The online supermarket shelf

(1)

The online supermarket shelf

The impact of different online shelf layouts on

shelf evaluation and shopper behavior

(2)

2

The online supermarket shelf

The impact of different online shelf layouts on

shelf evaluation and shopper behavior

Date: May 2013

Author: Karel Broer

Studentnumber: s1270974

Adress: Albert Cuypstraat 85-1

1072 CP Amsterdam Telephone number: +31 (0) 6 47 81 43 26

Email: kwabroer@gmail.com

University: University of Groningen

Department: Faculty of Economics and Business

Specialization: Msc. Business Administration Marketing Management Qualification: Master Thesis

(3)

3

MANAGEMENT SUMMARY

The supermarket shelf is an important tool to influence shopper behavior. Prior research has shown that the layout of the supermarket shelf can have a strong impact on decision making and sales. Over the past decade, the way people shop has changed. Due to the rapid development of the internet, consumers buy more and more products online. Online supermarket retailers use different online shelf layouts to present their products. Little research is done to investigate the effects of different online supermarket shelf layouts. Therefore, this research is dedicated to investigate the impact of different online supermarket shelf layouts on the evaluation of the shelf and shopper behavior. An experiment website is developed where the presentation format and the sorting of products are manipulated to investigate the impact on shelf evaluation and actual shopper behavior. The results suggest that the matrix presentation provides more benefits to shoppers than the list presentation format. Furthermore, the finding suggest that the way that products are sorted only had a very small impact on shelf evaluation and shopper behavior.

Keywords: online grocery shopping, online product presentation, shelf layout, online supermarket

(4)

4

PREFACE

This thesis is the conclusion of my Master Business Administration specialization Marketing Management at the University of Groningen. I would like to thank my supervisor prof. dr. Laurens Sloot for his patience, useful insights and guidance during the writing of this thesis. Furthermore, i would like to thank my second supervisor dr. Erjen van Nierop for his fast and useful feedback to finalize this thesis.

Additionally, I would like to thank my colleagues at NORM for their support during the writing of this thesis. I would like to give special thanks to Peter van Keulen for his support and for giving me the opportunity to write my thesis at the office.

Last but not least, I would thank my friends and family for supporting me. I would like to give special thanks to my parents and my girlfriend who always believed in me. Without their motivation and unconditional support this thesis would not have been possible.

Karel Broer,

(5)

5

1. INTRODUCTION

1.1 Background

When it comes to grocery shopping consumers have many options to choose from. A supermarket typically carries between 20,000 and 40,000 different items made by over 500 companies (Levy and Weitz, 2007). Previous research has shown that many of the purchase decisions are made in-store. According to Underhill (2000) supermarkets are places with a high rate of impulse buying, where 60 to 70 percent of the purchases are unplanned. According to research of EFMI Business School, 35 to 40 percent of purchase decisions are made in-store (EFMI Business School, 2010). Although there is a difference in the results with regard to in-store decision making, it is clear that a large amount of the purchase decisions are made in-store.

Motivating consumers to make unplanned, impulse purchases and providing them with a satisfying shopping experience are important examples of the objective to influence customer buying behavior for retailers (Levy and Weitz, 2007). Because of the high rate of in-store decision making in the supermarket, it is important for supermarket retailers to grab the attention of the consumer at the point of purchase. The supermarket shelf is an important factor that influences the buying behavior of consumers. According to Breugelmans et al. (2007) shelf display can make consumers' choice decisions much easier and plays an important role in attracting customer attention. Therefore, the supermarket shelf is an important tool for supermarket retailers (and manufacturers) to influence consumers' choice decisions.

Over the last years, the way people shop has changed. Due to the rapid development of the internet channel, consumers are not only able to shop at traditional physical stores, but also online at web shops. The emergence of the internet channel enabled consumers to shop for all kind of products without having to personally visit a physical store. This new type of shopping comes in different names like electronic shopping, e-shopping, internet shopping, online shopping, or web-based shopping.

(9)

9

the Dutch population has ever bought something over the internet. A recent study about the use of shopping channels in the Netherlands conducted by Blauw Research (2011a) showed that the store channel is still the largest shopping channel in the Netherlands. 63.5% of the purchases were made in the store (See Figure 1). The internet is the second largest retail channel in the Netherlands and almost 1 out of 3 purchases are made over the internet (30.6%). Furthermore, despite the economical crisis in the Netherlands, the online spending of products and services has increased with 9% to € 8,98 billion in 2011 (Blauw Research, 2011b). So, it is clear that the internet channel has become an important channel for retailers.

Figure 1 - Use of shopping channel in 2011 in the Netherlands (Blauw Research, 2011a)

The online channel is rapidly gaining ground on the store-based channel, but when it comes to grocery shopping, the distribution of the shopping channels is different. In the Netherlands, the vast majority of the grocery purchases are still done by using the store based channel. Recent research in the Netherlands showed that 94 % of all grocery purchases are made in a store-based shopping channel (Blauw Research, 2011a). The research also shows that only 4% of the grocery purchases are made over the internet. Although 96% percent of the Dutch population has access to the internet channel, only 4% uses the internet channel for purchasing groceries. Therefore, it can be argued that online grocery shopping is still in its infancy in the Netherlands.

1.2 Problem identification

Supermarkets are self-service stores where customers buy products without the help of service personnel. If customers are not be able to see a product they cannot buy the product. Therefore, the way products are presented is an important issue for supermarket retailers and manufacturers. In the past, many research has been done to investigate the impact of shelf display on choice decisions of consumers in traditional brick-and-mortar supermarkets. The position of a product on the shelf or the amount of space that a product receives (number of facings) can have a strong impact on

(10)

10

consumer choices. For example, items are more likely to be chosen if they are placed on more prominent shelf positions or if they receive more shelf space (Breugelmans et al., 2007). According to Breugelmans et al. (2007), the impact of shelf display is especially important when consumers are not highly involved with the purchase decision. Raijas and Tuunainen (2001) define grocery shopping as "buying behavior with low involvement decisions which are made frequently, due to low costs, with low information search and with low risk". Therefore, the layout and arrangement of the shelf can strongly influence the behavior of (online) grocery shoppers. Traditional supermarkets use shelves to present the majority of their assortment. As stated before, many research is done to investigate the influence of the layout of the shelf on shopping behavior. In contrast to the traditional supermarkets, online supermarkets use different layouts to present the available products. The page of an online store where products are being presented can be compared with the shelf in a traditional supermarket because it represents the point where purchase decisions are made. Yet, no research is done to investigate how products should be presented in an online supermarket. It will be interesting to see how online grocery shoppers evaluate and react on different online shelf layouts that are used in an online supermarkets. Therefore, this research will be dedicated to investigate how products should be presented in an online supermarket.

1.2.1 Research objective

As stated before, a large part of the grocery purchases are unplanned and made in-store. Therefore, it is important for retailers to influence the behavior of the shopper in the store. The way in which products are being presented (shelf layout) is an important factor to influence shopper behavior. Online grocery stores use different shelf layouts to present their products. The objective of this research is to investigate the effects of different online shelf layouts on the evaluation of the shelf and shopper behavior.

1.2.2 Research question

The research question is formulated as follows:

What is the impact of different online shelf layouts on shelf evaluation and shopper behavior?

1.2.3 Sub questions

To answer the main research question, the following sub questions are formulated:  What are important effects of the traditional supermarket shelf?

(11)

11

 What are important design features of the online shelf?

 What are important effects of the design of the online shelf?

1.3 Relevancy of the study

The layout of the supermarket shelf can have a strong impact on consumer behavior. Previous research about the effects of the supermarket shelf is focused on the traditional supermarket. Although nowadays only a small part of the Dutch population shops for their grocery at online supermarkets, it is expected that shopping for groceries online become more and more popular. For example, in 2011, Ahold (owner of the online Dutch online supermarket Albert.nl) announced that one of their strategic objectives is to triple online sales over the next five years (Ahold, 2011). The expected growth of the online grocery shopping and the lack of research about the online supermarket shelf makes this research very relevant for both (online) supermarket retailers and manufacturers.

1.4 Structure of the thesis

(12)

12

2. LITERATURE REVIEW

In this chapter, literature about the supermarket shelf will be discussed. Very little research is done to investigate the effects of the online supermarket shelf on consumer behavior. Therefore, existing literature about the traditional supermarket shelf will be discussed first. Thereafter, existing literature about the online shelf will be discussed.

2.1 The traditional supermarket shelf

The vast majority of products in the supermarket are presented on shelves on both sides of the aisles. According to Fernandez and Gomez (2005) the shelf can be described as "the length of the product display in the outlet". The length of the shelf can be measured in facings, products presented in front or in meters. A facing can be defined as a package that is presented full face on the shelf (Fernandez and Gomez, 2005). Previous research have showed that the supermarket shelf is a key influencer of in-store shopping behavior (e.g., Drèze et al., 1994; Van Nierop et al., 2008). The most important shelf effects that can have a strong impact on consumers' choice decisions in the supermarket are the number of facings (shelf space), the vertical shelf position, and the horizontal shelf position (See figure 2).

(13)

13

When a product receives more shelf space (number of facings) it requires more shelf space. However, retailers have a fixed amount of shelf space at their disposal and shelf space is one of the most valuable and scarcest resources of a (supermarket) retailer (Fernandez and Gomez, 2005; Amrouche and Zaccour, 2007; Van Nierop et al., 2008). Therefore the effects of shelf space is an important issue for retailers. Besides the amount of space (number of facings) that products receive, the location of the products on the shelf is important for supermarket retailers. When it comes to the position of a product on the shelf, a distinction can be made between horizontal and vertical shelf position. The horizontal position can best be described as the horizontal distance from the beginning of the shelf to the end of the shelf (from the dominant walking direction). The vertical shelf position can best be described as the height (or shelf) where a product is placed. When products are placed on more prominent positions they have a higher probability that they will be chosen (Drèze et al., 1994; Breugelmans et al., 2007).

Shelf management is important for retailers as well as manufacturers. While (supermarket) retailers want to maximize the sales and profits of the product category as a whole, manufacturers want to maximize the sales and profits of their own products and brands (Drèze et al., 1994). To receive more space and better shelf positions for their products, manufacturers are willing to pay significant premiums (Drèze et al., 1994). On the other hand, retailers charge slotting allowances is some product categories. Bloom et al. (2000), define slotting allowances as "a family of marketing practices that involves payments by manufacturers to persuade downstream channel to stock, display, and support new products". This makes the process of determining how much to allocate to products even more complicated. Therefore, the process of determining the amount of space and the position of products on the shelf is of great importance for supermarket retailers.

Research about shelf space and shelf layout have been of interest to researchers for many years. Already in the 1960s and 1970s experiments were done to measure the effects of shelf space, for example Cox (1964), Kotzan and Evanson (1969) and Curhan (1972). More recent research in the field of shelf management measured the effects of the layout of the shelf, for example Drèze et al. (1994), Christenfield (1995), and Van Nierop et al. (2008). The relevant literature about shelf space will be discussed in subparagraph 2.1.1. In subparagraph 2.1.2 relevant literature about the layout of the traditional supermarket shelf will be discussed.

2.1.1 Shelf space

(14)

14

out-of-stock (Drèze et al., 1994). It is obvious that consumers are not able to buy products when they are not on the shelf. Therefore, products that sell better usually receive more facings. Preventing products of being OOS is important because of the negative reactions of consumers to the occurrence of OOS. Previous research about consumers reactions to OOS has shown that consumers can delay the purchase or even go to another store when the preferred product is not available (Borin and Farris, 1995). As a result, the event of OOS can have a negative impact on the sales and profits of a retailer. Secondly, changes in the amount of shelf space allocated to products can increase the attention of consumers (Drèze et al., 1994). As stated before, supermarkets are places with a high rate of impulse buying. So, when a product receive more facings, it is likely that it would be visually perceived better, which can result in higher sales.

(15)

15

extra faces can have a strong impact on sales, especially from low-market-share brands. Although not all studies about the relationship between shelf space and sales show the same results, it is clear that the number of facings can positively affect the choices of consumers, and therefore it is an important effect of the supermarket shelf.

2.1.2 Shelf positioning

Besides the amount of shelf space allocated to products, the position of a product on the shelf can also influence the choices of consumers. According to Drèze et al. (1994) the position of a product on the shelf is far more important than the number of facings. When it comes to the position of the shelf a distinction can be made between the horizontal en vertical shelf position. In this part, the effects of horizontal and vertical shelf positioning found by researchers will be discussed.

Horizontal shelf positioning

(16)

16

the point where the shopper enters the aisle, so again it depends on the walking route of the shopper. They also found that products placed on the beginning of the shelf are more sensitive to promotions than products that are placed in the middle of the shelf. Furthermore, they found that shoppers are less price sensitive when products are placed in the middle horizontal shelf position.

Vertical shelf positioning

Since the beginning of supermarket retailing manufacturers and marketers have argued that when it comes to vertical positioning that an eye-level shelf position provide maximum sales (Hubbard, 1969). According to Breugelmans et al. (2007) from all the shelf effects the vertical shelf position seems to have the strongest effect. They state that the probability of being selected is significantly higher for products that are placed on hand- or eye-level positions. However, Frank and Massy (1970) found that the vertical position of a product on the shelf had only a modest effect on sales or even had no effect on sales at all. On the other hand, Drèze et al. (1994) found that the vertical position of a product on the shelf can have a positive effect on sales. They found that a vertical position slightly below eye-level was the most favorable. Furthermore, they found that, when it comes to facings, adding extra facings at eye-level is far more effective than adding extra facing on the bottom shelf. So placing products on eye-level has a positive effect on sales. Research performed by Van Nierop et al. (2008) showed a similar result. They also found that the vertical shelf position can have a positive effect on sales. Their research showed that, until eye-level, a higher vertical shelf position generates more sales. When the vertical shelf position becomes higher than eye-level, the effect on sales declines. Furthermore, they found that shoppers become more sensitive to changes in price when a product is placed higher on the shelf. They also found that the vertical position on the shelf can influence the effectiveness of promotions. Promotions are more effective for products that are placed on the bottom shelf than for products that are placed on a higher vertical shelf position.

2.2 The online supermarket shelf

(17)

17

purchase at the online supermarket. The product listing page is one of the most relevant pages of a website in the product searching process (Schmutz et al., 2010). Previous research has shown that the design of a product listing page can have a great influence on online shopping behavior (e.g. Lohse & Spiller, 1998; Hong et al., 2004a; Shun & Yunjie, 2008: Schmutz et al., 2010). Lohse and Spiller (1998) found that improvements in the design of a product listing page can lead to higher sales and store traffic. Schmutz et al. (2010) also found that the design of the product listing pages can have an important effect on decision-making and sales. Furthermore, they state that improving the usability of an E-commerce website is key to improve customer satisfaction, increase sales and enlarge customer retention. Therefore, the design of product listing pages is an important aspect for online supermarket retailers.

There are a number of different design features of product listing pages. The two relevant design features for this research are the presentation format and the sorting of products on a product listing page. The presentation format is also called the information format and is defined as the presentation and organization of information about the available products and their description (Cooper-Martin, 1993; Hong et al., 2004a). For this research, the term presentation format is used. Sorting of products refers to the order in which products are sorted (i.e. descending prices, ascending prices, alphabetical by name). The presentation format will be discussed in subparagraph 2.2.1. In subparagraph 2.2.2, the sorting of products will be discussed.

2.2.1 Presentation format

The two most popular presentation formats are the list presentation format and the matrix presentation format. A list presentation format displays only one product on each row (see figure 3). A matrix presentation format displays more than one product on each row (see figure 4). It has to be noted that there are a variety of other presentation formats available (e.g. presentations of products in context and iconic presentations in irregular positions), but list and matrix presentation formats are the most relevant for most commercial websites and other presentation formats will not be discussed in this research.

(18)

18

the attitude of users towards using the website is more positive when a products are presented in a list presentation format. In another research Hong et al. (2004b) investigated how the fit between shopping tasks (browsing vs. searching) and the presentation format (list vs. matrix) influenced online shopping behavior. They found that that a list presentation format is best when consumers are browsing for particular products. When consumers are searching for products, the matrix presentation format is better. More recent research by Schmutz et al. (2010) shows that the list presentation format is better than matrix presentation format when the shopping task involves comparing different products. So, the results of previous research suggest that in most cases a list presentation format is preferred.

Figure 3 - Screenshot of an online supermarket in list presentation format

(19)

19

2.2.2 Sorting

(20)

20

3. HYPOTHESES & CONCEPTUAL MODEL

This chapter will cover the hypotheses and the conceptual model for this research. In order to develop the hypotheses and conceptual model, the independent and dependent variables will be described first.

3.1 Independent variables

The independent variables that are used for this research relate to the design of the online supermarket shelf. The literature review has showed that the presentation format and the way that products are sorted can significantly influence consumer shopping behavior.

3.1.1 Presentation format

The two most commonly used presentation formats for product listing pages are the list presentation format and the matrix presentation format. The list presentation format displays one product on each row and the matrix presentation format displays more products on each row. For this research the online shelf is presented in list or in matrix presentation format.

3.1.2 Sorting

Online retailers use different sorting methods to present their products. Little research is done to investigate the effects of sorting of products on online (grocery) stores. For this research the four most frequently used ways of sorting will be investigated. These are sorting by ascending prices, sorting by descending prices, sorting by market share (High to Low), and sorting by alphabet (A to Z).

3.2 Dependent variables

The literature review has shown that the supermarket shelf can have a great influence on shopping behavior. No research is done to investigate the impact of different presentations of the online supermarket shelf on the evaluation of the supermarket shelf and shopping behavior. The dependent variables that will be used for this research will be discussed in the next subparagraphs.

3.2.1 Shelf evaluation variables

(21)

21

used to measure how consumers evaluate the supermarket shelf. The shelf benefits that are used for evaluation of the online shelf are perceived efficiency and perceived choice. Furthermore, the online shelf can be evaluated on the basis of attractiveness of the online shelf. The variables that are used for the evaluation of the shelf will be described briefly.

Perceived Efficiency

Perceived efficiency measures the extent to which the shelf contributes to speed of the shopping process. The speed of shopping is an important factor when it comes to online shopping because saving time is one of the key advantages of online (grocery) shopping (Breugelmans & Campo, 2011; Verhoef & Langerak, 2001). Therefore, it will be interesting to investigate the effect of online shelf layout on perceived efficiency.

Perceived Choice

Perceived choice measures the extent to which the shelf contributes to the desired depth and breadth of the assortment. Hong et al. (2004a) found that the design of the online shelf can have a significant influence on the recall of product images and brands. Therefore it will be interesting to investigate whether the design of the online shelf influences the perceived choice of online grocery shoppers.

Overall shelf attractiveness

This variable measures the extent of shelf attractiveness of shoppers. As stated before, the design of a product listing page can lead to higher sales and store traffic (Spiller, 1998). Furthermore, the design can have a significant impact on decision making (Schmutz et al., 2010). Therefore, it can be relevant to measure the level of shelf attractiveness of an online supermarket shelf. It can be expected that higher scores on overall shelf attractiveness can lead to higher sales an store traffic.

3.2.2 Reference price

(22)

22

consumers are willing to pay (Bennet et al., 2003; Kwanho et al., 2012). No research is done to investigate the effects of the design of the shelf on the reference price. Therefore, it will be interesting to investigate if the way that the prices are sorted in the online shelf can influence the reference price.

3.2.3 Shopper behavior variables

The last variables that are used for this research relate to the shopper behavior of online grocery shoppers. These variables are time to buy and total spent.

Time to buy

Time to buy measures the time it takes for a shopper to buy a product. As stated earlier, time saving is one of the key advantages of online shopping and therefore it will be relevant to investigate whether different shelf presentations influence the time to buy (grocery) products.

Total spent

Total spent measures the price of the product that shoppers buy. This is also relevant for online retailers because it will be interesting to investigate whether different online shelf presentations influence the choice of products of online grocery shoppers.

3.3 Hypotheses

In this paragraph, the hypotheses for this thesis will be developed. Where possible, the hypotheses are based on existing literature about the (online) supermarket shelf. The hypotheses all relate to the design of the online shelf and how shoppers evaluate the shelf. In subparagraph 3.3.1 the hypotheses regarding the presentation format will be presented. In subparagraph 3.3.2, hypotheses regarding sorting will be presented.

3.3.1 Presentation format effects

(23)

23

better when consumers are searching for products (Hong et al., 2004b). Furthermore, the traditional supermarket shelf is more consistent with the matrix presentation format (multiple products on each row) and consumers are used to this type of presentation when they shop for groceries. For this research, it is expected that there will be differences in terms of perceived efficiency, perceived choice, overall shelf attractiveness, and time to buy between list and matrix presentation format. First, the matrix presentation format is more consistent with the traditional supermarket shelf because there are more products presented on each row. Therefore, it is expected that when the online supermarket shelf is presented in list presentation format the perceived efficiency is lower than when the online shelf is presented in matrix information format (H1a). Second, the different presentations require a different amount of space. A product list in matrix presentation format is more compact and requires less space than a product list in list presentation format. Therefore, it is likely that shoppers perceive the breadth and/or the depth of the assortment differently. A list presentation format requires more space and as a result consumers can get the impression that the assortment is larger. Therefore, it is expected that the perceived choice is higher when the online shelf is presented in list presentation format than in matrix presentation format (H1b). Third, although the literature review showed that the attitude of shoppers towards using the website is more positive when products are being presented in list presentation format, the matrix presentation format is more consistent with the traditional supermarket shelf. Therefore, it is expected that the overall shelf attractiveness is lower when products are being presented in list presentation format than in matrix presentation format (H1c). Finally, because the matrix presentation format is more consistent with the traditional supermarket shelf and because this presentation format requires less space, it is expected that the time to buy a product is longer when a products are presented in list presentation format than when products are presented in matrix presentation format (H1d).

H1a. Perceived efficiency will be lower in the list presentation format than in the matrix

presentation format.

H1b. Perceived choice will be higher in the list presentation format than in the matrix

H1c. Overall shelf attractiveness will be lower in the list presentation format than in the

matrix presentation format.

H1d. Time to buy will be longer in the list presentation format than in the matrix

(24)

24

3.3.2 Sorting effects

Another important design feature of the online shelf is the way how products are sorted. Although there is no extended literature available about the impact of sorting on shelf evaluation the literature review have showed that is can have an impact on online shopping behavior. Especially sorting by price can have a strong impact on online shopping behavior. First, the hypotheses for sorting by price will be developed (ascending versus descending and after that the hypotheses on the alphabetical (A-Z) versus market share (High-Low) will be developed.

Descending prices versus ascending prices

The literature review has showed that sorting by price can have a significant impact on online consumer behavior. No research is done to investigate the impact of different price sorting on the evaluation of the online shelf. For this research, it is expected that there will be differences in terms of reference price, total spent, and overall shelf attractiveness between descending and ascending price sorting. The literature review has showed that the willingness to pay of shoppers is higher when products are presented in descending price order. Furthermore, consumers tend to buy higher priced products when the products are presented in descending price order. This is also in line with the primacy effect that is discussed earlier in the literature review. Therefore, it is expected that the reference price and the total spent is higher when products are presented in descending price order than in ascending price order (H2b & H2c). When products are presented in descending price order, shoppers see the most expensive alternatives first and when products are presented in ascending price order, shoppers see the cheapest alternatives first. As a result different price sorting can have an impact on how consumers evaluate the overall attractiveness of the online supermarket shelf. Therefore, it is expected that the overall shelf attractiveness is higher when products are presented in ascending price (H2a).

H2a. Overall shelf attractiveness will be lower when products are sorted in descending

price order than in ascending price order.

H2b. Reference price will be higher when products are sorted in descending price order

than in ascending price order.

H2c. Total spent will be higher when products are sorted in descending price order than in

(25)

25

Market share (high-low) versus alphabetical (A-Z) versus

No research is done to measure the impact of sorting alphabetically and sorting by market share. For this research, it is expected that there will be difference in terms of perceived efficiency, perceived choice, reference price, time to buy, total spent, and overall shelf attractiveness. Firstly, when products are sorted by market share, the most frequently bought products are presented first. As a result, most shoppers will find the most popular products faster than when they are sorted alphabetically. On the other hand, when products are presented by market share, changes are high that products of the same brands are mixed up in the shelf. As a result, it takes more time and effort when shoppers are searching for a particular product. Therefore, it is expected that the perceived efficiency is lower and the time to buy is longer when products are sorted by market share than when they are sorted by alphabet (H3a & H3e). Secondly, when products are sorted by alphabet, brands are sorted together and shoppers can easily see how many different brand there are. When products are sorted by market share, changes are high that the brands are mixed up in the shelf. As a result, it is likely that this has an impact on the perceived breadth and depth of the assortment. Therefore, it is expected that the perceived choice is higher when products are sorted by market share than when they are sorted by alphabet (H3b). Thirdly, when products are sorted by market share, the most popular products are presented first. Not infrequently, the most popular products in a category are products of a premium brand. Therefore, it is expected that the reference price and the total spent are higher when products are sorted by market share than when they are sorted by alphabet (H3d & H3f). Finally, because of the previous arguments, it is expected that the overall shelf attractiveness is higher when products are sorted by market share than when they are sorted by alphabet (H3c).

H3a. Perceived efficiency is lower when products are sorted by market share than when they are sorted by alphabet.

H3b. Perceived choice is higher when products are sorted by market share than when they

are sorted by alphabet.

H3c. Overall shelf attractiveness is higher when products are sorted by market share than

when they are sorted by alphabet.

H3d. Reference price is higher when products are sorted by market share than when they

(26)

26

H3e. Time to buy is longer when products are sorted by market share than when they are

sorted by alphabet.

H3f. Total spent is higher when products are sorted by market share than when they are sorted by alphabet.

3.4 Conceptual model

The hypotheses lead to the following conceptual model:

Figure 5 - Conceptual model

Online shelf Layout

Presentation format List vs. Matrix Sorting Descending prices vs. Ascending prices

Market share (High-Low) vs. Alphabetical (A-Z) Shelf evaluation Reference Price Shopper behavior Perceived Efficiency Perceived Choice Total spent Time to buy Overall Shelf Attractiveness

(27)

27

4. RESEARCH DESIGN

4.1 Type of research

According to Malhotra (2007), three types of research designs can be distinguished; exploratory research, descriptive research, and causal research. Exploratory research can be used to gain understanding of, or insight into a problem that is not been clearly defined. The main objective of descriptive research is to describe market characteristics or functions. Causal research can be used to determine cause and effect relationships. The first part of this research (chapter 2) can be classified as exploratory research and serves as input for the next part of this research. The main objective of the second part of this research is to determine the effect of different online supermarket shelf layouts on the evaluation of the online supermarket shelf and shopper behavior. Therefore, the second part of this research can be classified as a causal research.

4.2 Research method

According to Malhotra (2007), the main method for conducting causal research is experimentation. Experimentation is used to measure the relationships between the variables in a relatively controlled environment. For this research, an experiment website is developed where the layout of the online shelf (independent variables) will be manipulated to measure the effect on the evaluation of the online supermarket shelf and shopper behavior (dependent variables). To obtain the required data for this research, an online survey is developed where the experiment website used in combination with an online questionnaire.

4.3 Experiment website

(28)

28

shelf layouts. The reason for this is to limit the number of respondents that are needed for this research. The presentation formats that are used are the list presentation format and the matrix presentation format. In the list presentation format, respondent will see one product on each row (see figure 6). In the matrix presentation format, respondents will see three products on each row (see figure 7). Furthermore, four different types of sorting formats will be used. The sorting formats that are used are sorting by descending prices (high to low), sorting by ascending prices (low to high), sorting by alphabet from A to Z, and sorting by popularity (market share) from high to low. The two presentation formats and the four sorting formats result in 8 different online shelf layouts. Since there are four different product categories, a total of 32 shelf layouts are used for this research. Finally, it has to be noted that the information content (e.g. number of products, image size, prices, and product information) was held constant across all the 32 different online shelf layouts.

(29)

29

Figure 7 - Experiment website in matrix presentation format

4.3 Data collection

To collect the data for this research, an online survey is developed. The online survey consist of a experiment website in combination with an online questionnaire. The reason for using an online survey is because the advantages of this method is by far the fastest in obtaining data from a large number of respondents at the lowest costs (Malhotra, 2007; Ilieva et al., 2002). Moreover, an online survey is very appropriate for this research because the problem at hand is about the layout of the online supermarket shelf. The online survey is created in Simstore360, a system of the research company NORM Research & Consulting.

4.3.1 Target population and sample

The target population of this research are Dutch shoppers between 18 and 65 years old who are fully or partly responsible for grocery shopping. It as to be noted that although this research is about the online supermarket shelf, the target population is set on all grocery shopper (not only online grocery shoppers). The reason for this is because only 3% of the grocery purchases are done by using the online channel (Blauw Research, 2011a). As a result, it would be very difficult and time consuming to find enough respondents. In total, 32 online shelves should be evaluated by the respondents (8 online shelf layouts of 4 product categories). The minimum number of evaluations is set on 25. According to Baarda et al. (2003), the minimum for most statistical analysis is 25 respondents. As a result, a minimum of 800 respondents is needed for this research. Because of resource constraints, it was decided that each respondent has to evaluate 4 online shelf layouts. Therefore, the sample size for this research is set at a minimum of 200 respondents. To obtain the required respondents a representative online panel of the data collection company GMI is used.

(30)

30

4.3.2 Design of the questionnaire

The online questionnaire is designed to investigate the effect of the 8 different online shelf layouts on shelf evaluation and shopper behavior. In addition, the questionnaire contains question about demographics, supermarket shopping, and online shopping. These additional questions can be used to describe the characteristics of the respondents. The dependent variables perceived efficiency, perceived choice, and overall shelf attractiveness are constructs based on multiple questions. A complete overview of the variables and the corresponding questions can be found in appendix 3. The dependent variables perceived efficiency and perceived choice are measured with a 5-point Likert scale which require respondents to indicate the level of agreement or disagreement with the statements. The dependent variable overall shelf attractiveness is measured with a 10-point Likert scale which requires respondents to choose between two opposites.

(31)

31

Coke Beer Spaghetti Toilet paper

Group 1 List Alphabetical (A-Z) Matrix Price descending (H-L) List Price ascending (Low-High) Matrix Market share (High-Low) Group 2 List Price descending (High-Low) Matrix Price ascending (Low-High) List Market share (High-Low) Matrix Alphabetical (A-Z) Group 3 List Price ascending (Low-High) Matrix Market share (High-Low) List Alphabetical (A-Z) Matrix Price descending (High-Low) Group 4 List Market share (High-Low) Matrix Alphabetical (A-Z) List Price descending (High-Low) Matrix Price ascending (Low-High) Group 5 Matrix Alphabetical (A-Z) List Price descending (High-Low) Matrix Price ascending (Low-High) List Market share (High-Low) Group 6 Matrix Price descending (High-Low) List Price ascending (Low-High) Matrix Market share (High-Low) List Alphabetical (A-Z) Group 7 Matrix Price ascending (Low-High) List Market share (High-Low) Matrix Alphabetical (A-Z) List Price descending (High-Low) Group 8 Matrix Market share (High-Low) List Alphabetical (A-Z) Matrix Price descending (High-Low) List Price ascending (Low-High)

Table 1 - Groups for the questionnaire

4.4 Plan of analysis

The results of the questionnaire will be analyzed by using different statistical test. To analyze the results, the statistical program SPSS 17.0 will be used. A database based on respondents (n=339) will be used for the descriptive statistics and the control check. A database based on evaluations (n=1356) will be used for further analysis.

4.4.1 Descriptive statistics

The first step of the data analysis is to describe the results of the questions regarding demographics, grocery shopping, and online shopping. For the descriptive statistics, the database based on respondents (n=339) will be used. The results of this analysis are used to describe the characteristics of the respondents who participated in this research. The results of the descriptive analyses will be described briefly.

4.4.2 Control check groups

(32)

32

same composition. The different variables are not measured on the same scale and therefore different statistical test will be used to analyze the differences between the 8 groups. When the variable is measured on a nominal scale, a Chi-square test will be used. The Chi-square test can be used to check if there is a significant difference between two or more groups on a variable that is measured on nominal scale (Baarda et al., 2003). When the variable is measured on an ordinal scale, a Kruskal-Wallis test will be used. The Kruskal-Wallis test can be used to check if there is a significant difference between more than two groups on a variable that is measured on an ordinal scale (Baarda et al., 2003). When the variable is measured on interval/ratio scale, a one-way ANOVA test will be used. The one-way ANOVA can be used to measure if there is a significant difference between more than two group on a variable that is measured on inter/ratio scale (Baarda et al., 2003). The level of significance is set on 0.05. So there are significant differences between groups if p= 0.05 or lower.

4.4.3 Internal consistency reliability

The next step of the data analysis is to check the reliability of variables that are measured by multiple questions. For this analysis, the database based on evaluations (n=1356) will be used. The variables that are measured by multiple questions are perceived efficiency, perceived choice, and overall shelf attractiveness. These variables and their corresponding questions can be found in appendix 3. Before the analysis the reversed question have to be recoded. To test the reliability of the multiple item variables the internal consistency reliability is checked to make sure the new variables can be used for further analysis. The internal consistency reliability is checked by using Chronbach's alpha. The Chronbach's alpha should have a value between 0.6 and 1 to indicate a satisfactory internal consistency reliability (Malhotra, 2007). When Chronbach's alpha is lower than 0.6, the alpha if item deleted is checked to find out if Chronbach's alpha is sufficient when one item is deleted. When Chronbach's alpha is 0.6 or higher, the new variable can be used for further analysis.

4.4.4 Hypothesis testing

(33)

33

(34)

34

5. RESULTS

In total, 476 respondents started the online survey. 65 respondents were screened out because they did not meet the target group (age or not responsible for grocery shopping). The results of 62 respondents are deleted because they did not complete the survey (59 respondents) or because of extreme values (13 respondents). As a result, the results of 339 respondents are used to describe the characteristics of the sample. For the descriptive statistics and the control check of the 8 different groups the database based on respondents (n=339) is used. In the online survey, each respondent is asked to evaluate 4 different online supermarket shelves, therefore, the database based on evaluations (n=1356) is used for further analysis. In this chapter the results of this research will be presented.

5.1 Descriptive statistics

In this paragraph, the descriptive statistics of the research will be described. First, the demographics of the sample will be described. Second, grocery shopping related characteristics will be described. Finally, the online shopping characteristics of the sample will be described. A complete overview of the descriptive statistics can be found in appendix 5.

5.1.1 Demographics

(35)

35 CBS 2011 Current research (n=339) Gender Male 49% 50% Female 51% 50% Age 18-34 32% 50% 35-50 37% 27% 51-65 31% 23% Household size 1 person 36% 32% 2 or more persons 64% 68%

Table 2 - Representativeness of the sample (source: CBS Statline)

5.1.2 Grocery shopping

Besides the demographic variables, some variables about grocery shopping are investigated. 90.3% of the respondents (sometimes) shop for their groceries at the Dutch supermarket retailer Albert Heijn. So, almost all the respondents are familiar with the Albert Heijn products that are used in this research. Furthermore, the majority of the respondents visit the Albert Heijn at least once a week (45.1%). 80.8% of the respondents sometimes buys a bottle of coke, 56.6% sometimes buys a crate of beer, 84.1% sometimes buys spaghetti, and 94.1% sometimes buys toilet paper. When looking at average weekly spending on groceries, the largest part of the respondents (44.2%) spends on average between 50 and 100 Euros a week on grocery shopping.

5.1.3 Online shopping

Since only a small percentage of grocery shopping is done in an online supermarket in the Netherlands, several variables to describe online shopping are investigated. 81.1% of the respondents sometimes visit online shops and 78.5% of the respondents sometimes buy products online. When it comes to online grocery shopping, only 38.4% of the respondents sometimes visit an online supermarket and only 17.2% of the respondents sometimes buy products at an online supermarket. Only 9.1% of the respondents sometimes shop for their groceries at Albert.nl (the online supermarket of Albert Heijn). These results show that the major part of grocery shopping still is done in the ‘traditional’ supermarket.

5.2 Control check groups

(36)

36

corresponding tests, and the results can be found in appendix 6. The 8 different groups are compared on a number demographic, grocery shopping, and online shopping variables. The demographic variables that are used are gender, age, household composition, number of kids living at home, education, social status, and income. The grocery shopping related variables that are used are shopping frequency at Albert Heijn and average weekly spending on groceries. The online shopping related variables are online shopping experience, online shop visiting frequency, online shop buying frequency, online supermarket visiting frequency, online supermarket buying frequency, and shopping at Albert.nl. The results of the statistical test are shown in table 3. The results show that there are no significant differences on the variables between the 8 different groups.

Variables Sig.

Demographic

Gender 0.408

Age 0.094

Household composition 0.180

Number of kids living at home 0.563

Education 0.560

Social status 0.153

Income 0.532

Grocery shopping related

Shopping frequency at Albert Heijn 0.981 Average weekly spending on groceries 0.178

Online shopping related

Online shopping experience 0.310

Online shop visiting frequency 0.212

Online shop buying frequency 0.283

Online supermarket visiting frequency 0.540 Online supermarket buying frequency 0.665

Shopping at Albert.nl 0.913

Table 3 - Results comparing groups

5.3 Internal consistency reliability

(37)

37

First, the Chronbach’s alpha for perceived efficiency is tested. Six new variables are computed to measure perceived efficiency. These are perceived efficiency for list format (Perceived efficiency List), perceived efficiency for matrix format (Perceived efficiency Matrix), perceived efficiency for sorting by descending prices (Perceived efficiency HL), perceived efficiency for sorting by ascending prices (Perceived efficiency LH), perceived efficiency for sorting by alphabet (Perceived efficiency AZ), and perceived efficiency for sorting by market share (Perceived efficiency MS). The new variables and the corresponding Chronbach’s alpha are shown in table 4. The Chronbach’s alpha of all the new perceived efficiency variables is above the minimum of 0.6 and therefore they can be used for further analysis.

New variable Chronbach’s alpha

Presentation Format Perceived efficiency List 0.852

Perceived efficiency Matrix 0.842

Sorting

Perceived efficiency HL 0.876 Perceived efficiency LH 0.824 Perceived efficiency AZ 0.824 Perceived efficiency MS 0.859 Table 4 - Chronbach’s alpha perceived efficiency

The second variable that is measured by multiple questions is perceived choice (Perceived choice). Again, the Chronbach’s alpha of the six new variables are tested. The results are shown in table 5. The Chronbach’s alpha of all the new variables is above 0.6 and therefore they can be used for further analysis.

Presentation Format Perceived choice List 0.691

Perceived choice Matrix 0.696

Sorting

Perceived choice HL 0.668

Perceived choice LH 0.684

Perceived choice AZ 0.695

Perceived choice MS 0.727

Table 5 - Chronbach’s alpha perceived choice

(38)

38

Presentation Format Attractiveness List 0.929

Attractiveness Matrix 0.928 Sorting Attractiveness HL 0.926 Attractiveness LH 0.927 Attractiveness AZ 0.935 Attractiveness MS 0.925

Table 6 - Chronbach’s alpha overall shelf attractiveness

5.3 Hypothesis testing

In this paragraph, the results of testing the hypotheses will be presented. First, the results of the hypotheses regarding the presentation format will be presented. Second, the results of the hypotheses regarding sorting will be presented. The hypotheses are tested by using several multiple regression analyses. A complete overview of the results can be found in appendix 8.

5.3.1 Presentation format effects

Several multiple regression analyses are conducted to test the hypotheses regarding the presentation format. The dummy variables that are used as predictors for the regression are list presentation format, matrix presentation format, sorting by descending prices, sorting by ascending prices, sorting by market share (High-Low), and sorting by alphabet (A-Z). The reference variables for the regression analyses are matrix presentation format and sorting by descending price.

H1a. Perceived efficiency will be lower in the list presentation format than in the matrix presentation format.

A multiple regression analysis was used to test if the design of the online supermarket shelf significantly predicted the perceived efficiency. The results of the regression indicated that 0.2% of the variance in perceived efficiency is explained by the design of the online supermarket shelf. The design of the online supermarket shelf did not significantly predicted perceived efficiency (R²= 0.005, F(4, 1351)= 1.687, p= 0.150). Therefore, hypothesis H1a is rejected.

H1b. Perceived choice will be higher in the list presentation format than in the matrix

(39)

39

p= 0.021). The significant variables are list presentation format (β= 0.73, p= 0.007) and sorting by ascending price (β= 0.067, p= 0.043). The results suggest that the perceived choice is higher when products are presented in list presentation format than when they are presented in matrix presentation format. Therefore, hypothesis H1b is accepted. Furthermore, the results suggest that the perceived choice is higher when the product are sorted by ascending prices than when they are sorted by descending prices.

H1c. Overall shelf attractiveness will be lower in the list presentation format than in the

matrix presentation format.

A multiple regression analysis was used to test if the design of the online supermarket shelf significantly predicted the overall shelf attractiveness. The results of the regression indicated that 0% of the variance in overall shelf attractiveness is explained by the design of the online supermarket shelf. The design of the online supermarket shelf did not significantly predicted overall shelf attractiveness (R²= 0.003, F(4, 1351)= 1.075, p= 0.367). Therefore, hypothesis H1c is rejected.

H1d. Time to buy will be longer in the list presentation format than in the matrix

A multiple regression analysis was used to test if the design of the online supermarket shelf significantly predicted the time to buy a product. The results of the regression indicated that 0.3% of the variance in time to buy is explained by the design of the online supermarket shelf. The design of the online supermarket shelf significantly predicted time to buy (R²= 0.006, F(4, 1351)= 2.148, p= 0.073). The significant variable is list presentation format (β= 0.047, p= 0.085). The results suggest that the time to buy a product is shorter when products are presented in matrix format than when they are presented in list presentation format. Therefore, H1d is accepted.

5.3.2 Sorting effects

First, the test results of the hypotheses regarding sorting by descending prices versus sorting by ascending prices will be presented. The dummy variables that are used as predictors for the regression are list presentation format, matrix presentation format, sorting by descending prices, sorting by ascending prices, sorting by market share (High-Low), and sorting by alphabet (A-Z). The reference variables for the regression analyses are matrix presentation format and sorting by descending price.

H2a. Overall shelf attractiveness will be lower when products are sorted in descending

(40)

40

A multiple regression analysis was used to test if the design of the online supermarket shelf significantly predicted the overall shelf attractiveness. The results of the regression indicated that 0% of the variance in overall shelf attractiveness is explained by the design of the online supermarket shelf. The design of the online supermarket shelf did not significantly predicted overall shelf attractiveness (R²= 0.003, F(4, 1351)= 1.075, p= 0.367). Therefore, hypothesis H2a is rejected.

H2b. Reference price will be higher when products are sorted in descending price order

than in ascending price order.

A multiple regression analysis was used to test if the design of the online supermarket shelf significantly predicted the reference price. The results of the regression indicated that -0.2% of the variance in reference price is explained by the design of the online supermarket shelf. The design of the online supermarket shelf did not significantly predicted reference price (R²= 0.001, F(4, 1351)= 0.203, p= 0.937). Therefore, hypothesis H2b is rejected.

H2c. Total spent will be higher when products are sorted in descending price order than in

ascending price order.

A multiple regression analysis was used to test if the design of the online supermarket shelf significantly predicted the total spent. The results of the regression indicated that -0.2% of the variance in total spent is explained by the design of the online supermarket shelf. The design of the online supermarket shelf did not significantly predicted total spent (R²= 0.001, F(4, 1351)= 0.185, p= 0.946). Therefore, hypothesis H2c is rejected.

Next, the test results of the hypotheses regarding sorting by markets share (high-low) versus sorting by alphabet (a-z) will be presented. The dummy variables that are used as predictors for the regression are list presentation format, matrix presentation format, sorting by descending prices, sorting by ascending prices, sorting by market share (High-Low), and sorting by alphabet (A-Z). The reference variables for the regression analyses are matrix presentation format and sorting by market share.

H3a. Perceived efficiency is higher when products are sorted by market share than when

they are sorted by alphabet.

(41)

41

the online supermarket shelf did not significantly predicted perceived efficiency (R²= 0.005 F(4, 1351)= 1.687, p= 0.150). Therefore, hypothesis H3a is rejected.

H3b. Perceived choice is higher when products are sorted by market share than when they

A multiple regression analysis was used to test if the design of the online supermarket shelf significantly predicted the perceived choice. The results of the regression indicated that 0.6% of the variance in perceived choice is explained by the design of the online supermarket shelf. The design of the online supermarket shelf significantly predicted perceived choice (R²= 0.008, F(4, 1351)= 2.894, p= 0.021). The significant variable was list presentation format (β= 0.73, p= 0.007). The results suggest that the perceived choice is higher when products are presented in list presentation format than when they are presented in matrix presentation format. No significant results were found regarding sorting by market share versus sorting by alphabet. Therefore, hypothesis H3b is rejected.

H3c. Overall shelf attractiveness is higher when products are sorted by market share than

when they are sorted by alphabet.

A multiple regression analysis was used to test if the design of the online supermarket shelf significantly predicted the overall shelf attractiveness. The results of the regression indicated that 0% of the variance in overall shelf attractiveness is explained by the design of the online supermarket shelf. The design of the online supermarket shelf did not significantly predicted overall shelf attractiveness (R²= 0.003, F(4, 1351)= 1.075, p= 0.367). Therefore, hypothesis H3c is rejected.

H3d. Reference price is higher when products are sorted by market share than when they

A multiple regression analysis was used to test if the design of the online supermarket shelf significantly predicted the reference price. The results of the regression indicated that -0.2% of the variance in reference price is explained by the design of the online supermarket shelf. The design of the online supermarket shelf did not significantly predicted reference price (R²= 0.001, F(4, 1351)= 0.203, p= 0.937). Therefore, hypothesis H3d is rejected.

H3e. Time to buy is longer when products are sorted by market share than when they are

sorted by alphabet.

(42)

42

the online supermarket shelf significantly predicted time to buy (R²= 0.06, F(4, 1351)= 2.148, p= 0.073). The significant variables are list presentation format (β= 0.047, p= 0.085), sorting by ascending price (β= -0.075, p= 0.25), and sorting by alphabet (β=-0.058, p= 0.081). The results suggest that the time to buy a product is shorter when products are presented in matrix format than when they are presented in list presentation format. Furthermore, the results suggest that the time to buy is shorter when products are sorted by ascending price then when they are sorted by market share. Finally, the results suggest that the time to buy a product is shorter when products are sorted by alphabet than when they are sorted by market share. Therefore, hypothesis H3e is accepted.

H3f. Total spent is higher when products are sorted by market share than when they are sorted by alphabet.

The online supermarket shelf

The online supermarket shelf

The impact of different online shelf layouts on

shelf evaluation and shopper behavior

The online supermarket shelf

The impact of different online shelf layouts on

shelf evaluation and shopper behavior

MANAGEMENT SUMMARY

PREFACE

TABLE OF CONTENTS

1. INTRODUCTION

1.1 Background

1.2 Problem identification

1.3 Relevancy of the study

1.4 Structure of the thesis

2. LITERATURE REVIEW

2.1 The traditional supermarket shelf

2.2 The online supermarket shelf

3. HYPOTHESES & CONCEPTUAL MODEL

3.1 Independent variables

3.2 Dependent variables

3.3 Hypotheses

3.4 Conceptual model

4. RESEARCH DESIGN

4.1 Type of research

4.2 Research method

4.3 Experiment website

4.3 Data collection

4.4 Plan of analysis

5. RESULTS

5.1 Descriptive statistics

5.2 Control check groups

5.3 Internal consistency reliability

5.3 Hypothesis testing