive LT effect is found for features of premium brands, the total effect is the highest for premium brands in food. These ST and LT effect can be explained by the fact that a premium brand is focused on creating brand equity by awareness and image. As mentioned, feature is a printed door-to-door flyer and printed advertisement influence brand equity (Keller 2013). Additionally, the interaction between both in-store promotion tools is found to be positive for premium brands and negative for value brands. The effect for value brands is slightly negative in the ST and even more negative in the LT. For value brands this effect can be explained by the price-cut proxy effect, consumer lack of motivation to carefully process the information of the display and/or feature. When consumers realize that a display and/or feature is not accompanied with a price promotion the positive effect of sales will disappear even in the LT (Zhang 2006).

Concluding, overall the price is the most effective tool for the consumer in the ST and LT effect. This ST effect will even increase by the use by one of the in-store promotions, feature and display. For the LT effect the use of a price promotion with an in-store feature will be most effective. Furthermore, differences are found between premium brands and value brands. The consumers of value brands are more price sensitive, which will even increase by the use of the in-store promotions. For premium brands this sensitivity is lower, while the interactions are also less effective. In addition, features are found to positive effect the ST and LT sales. Value brands in drink are more ST focused by different interaction, while premium brands in food found to have more LT interaction effects. Thus, this study estimated the most effective tool in ST and LT, possible interaction and differences between premium brands and value brands.

6.3 Managerial implications

In addition to the theoretical implications mentioned in the previous section, managerial implications are found. The results of this study provide several opportunities for brand managers in the FMCG-sector. A brand manager should carefully consider their expenditure in marketing tools based on their brand characteristics.