Choice of prevailing delivery methods in e-grocery: a stated preference ranking experiment

Christina Milioti (Department of Civil Engineering, University of West Attica, Athens, Greece)
Katerina Pramatari (Department of Management Science and Technology, Athens University of Economics and Business, Athens, Greece)
Eleni Zampou (Department of Management Science and Technology, Athens University of Economics and Business, Athens, Greece)

International Journal of Retail & Distribution Management

ISSN: 0959-0552

Article publication date: 18 November 2020

Issue publication date: 26 January 2021

1231

Abstract

Purpose

The main purpose of this research is to investigate acceptability of different delivery methods in e-grocery (home delivery, pick-up from store, pick-up from locker) and the respective willingness of customers to pay for them using a stated preference ranking experiment.

Design/methodology/approach

Data collected involved two countries (Greece and UK) with different level of e-grocery development and two different distribution conditions (weekly and urgent order). Rank-ordered logit model is used to analyse the ranking experiment and calculate the willingness-to-pay (WTP) measures. Delivery mode, cost and time window are used as independent variables.

Findings

Results indicated that home delivery and picking-up from locker appear to be clearly preferable than picking-up from store. However, home delivery seems to hold a stronger competitive position over the other delivery methods, especially in the weekly order and in the UK market. The pick-up from locker option appears to be a competitive delivery mode for consumers who are cost sensitive and shop under urgent conditions. Willingness to use and pay for picking-up from locker increases significantly in the context of same-day delivery.

Practical implications

The information provided in this study will help retailers to design and implement distribution schemes that can meet consumers' preferences for e-grocery. WTP differences among the consumer groups and the distribution conditions examined can have a considerable impact on the evaluation of marketing and pricing strategies applied by e-retailers.

Originality/value

Consumer preference and the respective WTP for different delivery methods in e-grocery, especially for the pick-up from locker option, have not been systematically investigated.

Keywords

Citation

Milioti, C., Pramatari, K. and Zampou, E. (2021), "Choice of prevailing delivery methods in e-grocery: a stated preference ranking experiment", International Journal of Retail & Distribution Management, Vol. 49 No. 2, pp. 281-298. https://doi.org/10.1108/IJRDM-08-2019-0260

Publisher

:

Emerald Publishing Limited

Copyright © 2020, Emerald Publishing Limited


1. Introduction

Over the last years, e-grocery growth has been slower than expected as shown by the small share of e-grocery in the total grocery market. Delivery constraints are recognized as the main barriers of adopting e-grocery (Hays et al., 2005). Preservation conditions of fresh products, on-time delivery requirements and low value to volume of goods ratio make last mile distribution a complex and costly activity. On the other hand, consumers' willingness to pay (WTP) in e-grocery is low (Vyt et al., 2017). Several delivery schemes including home delivery and click-and-collect have been developed to attract consumers and make e-grocery delivery more efficient (Saskia et al., 2016; Vakulenko et al., 2019). The global lock-down conditions created by COVID-19 and the respective increasing demand for e-grocery have also brought to the forefront the discussion about e-grocery delivery methods.

The aim of this research is to investigate acceptability of different delivery methods in e-grocery and the respective willingness of customers to pay for them. The analysis is based on a stated preference ranking experiment that is used to assess delivery mode choice options in e-grocery market. We focus on three essential parameters that shape delivery conditions, namely delivery method, time window and cost, and we attempt to evaluate their importance according to consumer perceptions. Three delivery methods were evaluated. Picking-up from the store, which was taken as a basis for comparison, home-delivery and picking-up from a locker. Time window and cost are closely related. In the case of home delivery, a large time window decreases delivery cost and is convenient for the retailer but inconvenient for the customer. The reverse is true for the click-and-collect method (picking-up from store or locker).

Data collected involved two countries (Greece and UK) with different level of e-grocery development and two different distribution conditions (weekly and urgent). A rank-ordered logit model is implemented to analyse the ranking question. Delivery mode option, cost and time window are used as independent variables. WTP measures for consumers with different characteristics are also obtained. Results are presented in four successive steps (studies): Study 1 investigates distribution choices of consumers for a weekly shopping list purchase (scenario 1). Study 2 explores consumer choices in an urgent e-grocery purchase setting (scenario 2). Study 3 tests the two scenarios in a different market with different socio-economic conditions and different level of maturity with respect to online grocery, in order to explore whether there exist differences in the WTP measures in the two markets. Finally, study 4 investigates WTP measures for different groups of consumers. Apart from the theoretical contributions of this study, the analysis is expected to contribute to an informed design of delivery options for online grocery retailing and be mutually beneficial to both the online retailers and the customers.

2. Background

Several delivery schemes have been developed in e-grocery retailing to make it efficient and profitable. Hübner et al. (2016) summarised the main schemes in last-mile distribution of e-grocery retailing. The design characteristics of each scheme include delivery mode, delivery time and delivery area. With respect to delivery mode, home delivery and click-and-collect are the main delivery methods in e-grocery. Two concepts of home delivery exist: attended and unattended home delivery, which differ in whether they require the presence of the customer or not. Attended home delivery has the largest market share (Hübner et al., 2016). Methods for improving attended home delivery both from the consumer and the e-grocer's point of view have been proposed by several researchers (Tanskanen et al., 2002; Punakivi and Saranen, 2001; Yrjola, 2001; Chen et al., 2012; Liao et al., 2011). Recently, Pan et al. (2017) proposed a data mining approach to enhance attended home delivery in e-grocery by reducing travelled distance and increasing the success rate of first round delivery.

Unattended delivery, using either customer-specific reception boxes installed at the customer's yard or cooled delivery boxes, enables the delivery of an online order without the presence of consumers (Punakivi et al., 2001; Hübner et al., 2016) and is considered to be more efficient in terms of time and cost both for consumers and for e-grocers (Tanskanen et al., 2002). Consumers' perceptions about unattended delivery in e-grocery were investigated by Goethals et al. (2012), using data collected from 245 face-to-face interviews with French consumers. Results showed that significant differences in unattended delivery adoption exist among consumers belonging to different age groups.

The click-and-collect option allows for ordering online and picking the order from the store or a locker. Picking the order from the store is easier for the retailer, since it does not require any additional infrastructure and respective investment cost. However, in-store picking is less convenient for the consumer, who needs to drive to the store to pick-up the order. On the other hand, picking the order from a locker saves time and cost, since consumers can pick their order on their way home. From the retailers' perspective, although this delivery method requires investments (Vyt et al., 2017) it helps in reducing distance travelled and related costs (Weltevreden, 2008; Punakivi and Tanskaren, 2002) compared to home delivery. Although the development of the click-and-collect points in the context of e-commerce has been investigated in the literature (Morganti et al., 2014; Kedia et al., 2017; Vakulenko et al., 2018; Milioti et al., 2020), research in the context of e-grocery retailing remains limited. Critical success factors for the click-and-collect model in e-grocery are summarised by Colla and Lapoule (2012). Implementation of efficient marketing strategies and operation of the click-and-collect service in not densely populated areas were found to affect its successful implementation. Consumers' acceptability and WTP are considered to be very important factors for the implementation of such a service, especially considering the fact that recent researchers have shown that one of the main reasons for developing the grocery pick-up is to make the customer more loyal (Vyt et al., 2017). Some studies have pointed out that WTP in e-grocery delivery is low (Hübner et al., 2016; Goethals et al., 2012), and others have found that price is a crucial factor influencing consumers' delivery choice (Fikar et al., 2019).

Stated preference methods have been widely used in transportation and marketing research to analyse consumers' behaviour (Ben-Akiva et al., 1992) and to estimate consumers' WTP for them. The application of stated preference methods in e-commerce delivery is rather limited (Collins, 2015; de Oliveira et al., 2017). Recently, de Oliveira et al. (2017) used stated preference data to analyse potential demand for automatic delivery stations (lockers) in Brazil. However, the aforementioned studies do not refer exclusively to the e-grocery market. The contribution of this study is that it applies a stated preference ranking experiment to analyse consumers' preferences of existing and alternative delivery modes in e-grocery and calculate acceptable price levels for consumers. The particular research questions aim at investigating how delivery preferences (delivery mode, time window) are affected by (1) different shopping list features (weekly or urgent), (2) different state of e-grocery market development (low and high maturity) and (3) consumer characteristics.

3. Methodology

3.1 General framework

Acceptability of alternative delivery methods in e-grocery is investigated by organising the research procedure in the following steps. First, taking into account related studies in the literature, the factors that may affect delivery choice in e-grocery, such as cost, time constraints and consumer characteristics, were identified. Then, a questionnaire was designed to provide the necessary data (Section 3.2) which were subsequently analysed using statistical indexes (descriptive statistics) and econometric modelling to assess which factors have statistical significance on consumers' delivery choice (Section 3.3).

3.2 Questionnaire and experimental design

A detailed questionnaire was designed, in order to capture the current e-grocery consumer behaviour (habits, levels of satisfaction), delivery mode choice options, as well as the characteristics of respondents which might influence their choices. The five-part questionnaire is briefly described below.

Part A: Grocery Online buying habits. Information about online buying frequency, online buying experience (when was the first online purchase), buying purpose (special occasion, weekly needs, etc.), average spending, preferred delivery method and delivery time was recorded.

Part B: Consumer satisfaction level. Questions focusing on customer satisfaction with the delivery process were included in this part, based on the fulfilment dimension of the e-SERVQUAL model that examines whether the products are delivered the promised time and well-packaged (Zeithaml et al., 2002).

Part C: Consumer personality traits and psychology. Price consciousness and perceived time pressure have also been included in the questionnaire as factors that might influence consumers' distribution choices in e-grocery. The Price Consciousness construct describes the tendency of an individual to look for the best price in each purchase, focusing on paying low prices (Alford and Biswas, 2002). The Perceived Time Pressure construct describes the perceived pressure consumers feel during their grocery shopping concerning time spent on it (Veirmer and Kenhove, 2005).

Part D: Ranking question. In order to investigate delivery choice options in e-grocery market, we examined six different options concerning delivery conditions that differ in three attributes: the delivery method, the size of the time window and the delivery charges. We have chosen the values of the attributes, considering that five euros is the maximum value that most customers would be prepared to pay and one hour is the minimum time interval for a time window delivery and that, in general, home delivery is more expensive than click-and-collect. Having three different attributes with three and two levels each, the possible combinations to be presented to the participants would be 18. Since participants would be asked to rank the possible combinations, 6 alternatives (options) out of the 18 were chosen to be presented basically according to their degree of possible realisation. The design is balanced meaning that each level of each parameter appears the same number of times within the presented alternatives. Respondents were asked to rank these alternatives according to their preference, assuming that they were making a weekly order online in the first ranking question and that they were making an urgent order to prepare a meal for an unplanned friends' visit in the second. The alternatives that have been chosen and included in the ranking question are presented in Figure 1.

Part E: Demographics. Finally, demographic characteristics of the respondents (age, gender, occupation, income) were included in the final part of the questionnaire.

3.3 Econometric modelling

In order to model the ranking question, the rank-ordered logit model was used (Beggs et al., 1981). The rank-ordered logit model can be applied to analyse how decision makers combine attributes of alternatives into overall evaluations of the attractiveness of these alternatives. Specifically, it was explored how the time window, the delivery method and the cost affect consumers' decisions. Rank-ordered data provide more information than a choice experiment that asks to choose the most preferred alternative (Hausman and Ruud, 1987).

Let J be the number of available alternatives (Figure 1). Based on the random utility framework, the random utility of individual i for alternative j is defined as:

Uij=Vij+εij
where Vij is the deterministic component of the utility and ɛij the random component of the utility, i = 1,…N indexes individuals and j = 1,…J indexes the different delivery options (items). The deterministic part of the utility function is modelled as
Vij=(β1HomeDelivery+β2Locker+β3TimeWindow+β4Cost)ij

Three dummy variables were used to describe the delivery mode: home delivery, locker and store. Since n−1 dummy variables should be included in the model, the binary variable store was excluded from the model and represents the reference delivery method. This note is very important for the interpretation of the results. A zero-one coding was used to express the convenience of the time window. Zero corresponds to a convenient time window and one to a non-convenient one. Note that in the home delivery case, the one-hour time window is more convenient than the three-hour one. However, in the click-and-collect (pick up from locker or store) option the reverse is true. The variable cost has three levels, 0 (free of charge), 2 and 5 euros. It should be noted that respondents who answered that none of the delivery options satisfies them (approximately 18% of the total questionnaires) were excluded from the analysis.

Let yij be the rank that a consumer i gives to delivery option (item) j. The vector yi = (yi1, ….yiJ)' denotes the full response of consumer i to the ranking experiment. An equivalent notation ri = (ri1, …riJ)', denotes the item number that received rank j by consumer i.

A consumer ranks the utilities from the most preferred (higher utility) to the least preferred (lower utility). Thus, a full ranking implies that:

Uiri1>Uiri2>>UiriJ

The rank-ordered logit model is a generalisation of the multinomial logit model: first, you choose 1 item (your favourite) from the full set of options available to you; then, you choose the next favourite from the remaining items, and you continue until some limit is reached (Fok et al., 2012). According to Beggs et al. (1981) and Chapman and Staelin (1982), the probability of observing a particular ranking is a product of multinomial logit probabilities:

(Ri)=j=1J1exp(Virij)l=jJexp(Viril)

However, Chapman and Staelin (1982) noted that for the less preferred items, the random utility assumption does not always hold, probably due to the fact that the respondent has no experience with some of the items or due to the fact that respondents tend to find the least preferred items less important and rank them randomly. In practice, this means that if the least preferred items are not ranked according to the underlying utility model, the use of those ranks in the estimation will lead to a bias in the parameter estimates. Thus, the k lower rankings should not be included in the estimation procedure (Chapman and Staelin, 1982). For these reasons, in our case only the first three ranks were taken into consideration, while the remaining k = 3 lower rankings were excluded from the analysis.

A useful outcome of the rank-ordered logit model is the estimation of the amount an individual is willing to pay to obtain benefit from undertaking some specific action. WTP can be calculated as the ratio of corresponding coefficients β1 (Home delivery), β2 (Locker) and β3 (Time window) to the cost coefficients β4 (Cost). Of course, results obtained depend to some extent on the values of the attributes used to construct the six alternatives in the ranking experiment.

3.4 Data collection and sample characteristics

Greece and UK are two countries with different level of maturity with respect to online grocery retailing; in 2017, the share of consumers who purchased food or groceries online was 3 and 29% in Greece and UK respectively (Eurostat, 2020). The questionnaires were completed in both countries online (during February 2016 in Greece and during November 2016 in UK). Consumers were invited to participate through banners that were placed on popular e-supermarket web sites. The sample consists of consumers that have shopped grocery online at least once. The data collection in Greece resulted in 170 fully completed questionnaires, while the UK survey comprised 367 responses. Demographic characteristics of the two samples and consumers' e-grocery habits in the two countries are presented in Table 1.

4. Results

4.1 Study 1: weekly order in the Greek market

In the context of the choice experiment, respondents were asked to rank the six different delivery options according to their preference, assuming that they were making an order online for their weekly grocery needs in the first scenario (Figure 2).

Distribution preferences of consumers were accessed using two complementary approaches. First, by ranking the alternatives included in the choice experiment according to consumers' preferences (descriptive statistics). Second, by using econometric analysis to explain the ranking (dependent variable) using as explanatory variables the cost, the time window and the delivery mode. This method provides WTP measures for the various factors examined.

4.1.1 Study 1: descriptive statistics

To assess the ranking of each alternative on the basis of the sample characteristics, two measures were adopted. First, the percentage of first choices of each alternative in the sample (R1) and, second, the average value of ranking in the sample (R2). Results appear in Table 2. The two measures R1 and R2 give almost the same results, especially for the first two places, thus indicating the robustness of the ranking (Table 2). Pick-up from locker and home delivery occupy the first and the second place respectively in both measures, while the remaining alternatives follow by a clear margin. Specifically, the option that was ranked first by 43.50% of the respondents was to collect the order from a locker that was closer to them than the retail store, with a pre-defined pick-up time window of one hour and free-of-charge (first place: 43.50%, M: 2.07). The second choice was home delivery with a time window of three hours and a charge of two euros (first place: 30.40%, M: 2.43). As the third option in order of preference, looking at the mean value, we can consider the alternative of picking-up the order from the store with one-hour time window and free of charge (first place: 10.10%, M: 3.15). However, slightly more respondents gave their first choice to home-delivery with one-hour time window and 5 euros charging (first place: 13.80%, M: 3.46).

4.1.2 Study 1: rank-ordered logit model results

In this section the rank-ordered logit model is used to analyse the rankings of the alternatives made by consumers. It uses richer information about the comparison of alternatives, namely, how consumers (decision-makers) rank the alternatives rather than just specifying the alternative that they like best. The dependent variable for the rank-ordered logit model is the respondents' ranking in the choice set and is expressed as a function of the delivery attributes (independent variables). Estimation results including estimated coefficients (βi), statistical significance of the parameters and WTP measures are presented in Table 2. All the coefficients (βi) have the expected signs. The negative sign on the coefficient related to price indicates that cost negatively affects choices; increases in the cost variable decrease the likelihood of the alternative to be ranked first. The negative sign of the coefficient related to time window decreases the utility function indicating that a non-favourable time window decreases the likelihood of the alternative to be ranked first, as one expects. However, this variable was not found to be statistically significant in the scenario of the weekly shopping list, probably indicating that consumers do not base their choices on the delivery (in case of home delivery) or picking (in the case of locker and store) time window when they make their weekly e-grocery shopping. The coefficients of the variables “pick-up from locker” and “home delivery” depict the contrast with the reference category “pick-up from store”. These positive estimates show that consumers are more likely to choose the option of home delivery and pick-up from locker compared to pick-up from store. Specifically, the coefficients of 2.09 and 0.97 for home delivery and pick-up from locker respectively reveal that home delivery is more preferred than the pick-up from locker option, holding all other parameters constant. Thus, model results show a clear advantage of home delivery, if distribution mode is explained independently. The fact that the option for pick-up from locker free of charge with one-hour time window was ranked first (Table 2) demonstrates the impact that the values of time window and cost have on consumer preference.

The preference towards the home delivery method is also reflected in the WTP measures. WTP measures correspond to the amount of money that an individual would be willing to pay for achieving a benefit in time window or choosing a particular delivery method. WTP measures for a specific variable are derived by dividing the coefficient of this variable by the coefficient of the cost variable (Hensher et al., 2005). Estimation results show that, compared to picking-up their order from the store (reference category), consumers are willing to pay an additional fee of 3.64 euros for home delivery and an additional fee of 1.69 euros for the pick-up from locker option.

4.2 Study 2: urgent order in the Greek market

The second scenario that was examined included an urgent order that should be delivered on the same day to consumers. The order includes vegetables, cheese wine etc. and costs 40 euros. On one hand, same day delivery increases complexity for the e-retailers both in terms of cost and distribution planning (Hübner et al., 2016). On the other hand, same day delivery is necessary in the context of e-grocery, since the delivered products are used to cover daily necessities. The aim of this study is to investigate whether consumers change their distribution choices in case of an urgent order.

4.2.1 Study 2: descriptive statistics

According to the descriptive statistics presented in Table 3, the first choice of the respondents remains the option of picking-up the order from a locker located closer to home than the retail store with one-hour time window and free of charge (first place: 46.50%, M: 1.89). It should be noted that free of charge pick-up from locker is more preferred in the urgent order compared to the weekly order (mean value of 1.89 vs 2.07). Since response and delivery time is the issue in this scenario, the most convenient and simultaneously one of the most expensive options comes to the second place of preference. This is home delivery with one-hour time window and five-euro charge (first place: 32.50%, M: 2.90) followed by picking-up from store with one-hour time window and free of charge.

As expected, the distribution option of home delivery with one-hour time window and 5 euros charge is more preferred in the urgent order compared to the weekly order (mean 2.99 vs 3.46), while home delivery with 3 h time window and 2 euros charge is more preferred in the weekly order compared to the urgent order (mean 2.43 vs 3.01).

4.2.2 Study 2: rank-ordered logit model results

Estimation results of the rank-ordered logit model for the urgent order scenario are presented in Table 3. All the variables included in the models are statistically significant at the 5% level or higher.

The fact that the variable “Time window” is statistically significant in the 2nd scenario (urgent order) indicates that convenience of the time window is now one of the factors that shape consumers' choices. Specifically, Greek consumers are willing to pay 1.16 euros for a more convenient time window in the case of the urgent order.

Similarly to the model of the weekly shopping scenario, home delivery and pick-up from locker are more preferred compared to pick-up from store in the urgent order scenario. Thus, as expected, consumers are willing to pay more for the home delivery option and less for the pick-up from locker option. Comparing the weekly and the urgent order, consumers are willing to pay more for the pick-up from locker option in the urgent order scenario. This finding indicates that picking-up from locker is more attractive in the case of a small urgent order rather than in the case of a planned weekly order.

4.3 Study 3: UK market (weekly and urgent order)

We ran the ranking experiment also in UK, in order to compare distribution choices of consumers in the two markets that exhibit different socio-economic conditions and have different level of maturity with respect to online grocery retailing.

4.3.1 Study 3: descriptive statistics

In the weekly shopping scenario, the choice that has been most preferable of all the presented ones is the home delivery option with a three-hour time window at the cost of £2 (first place: 38.00%, M = 2.18). In comparison, Greek consumers show a preference for free-of-charge pick-up from locker with one-hour time window (Table 4).

Continuing with the urgent scenario presented to the participants and their corresponding preferences, picking-up the order from a locker closer to home with one-hour time window free of charge comes first in preference between the proposed options both in UK and in Greece; however, this distribution option is more preferred in Greece (M = 1.89 vs M = 2.21 for UK). Interestingly, there is a similar behaviour in the two countries for the urgent order despite all other differences in consumers' socio-economic conditions and market maturity level.

4.3.2 Study 3: rank-ordered logit model results

Estimation results and the respective WTP measures for the UK market both for the weekly and the urgent scenario are presented in Table 5.

Comparing the WTP measures in the two countries, Greek consumers are willing to pay less for the home delivery option compared to UK, especially in the weekly order, while WTP for the pick-up from locker option is almost equal in the two countries both in the weekly and in the urgent scenario.

Results also indicate that Greek consumers are willing to pay more for a more convenient time window compared to UK consumers in the urgent order scenario. In both countries, comparing the weekly and the urgent order, consumers are willing to pay less for the home delivery option and more for the pick-up from locker option in the urgent order. In UK, WTP for the home delivery option is significantly reduced in the urgent order (5.38 vs 3.68).

4.4 Study 4: different consumer groups

The final choice is also shaped by a combination of consumers' personality traits and satisfaction. Consumers' perceptions about time pressure, price consciousness and satisfaction with existing delivery services are some of the possible factors that may affect the final choice of delivery method. In order to explore distribution choices in e-grocery for different consumer groups, we divided the sample into sub-groups based on three constructs: (1) Price consciousness, (2) Perceived time-pressure, (3) Satisfaction with home delivery fulfilment. Different subgroups were formed by exploiting information derived from Part B and C of the questionnaire. Subsequently, we ran different models for each group and we calculated the WTP measures. Table 6 presents the WTP measures for the different consumer groups. Results are similar in the two studies conducted in Greece and the UK. For simplicity and clarity, results are presented for the Greek sample, both for the weekly and the urgent order scenario.

The price consciousness construct consists of five items measured on a five-point Likert scale (Alford and Biswas, 2002). The alpha coefficient for the five items is 0.877, suggesting that the items have relatively high internal consistency. The k-means clustering algorithm was used to classify consumers into high and low price-consciousness groups. Results from the econometric analysis showed that, as expected, price conscious consumers are willing to pay less both for the delivery and the pick-up from locker option. In the weekly order scenario, WTP measures do not vary significantly between the two groups. However, in the urgent order scenario, there is a 34% and 23% change in the WTP for home delivery and pick-up from locker option respectively between the two groups (price conscious and non-price conscious). Interestingly, price conscious consumers pay more for the home delivery option in the weekly order scenario compared to the urgent one, but slightly less for the pick-up from locker. Time window is not significant for the price conscious consumers, probably indicating that this group category bases its choices on the cost and the delivery method attributes.

Concerning the time pressure construct, the value of Cronbach's alpha is 0.908, indicating that the five items used measure the same underlying concept. The variation in the WTP measures between the two groups is not significant (in magnitude). Specifically, in the weekly order scenario, WTP measures for the home delivery and the pick-up from locker option are almost equal in the two groups. In the urgent order scenario, consumers with perceived time pressure are willing to pay slightly more for the home delivery and the pick-up from locker options compared to those not feeling under time pressure. As expected, time window is statistically significant for the former group, while the opposite applies for the latter. Specifically, consumers feeling under time pressure are willing to pay 1.41 euros for a more convenient time window.

Satisfaction with delivery fulfilment increases the WTP for the home delivery option, especially in the weekly order scenario (56% change in the WTP) and decreases the WTP for the pick-up from locker option in the urgent order scenario. Moreover, in the urgent order scenario, time window is significant and shows an increased WTP for those who are satisfied with the delivery services.

5. Discussion

Understanding consumers' preferences for different delivery methods is essential for designing effective e-grocery delivery schemes and efficient pricing policies.

The results of this study demonstrate that, in the case of a planned weekly order, consumers show a clear preference for home delivery and picking-up from locker over picking-up from the store both for the Greek and the UK case. This is evident both from the descriptive statistics and the econometric model results. According to the econometric analysis, consumers show a relative preference for home delivery compared to the pick-up from locker option and are willing to pay more for it. This preference is more distinct in the UK market, where home delivery in e-grocery is more developed (Hübner et al., 2016) and consumers, especially those who are satisfied with home delivery services, show a resistance to change. However, when the method of delivery is combined with specific cost and time window attributes, consumers' acceptability of the pick-up from locker option is increased significantly. These findings are in line with the studies of de Oliveira et al. (2017) and Hood et al. (2020) who found that although home delivery is the most attractive delivery solution, the click-and-collect (pick-up from locker) option shows great potential demand for consumers.

In the context of an urgent order (same-day delivery), WTP for the pick-up from locker option is slightly increased, while WTP for the home delivery option is decreased compared to the weekly order. This applies also to the UK case experiment (Study 3), confirming that picking-up from locker increases its competitive position in the urgent order. This information is very useful for policymakers, due to the fact that, on one hand the cost of home delivery in the case of same day delivery increases significantly, and on the other, the demand for same day delivery by consumers is showing an increasing trend (Lin et al., 2018). Thus, the pick-up from locker option appears to be beneficial both for consumers and for e-retailers. Since the variable “time window” is statistically significant in this scenario, an additional fee for a convenient (more flexible) time window could be applied, especially for the urgent (same day delivery) orders. Specifically, consumers are willing to pay an additional fee of 1.16 euros for choosing a more convenient time window (one hour instead of three in the home delivery option and three hours instead of one in the pick-up from store/locker options).

The analysis per group of consumers revealed that consumers with perceived time pressure are willing to pay 1.41 euros for a more convenient time window. Moreover, group survey results indicated that an individual's level of price consciousness seems to influence delivery mode choice and the respective WTP. This finding is in line with Hood et al. (2020) who found that affluent households are more likely to choose the home delivery option.

An important finding of the study is that consumers are willing to pay more for a more convenient delivery method of their choice (i.e. home delivery and smaller time window) if they are in general satisfied with delivery. This means that prior experience and satisfaction levels with delivery methods may be an important indicator for guiding pricing decisions. For example, dynamic pricing schemes may adjust pricing according to customer satisfaction, which may be a known parameter given the increasing trend for capturing online customer satisfaction.

Thus, exploring delivery preferences by groups of consumers may provide some indication for the level of acceptable prices (upper and lower limits for the pricing) as well as for the price structure that should be applied to the various distribution conditions (urgent vs weekly order). In addition, it is argued that WTP differences among the groups examined can have a considerable impact on the evaluation of marketing and pricing strategies applied by e-retailers.

6. Conclusions

In this study, we conduct a stated preference ranking experiment in order to investigate delivery mode choice options in the e-grocery market and calculate consumers' WTP for various distribution schemes that differ in the delivery method (home delivery, pick-up from store, pick-up from locker) the cost and the convenience of the time window. Results obtained involved different distribution conditions (weekly and urgent order) for two countries, Greece (Study 1, 2) and UK (Study 3) with different level of e-grocery development.

Results of the study are similar in two different countries, Greece and the UK, despite varying socio-economic conditions and different level of maturity with respect to online grocery retailing. Home delivery appears to hold a strong position among the distribution modes examined, especially concerning the weekly order, while pick-up from locker can be developed to a competitive alternative for urgent orders in both markets, especially when low-cost policies are applied. Time-window has also a significant impact on choice in the case of urgent need for same-day delivery, and this is stronger for consumers feeling under time-pressure. Thus, the information provided in this study will help online grocery retailers to effectively design and implement distribution schemes taking into account the preferences and values of different consumer groups classified according to their perceptions on quality fulfilment, time value and cost considerations.

The higher adoption of click-and-collect delivery options may have important implications from a societal and environmental perspective as well, as it may affect carbon emissions, traffic congestion, the city landscape, etc. This could be an interesting path for further research following the findings of this study.

Although the study provides some important implications for retailers and logistics companies who plan to design and implement delivery services in the online grocery sector, we recognise the limitations of the study. This study used a scenario setting with predefined distances to store and the locker points, as well as pre-defined type of grocery orders (value and products). In future studies, a more extensive classification of the type of orders considering the order size, the type of categories, the value or even the shopping mission could be used to reveal any differences in consumers' delivery preferences and respective WTP measures. Proximity of the click-and-collect points (retail stores or lockers) is also an aspect that could affect delivery options and the respective WTP measures (Davies et al., 2019). The survey captures the perception of consumers that are located in metropolitan areas with high density of retail stores. Consumers in suburban or rural areas with lower density of retail stores and more limited product assortment may behave differently and are more willing to pay for home delivery or the pick-up from locker option. Considering the fast growing online grocery sector, useful insights could also be collected by conducting a longitudinal study that could show how the aforementioned findings evolve over time.

Figures

Available options

Figure 1

Available options

Weekly shopping list scenario

Figure 2

Weekly shopping list scenario

Demographic characteristics and online grocery habits of the consumers

Greek surveyUK survey
VariablesSample rate (%)VariablesSample rate (%)
Gender
Men56.6Men44.0
Women43.4Women55.0
Age
Age<218.4
Age 18–2614.8Age 22–3440.3
Age 27–4048.5Age 35–4426.9
Age 41–5530.1Age 45–5412.1
Age>556.6Age>5512.3
Occupation
Government employee8.8Full-time employed61.8
Private employee38.2House maker/house wife5.8
Self employed29.4Self employed7.3
Unemployed9.6Unemployed3.6
Student6.6Student10.2
Other5.9Other11.3
Income
Income<12.000€37.3Income<20.000£14.9
Income 12.001–22.000€29.9Income 20.001–39.999£33.7
Income 20.001–40.000€21.6Income 40.00–59.999£24.3
Income 40.001–60.000€5.2Income 60.00–79.999£15.6
Income>60.000€6.0Income >80.000£11.6
Delivery method preference*
Home delivery85.4Home delivery87.1
Pick-up from store12.2Pick-up from store5.7
Pick-up from locker8.7Pick-up from locker3.2
Average online basket
<40€18.9<40£10.7
41–60€23.141–60£32.0
61–80€25.461–80£29.1
81–120€21.981–120£22.7
121–150€7.7121–150£4.3
>150€3.0>150£1.2
Number of observations170 367

Rankings and model results of the weekly shopping list scenario (Greece)

Rankings and model results of the urgent shopping list scenario (Greece)

Rankings of the weekly and the urgent order in UK

Rank-ordered model results in UK

Weekly orderUrgent order
VariableCoefficientProbabilityWTP in pounds (euros)CoefficientProbabilityWTP in pounds (euros)
Home delivery1.950.0005.38 (6.46)1.780.0003.68 (4.42)
Pick-up from locker0.500.0001.38 (1.65)0.790.0001.64 (1.97)
Time window0.240.355 −0.350.135ns
Cost−0.360.000 −0.480.000
N observations276 289
Log likelihood−1056.09 −1178.26

WTP measures (euros) for the different consumer groups in Greece

Construct Cronbach's alphaScenarioAttributesConsumer groupsPercentage change
Non-price consciousPrice conscious%
Price consciousness a = 0.877Number of respondents (weekly/urgent)(46/41)(87/71)
WeeklyHome delivery3.763.23−14
Pick-up from locker1.691.567
Time window1.22(ns)0.22(ns)-
UrgentHome delivery4.102.71−34
Pick-up from locker2.101.62−23
Time window1.771.00 (ns)−43
Non-time pressuredTime pressured
Time pressure a = 0.908Number of respondents (weekly/urgent)(73/59)(60/53)
WeeklyHome delivery3.533.520
Pick-up from locker1.671.58−5
Time window0.96(ns)0.12 (ns)-
UrgentHome delivery3.173.26+2
Pick-up from locker1.71.89+11
Time window0.90 (ns)1.41+56
Non-satisfiedSatisfied
Satisfaction with delivery fulfilment a = 0.732Number of respondents (weekly/urgent)(62/53)(71/59)
WeeklyHome delivery2.654.13+56
Pick-up from locker1.511.79+19
Time window0.04(ns)0.38(ns)-
UrgentHome Delivery2.803.62+29
Pick-up from locker1.961.67−15
Time window0.88 (ns)1.48+68

Note(s): ns = not significant in the econometric model at the 10% level of significance

References

Alford, B.L. and Biswas, A. (2002), “The effects of discount level, price consciousness and sale proneness on consumers' price perception and behavioural intention”, Journal of Business Research, Vol. 55, pp. 775-783.

Beggs, S., Cardell, S. and Hausman, J. (1981), “Assessing the potential demand for electric cars”, Journal of Econometrics, Vol. 17 No. 1, pp. 1-19.

Ben-Akiva, M., Morikawa, T. and Shiroishi, F. (1992), “Analysis of the reliability of preference ranking data”, Journal of Business Research, Vol. 24 No. 2, pp. 149-164.

Chapman, R. and Staelin, R. (1982), “Exploiting rank ordered choice set data within the stochastic utility model”, Journal of Marketing Research, Vol. 19, pp. 288-301.

Chen, M.C., Hsu, C.L. and Lee, Y.Y. (2012), “Applying quality function development to develop the home delivery service model for specialty foods in traditional market”, Industrial Engineering and Engineering Management (IEEM), pp. 1741-1745, Hong Kong, 2012.

Colla, E. and Lapoule, P. (2012), “E-commerce: exploring the critical success factors”, International Journal of Retail and Distribution Management, Vol. 40 No. 11, pp. 842-864.

Collins, A.T. (2015), “Behavioural influences on the environmental impact of collection/delivery points”, Green Logistics and Transportation, Springer, Cham, pp. 15-34.

Davies, A., Dolega, L. and Arribas-Bel, D. (2019), “Buy online collect in-store: exploring grocery click and collect using a national case study”, International Journal of Retail and Distribution Management, Vol. 47 No. 3, pp. 278-291.

de Oliveira, L.K., Morganti, E., Dablanc, L. and de Oliveira, R.L.M. (2017), “Analysis of the potential demand of automated delivery stations for e-commerce deliveries in Belo Horizonte, Brazil”, Research in Transportation Economics, Vol. 65, pp. 34-43.

Eurostat (2020), available at: https://www.statista.com/statistics/700676/share-of-individuals-who-purchased-groceries-online-in-the-uk/ (accessed 17 September 2020).

Fikar, C., Mild, A. and Waitz, M. (2019), “Facilitating consumer preferences and product shelf life data in the design of e-grocery deliveries”, European Journal of Operational Research. doi: 10.1016/j.ejor.2019.09.039.

Fok, D., Paap, R. and Van Dijk, B. (2012), “A rank‐ordered logit model with unobserved heterogeneity in ranking capabilities”, Journal of Applied Econometrics, Vol. 27 No. 5, pp. 831-846.

Goethals, F., Leclercq-Vandelannoitte, A. and Tütüncü, Y. (2012), “French consumers' perceptions of the unattended delivery model for e-grocery retailing”, Journal of Retailing and Consumer Services, Vol. 19 No. 1, pp. 133-139.

Hausman, J.A. and Ruud, P.A. (1987), “Specifying and testing econometric models for rank-ordered data”, Journal of Econometrics, Vol. 34 Nos 1-2, pp. 83-104.

Hays, T., Keskinocak, P. and De López, V.M. (2005), “Strategies and challenges of internet grocery retailing logistics”, Applications of Supply Chain Management and E-Commerce Research, Springer, Boston, MA, pp. 217-252.

Hensher, D.A., Rose, J.M. and Greene, W.H. (2005), Applied Choice Analysis – A Primer, Cambridge University Press, New York.

Hood, N., Urquhart, R., Newing, A. and Heppenstall, A. (2020), “Sociodemographic and spatial disaggregation of e-commerce channel use in the grocery market in Great Britain”, Journal of Retailing and Consumer Services, Vol. 55, p. 102076.

Hübner, A., Kuhn, H. and Wollenburg, J. (2016), “Last mile fulfilment and distribution in omni-channel grocery retailing: a strategic planning framework”, International Journal of Retail and Distribution Management, Vol. 44 No. 3, pp. 228-247.

Kedia, A., Kusumastuti, D. and Nicholson, A. (2017), “Acceptability of collection and delivery points from consumers' perspective: a qualitative case study of Christchurch city”, Case Studies on Transport Policy, Vol. 5 No. 4, pp. 587-595.

Liao, S.H., Chen, Y.J. and Lin, Y.T. (2011), “Mining customer knowledge to implement online shopping and home delivery for hypermarkets”, Expert Systems with Applications, Vol. 38 No. 4, pp. 3982-3991.

Lin, J., Zhou, W. and Du, L. (2018), “Is on-demand same day package delivery service green?”, Transportation Research Part D: Transport and Environment, Vol. 61, pp. 118-139.

Milioti, C., Pramatari, K. and Kelepouri, I. (2020), “Modelling consumers' acceptance for the click and collect service”, Journal of Retailing and Consumer Services, Vol. 56, p. 102149.

Morganti, E., Dablanc, L. and Fortin, F. (2014), “Final deliveries for online shopping: the deployment of pickup point networks in urban and suburban areas”, Research in Transportation Business and Management, Vol. 11, pp. 23-31.

Pan, S., Giannikas, V., Han, Y., Grover-Silva, E. and Qiao, B. (2017), “Using customer-related data to enhance e-grocery home delivery”, Industrial Management and Data Systems, Vol. 117 No. 9, pp. 1917-1933.

Punakivi, M. and Saranen, J. (2001), “Identifying the success factors in e-grocery home delivery”, International Journal of Retail and Distribution Management, Vol. 29 No. 4, pp. 156-163.

Punakivi, M. and Tanskanen, K. (2002), “Increasing the cost efficiency of e-fulfilment using shared reception boxes”, International Journal of Retail and Distribution Management, Vol. 30 No. 10, pp. 498-507.

Saskia, S., Mareï, N. and Blanquart, C. (2016), “Innovations in e-grocery and logistics solutions for cities”, Transportation Research Procedia, Vol. 12, pp. 825-835.

Tanskanen, K., Yrjölä, H. and Holmström, J. (2002), “The way to profitable Internet grocery retailing–six lessons learned”, International Journal of Retail and Distribution Management, Vol. 30 No. 4, pp. 169-178.

Vakulenko, Y., Hellström, D. and Hjort, K. (2018), “What's in the parcel locker? Exploring customer value in e-commerce last mile delivery”, Journal of Business Research, Vol. 88, pp. 421-427.

Vakulenko, Y., Shams, P., Hellström, D. and Hjort, K. (2019), “Service innovation in e-commerce last mile delivery: mapping the e-customer journey”, Journal of Business Research, Vol. 101, pp. 461-468.

Veirmer, I. and Kenhove, P.V. (2005), “The influence of need for closure and perceived time pressure on search effort for price and promotional information in a grocery shopping context”, Psychology and Marketing, Vol. 22 No. 1, pp. 71-95.

Vyt, D., Jara, M. and Cliquet, G. (2017), “Grocery pickup creation of value: customers' benefits vs spatial dimension”, Journal of Retailing and Consumer Services, Vol. 39, pp. 145-153.

Weltevreden, J.W. (2008), “B2c e-commerce logistics: the rise of collection-and-delivery points in The Netherlands”, International Journal of Retail and Distribution Management, Vol. 36 No. 8, pp. 638-660.

Yrjola, H. (2001), “Physical distribution considerations for electronic grocery shopping”, International Journal of Physical Distribution and Logistics Management, Vol. 31 No. 10, pp. 746-761.

Zeithaml, V.A., Parasuraman, A. and Malhotra, A. (2002), “Service quality delivery through web sites: a critical review of extant knowledge”, Journal of the Academy of Marketing Science, Vol. 30 No. 4, pp. 362-375.

Further reading

Zhou, M., Zhao, L., Kong, N., Campy, K.S., Xu, G., Zhu, G., Cao, X. and Wang, S. (2020), “Understanding consumers' behavior to adopt self-service parcel services for last-mile delivery”, Journal of Retailing and Consumer Services, Vol. 52, p. 101911.

Acknowledgements

This research has been conducted in the context of the U-TURN H2020 project, co-funded by the European Commission. The authors would like to express their gratitude to all the project partners and especially to Professor Michalis Bourlakis, Dr. Emel Aktas and Dr. Dimitris Zissis from Cranfield University and Richard Walters and Alan Braithwaite from LCP for their support in data collection in the UK market. Moreover, they would like to express their great appreciation to Professor Panagiotis Miliotis for his valuable and constructive feedback during the writing of this paper.

Corresponding author

Christina Milioti is the corresponding author and can be contacted at: c.milioti@uniwa.gr

About the authors

Christina Milioti is an assistant professor at the Department of Civil Engineering of the University of West Attica (UNIWA). She holds a Diploma in Civil Engineering from the National Technical University of Athens (NTUA), a MSc degree in Mathematics of Production and Finance from the Athens University of Economics and Business and a PhD from NTUA in the research area of transport demand analysis. Her areas of research and expertise involve econometric transport modelling, travel behaviour and statistics, transport economics, traffic and transport surveys and supply chain management. She has published more than 40 papers in scientific journals and peer-reviewed academic conferences.

Katerina Pramatari is an associate professor at the Department of Management Science and Technology of the Athens University of Economics and Business and scientific coordinator of the ELTRUN/ SCORE research group. Her research interests lie in the areas of re- tail analytics, supply chain information systems, digital innovation and entrepreneurship. She has received various academic distinctions and scholarships and has published more than 100 papers in scientific journals, peer-reviewed academic conferences and book chapters, amongst them in Decision Support Systems, Journal of Information Technology, European Journal of Information Systems, Journal of Re- tailing, Journal of Business Research, Journal of Strategic Information Systems, etc.

Eleni Zampou is a research associate at the ELTRUN research laboratory, Department of Management Science and Technology, Athens University of Economics and Business (AUEB). She holds a PhD on Environmental Informatics in the supply chain from AUEB. She has received her Diploma in Computer Engineering and Informatics and her MSc in Computer Science from the Department of Computer Engineering and Informatics of the University of Patras. She also holds a MSc in Information Systems from AUEB. She has been awarded with scholarships for her PhD and MSc studies due to academic excellence. She has a multi-disciplinary background and her current research interests include environmental informatics, intelligent transport systems, sustainable supply chain management, collaborative logistics and supply chain management and e-commerce logistics. She has applied research and industry experience in the design and development of solutions in the aforementioned areas. She also has long-term experience in the coordination of international research activities in the area of information systems for logistics and supply chain management.

Related articles