Professional Documents
Culture Documents
Electronic Shopping: Gerald L. Lohse
Electronic Shopping: Gerald L. Lohse
Electronic
Shopping
Designing online stores with effective customer interfaces
has a critical influence on traffic and sales.
T
he global electronic market will have want an item in an online retail store if it is not sell-
a profound impact on commerce in ing. Nor should they conclude that a poor response
the 21st Century. While current U.S. to a given store design is due to the merchandising
sales in cyberspace ($1 billion in mix. It is important to look at the relationship
1995) are small in comparison to between sales and user interface design. Limited
total U.S. retail sales ($1.7 trillion in menus, poorly designed navigation, and the diffi-
1995) [3], U.S. cybersales projections for the year culty in comparing multiple products on the same
2000 range from $7 to $117 billion [9]. More screen all have adverse effects on electronic shopping
importantly, most experts predict a radical shift in [2]. Can customers find what they want in the stores?
how business will be conducted in the next century. Are customers aware of what products are available?
This shift not only has businesses scrambling to meet After all, diligence in browsing a store is not a virtue
this new marketing reality, but also raises many retailers should expect from its online customers.
important research questions about business strategy, We review online retail store attributes such as the
technical infrastructure, government policies, the number of links into the store, image sizes, number
electronic market demographics as well as how peo- of products, and store navigation features. By review-
ple will use the technology. ing the user interface features in the context of elec-
Issues of technology usage become critical as busi- tronic shopping, we hope to facilitate the process of
nesses and retailers attempt to exploit the boom in designing and evaluating alternative storefronts by
electronic marketing. There are large differences identifying key features that impact traffic and sales.
between a physical store and its electronic counter- While this article specifically addresses user inter-
part. A help button on the home page of the Web face design issues for electronic shopping, user inter-
shopping site replaces the sales clerk’s friendly advice face design is an important component of all software
and service. The familiar layout of the physical store development [10] with important implications for
becomes a maze of pull-down menus, product productivity [5]. Unfortunately, Web information
indices, and search features. Now more than ever, the systems (WIS) designers sometimes do not heed the
promise of electronic commerce and online shopping importance of the message advocated by user inter-
will depend to a great extent upon the interface and face design pundits such as Jakob Nielsen [11] and
how people interact with the computer. Vincent Flanders (www.webpagesthatsuck.com).
Account managers, production staff, and mer- Many of the lessons learned from online stores also
chant partners should not assume customers do not apply to WIS applications.
[
behavior and suggest cost of delivering text and
ways to improve retail images via the Web.
Web sites. Spiller and
Lohse surveyed 35 attrib-
Moving text has an While a picture is worth a
thousand words, small pictures
utes of 137 Internet retail overpowering effect are often poor quality and fuzzy.
stores to provide a classi- on the human Large pictures take too much
fication of the strategies time to download. Less than 8%
pursued in Web-based
peripheral vision of the total screen area contained
marketing [12]. Many of making it difficult to images, which is much lower
the descriptive statistics process information than a typical paper catalog.
of retail store features in
the Spiller and Lohse elsewhere on Service. Staff responsiveness to
study resonate with the the page. the customer service link is crit-
consumer reactions noted ical. The customer wants care-
by Jarvenpaa and Todd. ful, continuous, useful
Lohse and Spiller [8] communication, across geo-
measured 32 interface features for 28 online retail graphic barriers (worldwide shipping, multilingual
stores in August 1996 and identified store design sites), 24 hours a day, 365 days per year. Service
features that influence online store traffic and sales. includes sales clerk service for merchandise selection,
Based on these studies, we describe attributes that answers to frequently asked questions (FAQs) and
influence store traffic and sales for the following six credit, return, and payment policies. Most retail sites
categories: provide a pittance of service information. Almost one-
third did not provide any information on the compa-
Merchandise. Consumers infer information about ny’s history, policies, or background; 80% had less
quantity, quality, and variety of products from the than 10 lines of information [12]. This is a surprising
brand names or reputation of the physical store. Unfor- number since customers want to know who they are
tunately, not all products available in the merchant’s dealing with and to whom they are sending credit
catalog or real store are available online [4]. This leads card information. This is especially important for new
to disappointment as noted by this customer: “…the virtual companies solely operating on the Internet.
last thing I expected to see from LL Bean was to see Consumers want help with product selection
only one model of Swiss Army knife.” Customers pre- (such as size, color, fabrics), gift services, contact
fer a large product selection. However, only 5% of the information for sales representatives, a FAQ section
Internet stores had more than 500 products and 62% for speedy answers, information about the security of
had less than 50 products [12]. The number of prod- their transactions, company return, payment and
ucts in a store explains 17% of all variance in store traf- credit policies, information about shipping and han-
fic but had no effect on sales [8]. This implies big stores dling costs, guarantees, and statements about prod-
are less effective than small stores at converting traffic uct quality. In the survey by Spiller and Lohse, 95%
into sales perhaps because consumers are not finding of the stores did not have hyperlinks among related
products, only 25% had help for product-size selec- Promotion. Additional store visits and sales are
tion, less than 9% of the sites had an FAQ section, generated by promotion. Each hour of promotion on
and 47% did not offer interactive email. Eighty per- the electronic shopping mall entrance screen
cent of the participants in the Jarvenpaa and Todd explained 4% of the variance in sales and 1.4% of the
study had at least one negative comment about variance in store traffic [8]. Promotions involve sales,
Internet customer service [4]. Stores that offer a FAQ advertising, and appetizer features that attract cus-
section had more visits than those without such a tomers. While it is easy to spot the discount signs in
section [8]. Stores with a feedback section for cus- the local department store, comments such as “you
tomers to obtain help had increased sales. never get a sale,” “shopping is more fun in the real
Users are unwilling to tolerate delays associated store,” and “you never get to see head-to-head com-
with delivering audio, animation, graphics, and petition that you see in the mall” suggest that online
video. Some browsers do not handle Java, sound, ani- stores have a long way to go [4]. Spiller and Lohse
mations, and the like. The flood of consumer email noted that only 6% of the sites offered a ‘What’s
can create a customer relations disaster when compa- new’ section and 76% did not offer incentives or
nies do not respond quickly. To reduce the barrage of appetizers to attract and retain customers. Appetiz-
email, many companies use a FAQ section. One com- ers not only for getting customers to return but also
pany burdened consumers with separate forms for increase the enjoyment of shopping online.
each question posed to the service department. Store promotions include frequent buyer schemes,
[
are analogous to store window mative headlines, make judicious
displays and aisle products use of white space, and use multi-
with links to an individual ple levels of headings, highlight-
product. Figure 1 shows an Links must ing, and color text as visual cues.
end product page with navi- be active.
gation buttons. The browse Nothing will Checkout. The checkout process is
forward and browse back but- more complicated than necessary
tons allow customers to navi- drive away [4]. If the checkout process is too
gate from product to product. customers like long, customers balk and sales are
Without such buttons, the lost. For example, the checkout
consumer cannot look at mer-
a site full of process is different for every store.
chandise adjacent to this pro- dead links. This confuses the customer. Uni-
motion item nor access versally adopted standards would
information about the compa- alleviate this concern. Also, con-
ny’s reputation, returns poli- sumers enter a lot of repetitive
cies, and so forth. information such as name, address,
Web ads often use animation, scrolling text, and and credit card information. Ideally, this could be
other special effects to capture your attention. As entered once and allow the customer to checkout
noted by Nielsen [11], moving text has an overpow- once even though purchases are from multiple stores.
ering effect on the human peripheral vision making Thus, Web designers can use the interface to
it difficult to process information elsewhere on the improve the existing process.
page. The position of an ad on the page or a product The most common checkout method uses the
in a list has a large effect on consumer choice, even shopping cart metaphor [12]. At a real department
though the position says nothing informative about store, it is easy to undo a purchase during checkout.
the business [7]. Serial position matters because peo- Just tell the clerk that you have changed your mind
ple scan ads on a page sequentially and their scan- and only want the pants, but not the shirt. On the
ning is not exhaustive. As a result, people never read Internet, the undo button for one store emptied the
certain ads or product listings. entire shopping cart leaving the customer to start
over. Inconsistent menus do not allow customers to
Convenience. Store layout, organization features, as review the contents of the shopping cart from any
well as ease of use, all fall under the convenience page in the store. Often there is no access to the
umbrella. Comments from consumers such as “this is shopping cart unless a purchase is made.
not for computer illiterate people,” and “I had places Some order forms do not provide customers with
I wanted to go but couldn’t understand how” illus- important information such as when the order will
trate these convenience issues [4]. be shipped or whether an item is out-of-stock. For
General help functions might assist users in error example, to make returns or cancel an order, an order
recovery or find a particular function in the docu- number must often be provided. An electronic mail
The information superhigh- Unfortunately, electronic most systems offered today are
way in its incarnation as the shopping systems have had rel- proprietary structures that
Internet has the inherent atively limited success to date lack interoperability and cross
potential of enabling a global in supplanting or even aug- navigation. Despite the grow-
marketplace. Electronic prod- menting traditional physical ing number of companies that
uct catalogs (EPCs) form the and catalog shopping. Besides present their products on the
Web-based front end to this problems with basic usability Internet, a global search for
rapidly evolving marketplace. as Lohse and Spiller discuss, products and comparative
analysis of their features is
Figure 1. A federation of intelligent and mediating electronic product catalogs impeded by semantic differ-
ences between the EPCs. Thus,
even though buyers enjoy
WWW WWW broad access to different ven-
Client Client
Query
dors’ product specifications,
integration and evaluation of
WWW MEPC Query MEPC WWW product information still has
Server Server
to be performed manually. The
Differentiation Services
impediments to an effective
and efficient use of the Inter-
WWW net as a global marketplace at
Server
MEPC
the stage of product identifi-
Integration Services cation can be summarized as
DB
follows:
MEPC
text. Buyers have to trans- for the specification of prod- of an arbitrary number of cat-
late those semantics manu- uct information is needed. alog modules structured in a
ally. Especially in a global Based on such a language a recursive manner. The basic
environment as diverse as Mediating Electronic Product building blocks are intelligent
the Internet, where people Catalog (MEPC) [1, 2] can be EPCs, which are catalogs that
from many different cul- constructed. Under the term are augmented by a semantic
tural backgrounds meet and MEPC we understand a catalog layer (see Figure 2). Such a
a wide variety of languages that semantically integrates layer is necessary in order to
is in use, the issue of trans- several individual EPCs or allow for the classification
lation between and integra- other MEPCs into a federated and evaluation of the contents
tion of different system (see Figure 1), and of catalogs queried and to
vocabularies is exacerbated provides the following facilitate the semantic inte-
even further. The search functionality: gration of heterogeneous cat-
engines available on the alogs. Besides Q-Calculus [3]
Internet today are of little • Transparent search over (used in our implementation)
help as they are practically several EPCs in a distributed other methods and languages
all based on classic models environment for the formal representation
of information retrieval that • Support for multilingual of product information have
work on a strictly syntactic query resolution and been proposed by Kimbrough
basis. national differences (conver- [4] and Keller [5], who has
• Vendors tend to employ the sion of measurements, sizes, applied KQML to the modeling
user interface design of or currencies, for example) of EPCs.
their EPCs as a means of dif- • Inference mechanisms for Drawing from the informa-
ferentiation. Thus, cus- the evaluation of integrated tion repositories of MEPCs, dif-
tomers are continuously information ferentiation services can be
forced to get accustomed to • Interfaces to additional mar- established. Such services
new ways of interaction. ket services (such as pay- focus on niche markets by
ment or logistics services) extracting specialized areas of
In order to overcome these product information from cat-
problems a common language A MEPC federation is composed alog intermediaries, subse-