You are on page 1of 13

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0959-0552.

htm

Understanding shoppers expectations of online grocery retailing


Muriel Wilson-Jeanselme and Jonathan Reynolds
d Business School, Templeton College, University of Oxford, Oxford, UK Sa
Abstract
Purpose The purpose of this paper is to analyse the online preference structures of consumers. Design/methodology/approach Novel choice-based conjoint experiments are used and are administered online. A select group of high net worth online grocery shoppers are examined. Both qualitative and quantitative procedures are used to determine the most frequently cited attributes affecting online patronage. Findings Whilst there is no single attribute on which a retailer could develop a competitive edge, a signicant market advantage can be gained by being simultaneously best in class on the top four attributes. Practical implications This research approach has signicant practical application to a wide range of strategic marketing questions. Originality/value These ndings give focus to the management task facing marketing executives in the UK multichannel grocery market. How these ndings might be used within a marketing plan is illustrated. Keywords Electronic commerce, Retailing, Consumer behaviour Paper type Research paper

Shoppers expectations of online retailing 529

Introduction Retailing has always offered a variety of shopping options to consumers. Most recently, the internet has been a particular cause of excitement and attention, in terms of its rapid growth (IMRG, 2006), its impact on consumer behaviour (Hoffman and Novak, 1996; Evans and Wurster, 1997; Ghose and Dou, 1998; Reynolds, 2003; Grewal et al., 2004; Keen et al., 2004; Wilson-Jeanselme and Reynolds, 2005) as well as its impact on price (Sinha, 2000; Brynjolfsson and Smith, 2000; Baker et al., 2001; Wilson-Jeanselme and Reynolds, 2005) and on brand (Burt and Sparks, 2003; Reynolds, 2003). Much of this work has been particularly concerned with the acquisition of customers by rst movers in the market. But as online business-to-consumer markets mature, as in the UK, online consumers are beginning to face the kinds of competitive choices hitherto only available to them ofine. In this paper we explore potential store switching criteria and characteristics for existing online grocery shoppers. In other words, what might persuade consumers who already shop online for groceries with one supermarket to switch to another online provider? Wilson-Jeanselme and Reynolds (2005) pointed out the success of Tesco.com in the UK whose online market share represented an estimated 45 per cent of the online grocery market in 2004. Grewal et al. (2002) noted that commodity and quasi-commodity products, such as many grocery products, are by nature not highly differentiated and that as a result price may be the main factor when making the decision to buy them. If this is case how will the present market leader, Tesco,

International Journal of Retail & Distribution Management Vol. 34 No. 7, 2006 pp. 529-540 q Emerald Group Publishing Limited 0959-0552 DOI 10.1108/09590550610673608

IJRDM 34,7

530

withstand the growth of price based competitive offers from other providers such as Sainsburys, Asda and Waitrose? On the other hand, if the decision making process of buying groceries online is determined by non-price factors, and Tesco provides these appropriately, then it may be that the company can sustain its competitive advantage and market share growth. Our paper seeks to evaluate the non price factors that will induce online shoppers to choose a certain grocery retailer. The paper rst calculates a set of non price consumer utilities values for online shoppers using choice-based conjoint techniques. Choice-based conjoint analysis was chosen because it provides realism in the modelling of the buying process. It also provides a unique basis for examining the interaction effects of utility values between attributes (Hair et al., 1998). These ndings are then applied to real life online grocery marketing problems with suggestions for building more robust marketing and business plans. Designing choice-based conjoint analysis for the online grocery market The choice-based conjoint study used respondents drawn from the parents of pupils of four independent schools in south-east England. Income proles and cultural preferences were therefore similar for participating parents. Income proles would also suggest that they buy a higher proportion of high margin products and services. The respondents were mainly women from 27 to 50 years old and comprised a mixture of working and non-working mothers who had between one and three children. To calibrate the conjoint analysis we undertook qualitative interviews with shoppers to identify which online store choice attributes were important to them. Twenty in-home 45 minute qualitative interviews were conducted with a random sample of the research population. Ten attributes were selected to include in the choice-based survey. Each attribute was further divided into two or three levels better to reect the choices that might be available to shoppers (Table I). Following a short pilot, a number of modications were carried out to increase the internal consistency of the questionnaire and its structure. The main survey was conducted online using Sawtooth software (www.sawtoothsoftware.com/). The four schools which participated in the survey are considered to be the population of interest. We did not take a sample of parents because they were all contactable and each was theoretically equally likely to reply. Conjoint designs are in any case generally more robust than other multivariate techniques in terms of sample size effects on validity (Hair et al., 1998). The major risk to validity lies in the non-respondents. We undertook follow up work with the non-respondents to establish any possible causes of response bias. Because contacting the whole population of interest is so rare the assumption is made that some form of sampling is always necessary. The new approach we are demonstrating to understanding internet user preferences could of course be expanded to a much larger population using classic random sampling techniques. For example, if we wanted to generalise for parents of all independent schools in Surrey we could have taken a random sample of the Surrey schools rst and then a random sample of a consolidated list of all parents in the selected schools (These parents may be proxies for high net worth, high consumption grocery customers, which in turn could be the real business target). This approach could have also been applied to the whole of the UK. There are of course numerous difculties in these approaches not least the reluctance of independent schools to participate if there is no connection with them.

Level 1

Level 2

Level 3

Delivered in 48 hours

Special offer section New product section Help line number Delivery time Delivery cost Delivery time reliability Quality Ordering time Substitutes Discount on internet prices

Has a special offer section Has a new products section Help line number Delivered in 6 hours No delivery cost Delivery at agreed time No items below quality 20 minutes to place an order No substitutes 0 per cent discount on internet prices

No new special offer section No new products section No help line number Delivered in 24 hours 5 delivery cost Delivery an hour early or late Five items below quality 35 minutes to place an order Five substitutes 10 per cent discount on internet prices

1 hour to place an order

Shoppers expectations of online retailing 531

Table I. Attributes and level of attributes used for the main survey

IJRDM 34,7

532

The survey was in two parts. The rst section sought information on underlying demographics and loyalty behaviours (such as, for example, the value of the respondents house, whether the household was dual or single income, and number of shopping trips per annum). The second section comprised 14 on-screen choice tasks, which provided realistic buying contexts from which respondents could choose. In total the main survey achieved 261 completed responses (questionnaires were considered as incomplete if one screen or one question was not completed). The useable response rate was 12.7 per cent[1]. For validation purposes, two xed holdout tasks were included (Hair et al., 1998). They were the only two choice tasks constant across all respondents as the other tasks were randomised. There was a 99 per cent t between the holdout samples and the main survey data. Figure 1 shows an example of one of the 12 choice task screens. Shoppers preferences towards online grocery brands in the UK Logit analysis was used to calculate the consumer utility values that will provide the relative importance of levels of attributes. Utility values illustrate the extent of desirability for a certain attribute level: the higher the utility, the more desirable is the attribute level:
In conjoint analysis, utility is assumed to be formed by the combination of part-worth estimates for any specied set of levels with the use of an additive model, perhaps in conjunction with interaction effects (Hair et al., 1998, p. 421).

Utility values for the selected levels of attributes are shown in Table II which estimates the utilities for online grocery shoppers by brand in the UK. Table II displays shoppers preferences towards the main online grocery brands in the UK. For this particular sample, over 50 per cent of respondents were currently shopping online with Tesco.com for groceries (slightly higher than the national estimate).
Choice task example If you were considering buying groceries online for your next grocery supplies and these were the only alternative supermarket websites, which one would you choose?
Advisory emails Bad doorstep presentation Help line number Delivered in 6 hours 5 delivery cost Delivered on time 5 items below quality 1 hour to place an order No substitutes 0% discount for online prices Advisory emails Good doorstep presentation No help line number Delivered in 24 hours No delivery cost Delivered on time 5 items below quality No advisory emails Good doorstep presentation No help line number Delivered in 48 hours No delivery cost Delivered 30 minutes late or in advance No items below quality None: I wouldn't purchase my groceries online from any of these described websites

20 minutes to place an order 1 hour to place an order No substitutes 10% discount for online prices 5 substitutes 0% discount for online prices

Figure 1. Choice task example

Make your selection by clicking within the box with the mouse
Previous

Levels Has a special offer section Has a new product section Help line number Delivered in 6 hours Delivered in 24 hours No delivery cost Delivery at agreed time No items below quality 20 minutes to place an order 35 minutes to place an order No substitutes 10 per cent discount on internet prices

Utility values for Tesco (65 respondents) 0.11 0.01 0.06 0.32 0.10 0.22 0.29 0.47 0.57 0.02 0.31 0.23

Utility values for Sainsburys (50 respondents) 0.10 0.00 0.10 0.19 0.18 0.26 0.29 0.50 0.34 0.08 0.18 0.30

Utility values for Ocado (24 respondents) 0.11 0.00 0.04 0.30 0.06 0.38 0.58 0.72 0.12 0.24

Shoppers expectations of online retailing 533

Note: The gures in italic represent level utility values which are signicantly different to the within group mean for that particular level

Table II. Utility values for three main online grocery brands

Sainsburys was identied as the closest online competitor, followed by Ocado (Waitrose). Asda and Marks & Spencer are insignicant in terms of the research population. Our analysis therefore focuses on shoppers preferences for the top three brands only. This comparison made it very clear that the preference structure for Tescos consumers differs from both that of Sainsburys consumers, which differ again from those of Ocado. It seems that Tesco online shoppers are more time sensitive than Sainsburys consumers (illustrated by the relative utility values of delivery time and time to place an order). Tescos consumers are also more sensitive to extensive use of substitutes, if we compare the utility value of substitutes with Sainsburys and Ocado. In contrast, Sainsburys consumers seem to be slightly more price-sensitive, as we note a higher utility value for delivery cost and 10 per cent discount on the internet compared with its other two competitors. Ocado consumers show some distinct preferences for order placing speed, with 35 minutes to place an order appearing as a negative value (this is due to the fact that the raw utilities are zero-centred within each attribute (Table III)). The example above shows respondents preferring 20 minutes to place an order to the other two options. The attribute level 20 minutes to place an order has the highest consumer utility value for Ocado shoppers, which means that it will have a large positive impact on inuencing respondents. This utility is therefore a good indicator for customer acquisition and retention if the company performs well on that level. Another distinct preference expressed by Ocado shoppers is the importance of quality (58 per cent). They seem to want to receive products as if they have chosen them themselves.
Level 20 minutes to place an order 35 minutes to place an order 1 hour to place an order Utility 0.72 2 0.11 2 0.60 Table III. Comparing Ocado ordering time utility values

IJRDM 34,7

534

Determining the relative importance of consumer preferences To understand UK shoppers reactions to the offerings of online grocery retailers, we next look at the relative weighting of attributes. The aim here is to determine the relative importance of each attribute that is, the weighting of each level of each attribute for each retailer. To do this we consider how much difference each attribute could make in the total utility of an online offering. That difference is expressed as the range in the attributes utility values, with the absolute utility level difference being a proportion of the sum of all absolute differences. This is shown for online shoppers in Table IV. Table IV suggests that the preferences of Ocados shoppers are much better dened, compared to the other two supermarkets. The most important factor for Ocados shoppers is time to place an order (28 per cent) followed by quality (22 per cent). The other two signicant factors are the time factors delivery time and delivery time reliability but they are only half as important as the rst two (The levels of importance are ratios: an attribute with an importance of 20 (20 per cent) is twice as important as an attribute with an importance of 10). As a result, an online retailer that might seek to acquire Ocados consumers could safely position itself against these two very strong attributes namely ordering time and quality. In other words, by ensuring efciency in placing an order online and by guaranteeing that no items of the shopping basket are below quality, online retailers could gain market share from Ocados shopper segment. Similarly, to retain existing consumers, Ocado needs to ensure that its online retention strategy focuses on these two very strong factors. However, as Ocados shoppers seem to be extremely time sensitive, with a preference of over 28 per cent to place an order, it could be worth examining consumers market share of preference with and without the next two most important attributes (which are time related) namely delivery time and delivery time reliability. We undertake this market simulation as shown in Table V.
Tesco online shoppers Sainsburys online shoppers Ocado online shoppers (per cent) (per cent) (per cent) 21.77 17.34 15.50 11.44 10.70 8.49 8.12 4.06 2.21 0.37 16.67 19.84 14.68 7.14 11.51 11.90 10.32 3.97 3.97 0.00 28.24 22.75 11.76 4.71 14.90 9.41 2.35 4.31 1.57 0.00

Attributes Ordering time Quality Delivery time Substitutes Del. reliability Discount Delivery cost Special offer section Help line number New product section

Table IV. The preference structures given by the weighting of each attribute for online grocery shoppers across the three selected online supermarkets

Product shares of preference Shares Table V. Market simulation of the time dimension Scenario 1 Scenario 2 Scenario 3 13.41 34.40 52.19 Std. err. 3.71 3.43 3.69

Three scenarios were tested: (1) In scenario 1, it takes 20 minutes to place an order and there are no items below quality. However, goods are delivered within 48 hours and the delivery van can arrive 1 hour earlier or later. The other attributes are set as being realistic (i.e. 5 delivery cost, no discount, a help line number, a special offer section, a new product section and ve items have been substituted). These six factors will remain constant across the three scenarios to allow the testing of the rst four most important attributes. (2) In scenario 2, goods are now delivered in 24 hours instead of 48 hours and the delivery van is on time. (3) In scenario 3, the delivery van is still on time but it now only takes 6 hours to receive the order at home. By having the goods delivered on time within 6 hours, the share of preference increased by nearly 39 per cent compared with having them delivered early or late within 48 hours. By having the goods delivered on time but within 24 hours instead of 6 hours, the share of preference increased by 21 per cent. Consequently, if an online retailer improves its delivery time (from 24 to 6 hours) other things being equal, its consumers share of preference could increase by up to 18 per cent. This highlights the fact that delivery time and delivery time reliability need to be included in online marketing strategy as their impact on consumers share of preference is signicant and market share will otherwise be lost. Wilson-Jeanselme (2001) looked at maximising customer acquisition and retention by minimising and, even better, eliminating leakages on determinant attributes, since any leakage is an open door for competition to gain market share. In the case of Tesco, Table V suggests that Tescos shoppers are, like Ocados shoppers, sensitive to the time it takes to place an order. However, the weighting is less than for Ocados (22 vs 28 per cent). Time to place an order is therefore a determinant attribute for which Tesco needs to provide the best service, i.e. perform better than its competitors. Two further attributes: quality and delivery time are close to each other in terms of their level of importance. These in turn are followed by another two attributes whose percentage is above 10 per cent: substitutes and delivery time reliability. The question is to nd out which ones of these four attributes will, like ordering time determine the choice of consumers and which ones are important but not determinant in the purchase making decision. In this case, across all scenarios we decided to set-up the attribute ordering time with the level 20 minutes to place an order because we know that it is the level with the highest order of importance. It was therefore necessary to avoid any uctuations for this attribute as it will hide the effect of the others on consumers share of preference. In this case, four scenarios were tested: (1) scenario 1 tested quality by having no items below quality expectations while the delivery time substitutes and delivery time reliability will have a poor performance; (2) scenario 2 tested delivery time;

Shoppers expectations of online retailing 535

IJRDM 34,7

(3) scenario 3 tested substitutes; and (4) scenario 4 tested delivery time reliability. The results are shown in Table VI. Having no items below quality expectations in the shopping basket is by far the most preferred attribute after placing an order online in 20 minutes, with the least preferred being substitutes. The two attributes that will determine whether or not consumers will remain with Tesco are ordering time and quality. Similarly, to acquire consumers from Tesco, retailers need to focus internal resources on insuring product quality and the efciency/ease of use of their website. Personalisation could also induce efciency when placing an order. So far, we have employed a web-based technique to understand online consumers preferences as opposed to the retailers perceptions of their expectations. This was achieved by determining the relative importance of levels of attributes pre-selected by conducting in-depth interviews. The results not only could improve customer retention rates but also help in acquiring online shoppers from other retailers supermarket websites. Understanding ofine shoppers preferences Finally, we can also compare these preferences with the expressed preferences of those shoppers who have never used the internet to buy groceries online, in order to understand the strategies required to acquire shoppers. As for the previous part of the study, logit analysis was rst used to analyse the choice results of ofine shoppers, but in this case we are interested in their perceptions and expectations of online grocery buying, rather than actual experience. A utility was estimated for each level of each attribute. Table VII summarises the utility for each level of each attribute for the off-line grocery shoppers group. We then determine the weighting of each level of each attribute, with the objective of determining the relative importance of each attribute and quantifying it. Each attributes absolute utility level difference is expressed, as it was done for online shoppers, as a proportion of the sum of the total of the ten attributes absolute differences. This is shown in Tables VIII and IX. Table IX clearly shows that the two main determinant attributes for ofine shoppers to shop online are seen to be ordering time followed by quality. The spacing between ordering time and the next most important attribute of delivery time is 9.6 per cent. It seems therefore safe for a company to strategically position itself against these two determinant attributes to acquire consumers. When comparing online and ofine shoppers between Sainsburys, Ocado and Tesco for the attributes ordering time and quality the result shows that these two attributes are perceived as determinant for both groups. However, the weighting of
Scenarios Scenario Scenario Scenario Scenario 1 2 3 4 Shares 34.08 28.02 15.85 22.05 Std.err. 4.52 4.49 3.14 4.03

536

Table VI. Shares of preference

Attribute level Has a special offer section No special offer section Has a new product section No new product section Help line number No help line number Delivered in 6 hours Delivered in 24 hours Delivered in 48 hours No delivery cost 5 delivery cost Delivery at agreed time Delivery an hour early or late No items below quality Five items below quality 20 minutes to place an order 35 minutes to place an order 1 hour to place an order No substitutes Five substitutes 0 per cent discount 10 per cent discount

Utility 0.10 2 0.10 0.00 2 0.00 0.01 2 0.01 0.40 0.00 2 0.40 0.29 2 0.29 0.22 2 0.22 0.58 2 0.58 0.52 0.15 2 0.67 0.30 2 0.30 2 0.24 0.24

Shoppers expectations of online retailing 537

Table VII. Utilities for ofine grocery shoppers

Attribute Special offer section New product section Help line number Delivery time Delivery cost Delivery reliability Quality Ordering time Substitutes Discount Total

Highest utility level 0.1 0 0.01 0.4 0.29 0.22 0.58 0.67 0.3 0.24

Lowest utility level 2 0.1 0 2 0.01 2 0.4 2 0.29 2 0.22 2 0.58 2 0.67 2 0.3 2 0.24

Difference 0.2 0 0.02 0.8 0.58 0.44 1.16 1.34 0.6 0.48 5.62

Per cent proportion 3.56 0.00 0.36 14.23 10.32 7.83 20.64 23.84 10.68 8.54 100.00

Table VIII. Calculation of the weighting of each attributes for ofine grocery shoppers

these two attributes is higher for ofine shoppers. This could be due to a fear of spending too much time placing an order whilst not being familiar with the website, and a fear of not getting a quality as good as that which might be personally selected by the respondent. Two attributes of online shopping in particular have a different relative importance in the mind of ofine shoppers as opposed to online shoppers. The rst relates to the relative importance of any discount. This attribute has more importance for Sainsburys online shoppers (11.90 per cent) and for Ocados online shoppers (9.41 per cent) than it has for ofine shoppers (8.54 per cent). The second is delivery time reliability; less important for ofine as against online shoppers. It is in fact

IJRDM 34,7

538

14.9 per cent for Ocados shoppers. This might be due to the fact that Ocados shoppers do not spend much time at home and therefore an early or late delivery is of no use (This was stated by a number of Ocados shoppers during the in-depth qualitative interviews conducted to determine the levels). Ocado has also made much of its shorter delivery time slots and this may have been factor for those selecting Ocado. Finally, ofine shoppers may over time become more time sensitive, so that delivery time reliability may also gain in weighting once a poorly timed delivery has been experienced. This suggests that a customer acquisition strategy will need to differ from a retention strategy as consumers needs and fears seem to be different. In summary three sets of differences can be observed. The rst is that ofine shoppers are mainly concerned with the length of ordering time (reecting lack of experience), while online shoppers are also concerned with the reliability of the delivery and not just the time to place the order. Secondly, ofine shoppers seem to be generally more sensitive to expectations of quality than online shoppers (although Ocado online shoppers are even more sensitive to quality issues than ofine shoppers). It might therefore be important to ensure an especially high level of quality for rst time shoppers, since this may reect a concern that what they will be delivered is not what they would have selected themselves in store. Finally, ofine shoppers are concerned by the price component, i.e. both the delivery cost and the potential for a 10 per cent discount. Lower price seems to be an incentive to attract ofine shoppers online, whilst once they have shopped online the main concerns become the contents of the shopping basket and time. Conclusions and implications for the marketing planning process This paper has sought to explore expressed preference differences between shoppers patronising three UK online grocery retailers (Tesco.com, Sainsburys.co.uk and Ocado.co.uk) and ofine grocery shoppers. We conclude that, even if ordering time and quality come rst for online and ofine shoppers, the weighting of these factors is different between these two segments of shoppers as well as amongst online shoppers from different retailers. The weighting of ordering time and quality is higher for ofine shoppers than online shoppers (except for Ocados shoppers). Consequently, by advertising and ensuring that the purchase process will be quick and easy for a rst time buyer, companies could increase customer acquisition. This customer acquisition

Attributes Ordering time Quality Delivery time Substitutes Delivery cost Discount Delivery time reliability Special offer section Help line number New product section

Off-line shoppers (per cent) 23.84 20.64 14.23 10.68 10.32 8.54 7.83 3.56 0.36 0

Table IX. The preference structure given by the weighting of each attribute for the ofine grocery shoppers group

rate could also be improved by ensuring that best quality items to consumers are provided. Nevertheless, subsequent attributes exhibit a greater difference between these two groups. For example, delivery time reliability remains a concern for online shoppers while ofine shoppers are still focussing on the content of their shopping basket, with the importance of substitutes. This paper recommends a segmentation of consumers based on understanding their expressed preferences as opposed to more traditional segmentation methods. It also stresses the importance of looking at the interaction effects of different combinations of attributes rather than focussing on a single one. Market simulations have been used to conduct what-if scenarios that could also be used to investigate issues such as retailers website design, positioning and pricing strategy. By better understanding the customers expectations, the gap between what the customer actually wants and what the company thinks the customer wants will be better aligned. In addition other perceptual gaps will be closed if the internal logistics, communication and other resources focus on these preferences. Consequently, there will be fewer customer acquisition opportunities for competitors to exploit. The ndings of this research also demonstrate the power of the choice-based conjoint method in providing the objective raw marketing data so necessary for the building of realistic and effective marketing plans. For example, strategic marketing questions such as the potential changes in market share arising from different internet conguration choices can be quantied. How to increase the probability of converting non-internet shoppers into internet shoppers can also be identied. And price sensitivity in the internet channel can be explored quantitatively. From these research ndings the strategic level of the marketing plan can be assembled. For example, informed choices on how to position the internet channel in relation to the traditional store channel can be made as well as key factors that will affect turnover, namely price and type of channel offering. Uniquely market segmentation on preference structures allows the marketing plan to differentiate its approach in relation to different attractive sectors. Choice based conjoint data will allow the marketing plan to specify not only the top line nancial numbers but will also translate that data into a language which is operationally useful. For example, the research could benchmark the performance required on the attributes that really matter in the customers buy/do not buy decision. As a result, the marketing plan would properly provide strategic and operational direction for the rest of the rm. Our study, for example, demonstrated clear implications for supply chain management (time based attributes) and purchasing (quality and availability based attributes). Properly conducted choice based conjoint research should lower an online grocery retailers risk by signicantly improving the richness, clarity and overall quality of the marketing plan.
Note 1. The useable response rate could have been higher by reducing the number of choice tasks. The choice tasks represented 14 screens and the majority of respondents stopped at ten screens. These responses were considered to be unusable. About 12 per cent of returned questionnaires were rejected for this reason. The gross response rate was in the order of 25 per cent. Using a priori estimates of standard errors for attribute levels with 258 respondents

Shoppers expectations of online retailing 539

IJRDM 34,7

and total choice tasks of 3,096, shows that the efciency of any attributes level ranges from 0.9991 to 0.9998. This test of design efciency shows that the realised useable response rate poses limited scope for design threats to validity. Hair et al. (1998) suggest that design validity problems only begin to emerge at a response level of less than 130 for designs with average numbers of choice tasks, attributes and levels. References Baker, W.L., Lin, E., Marn, M.V. and Zawada, C.C. (2001), Getting prices right on the web, The McKinsey Quarterly, Vol. 2 No. 2, Special edition, pp. 54-63. Brynjolfsson, E. and Smith, M.D. (2000), Frictionless commerce? A comparison of internet and conventional retailers, Management Science, Vol. 46 No. 4, pp. 563-85. Burt, S. and Sparks, L. (2003), E-commerce and the retail process: a review, Journal of Retailing and Consumer Services, Vol. 10 No. 5, p. 275. Evans, P.B. and Wurster, T.S. (1997), Strategy and the new economics of information, Harvard Business Review, September-October, pp. 71-82. Ghose, S. and Dou, W. (1998), Interactive functions and their impacts on the appeal of internet presence sites, Journal of Advertising Research, Vol. 38 No. 2, p. 29. Grewal, D., Iyer, G.R. and Levy, M. (2004), Internet retailing: enablers, limiters and market consequences, Journal of Business Research, Vol. 57, pp. 703-13. Grewal, D., Levy, M. and Marshall, G.W. (2002), Personal selling in retail settings: How does the internet and related technologies enable and limit successful selling?, Journal of Marketing Management, Vol. 18, pp. 301-16. Hair, J.F., Tatham, R., Anderson, R. and Black, W. (1998), Multivariate Data Analysis, Prentice-Hall, Englewood Cliffs, NJ. Hoffman, L.D. and Novak, T.P. (1996), Marketing in hypermedia computer-mediated environments: conceptual foundations, Journal of Marketing, Vol. 60, pp. 50-68. Interactive Media in Retail Group (2006), E-christmas: online retail results, press release, 20 January, available at: www.imrg.org Keen, C., Wetzels, M., De Ruyter, K. and Feinberg, R. (2004), E-tailers versus retailers: which factors determine consumer preferences, Journal of Business Research, Vol. 57 No. 7, pp. 685-95, 11p. Reynolds, J. (2003), Prospects for electronic commerce, in Reynolds, J. and Cuthbertson, C. (Eds), Retail Strategy: The View from the Bridge, Butterworth-Heinemann, Oxford. Sinha, I. (2000), Cost transparency: the nets real threat to prices and brands, Harvard Business Review, Vol. 78 No. 2, pp. 43-8. Wilson-Jeanselme, M. (2001), Grocery retailing on the internet: the leaky bucket theory, The European Retail Digest, No. 30, pp. 9-12. Wilson-Jeanselme, M. and Reynolds, J. (2005), Competing for the online grocery customer: the UK experience, in Kornum, N. and Bjerre, M. (Eds), Grocery E-Commerce. Consumer Behaviour and Business Strategies, Edward Elgar, Abingdon. Further reading Hill, T. (2000), Manufacturing Strategy: Texts and Cases, Palgrave Macmillan, New York, NY. To purchase reprints of this article please e-mail: reprints@emeraldinsight.com Or visit our web site for further details: www.emeraldinsight.com/reprints

540

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.