0% found this document useful (0 votes)
6 views23 pages

5 TH IBC2011 Product Info Display

Uploaded by

bholu23sept2000
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
6 views23 pages

5 TH IBC2011 Product Info Display

Uploaded by

bholu23sept2000
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/281550972

CUSTOMER PRODUCT KNOWLEDGE AND INFORMATION DISPLAY


PREFERENCES André P. Calitz Sherwin Barlow

Conference Paper · September 2011

CITATION READS

1 3,202

2 authors, including:

André P. Calitz
Nelson Mandela University
268 PUBLICATIONS 1,529 CITATIONS

SEE PROFILE

All content following this page was uploaded by André P. Calitz on 07 September 2015.

The user has requested enhancement of the downloaded file.


CUSTOMER PRODUCT KNOWLEDGE AND INFORMATION DISPLAY
PREFERENCES
André P. Calitz Sherwin Barlow
Nelson Mandela Metropolitan University Nelson Mandela Metropolitan University
P O Box 77000 PO Box 77000
Port Elizabeth, 6031 Port Elizabeth, 6031
Tel No: +27 41 504 2639 Tel No: +27 41 504 2247
Fax No: +27 41 504 2831 Fax No: +27 41 504 2831
André.Calitz@nmmu.ac.za Sherwin.Barlow@nmmu.ac.za

ABSTRACT
The product information displays or presentation modes used on e-commerce websites are
factors which influence a customer’s buying decision and need to adapt depending on the
customer’s knowledge about the specific products considered. The provision of the most
appropriate product information display for a specific customer is beneficial, however,
difficult to determine. Customer profiling in the e-commerce domain has provided several
benefits for conducting business online, in creating a more personalised user interface for
customers, improving customer satisfaction and presenting the required level of product
information.

In this study the authors designed and implemented an e-commerce website selling three
different product ranges, namely groceries, wine and electrical products. The customer’s
knowledge about each product range was determined utilising a tested questionnaire. A
customer’s knowledge relating to each product category was classified as novice, intermediate
or expert. The products in each product range used three different product information
displays namely list, matrix and commented list in order to display product information to
customers based on their product knowledge.

Internet users were required to purchase products from the e-commerce website. The product
information display (list, matrix or commented list) varied depending on the product
knowledge classification (novice, intermediate or expert) of the customer. The product
information displays presented were evaluated utilising on-line and written questionnaires,
purchasing activities on the e-commerce website and eye-tracking evaluations. The research
identified that a statistically significant relationship exists between a novice customers’
product knowledge level and the preference for the matrix product information display, but no
statistical significant relationship was determined between the other product knowledge levels
and product information displays.
Keywords: Product information display, product knowledge, e-commerce.
1. INTRODUCTION
The Internet has changed the way people conduct business activities and businesses on the
Internet continuously seek new methods to effectively communicate with customers
(Schneider, 2007). E-commerce has provided a way to build personalised relationships with
customers using a customer profile. A customer profile is information about a customer’s
demographics, sales history, characteristics and activities (Eirinaki and Vazirgiannis, 2003).
Customer profiling has become an important tool in business today. Businesses use customer
profiling to create a business customer database that can be utilised for market segmentation
in order to identify new target markets and potential customers.

A new development in customer profiling is Customer Relationship Management (CRM).


CRM is used by businesses to create personalised relationships between the business and the
customer to improve customer loyalty (Liu, Lin, Chen and Huang, 2001). Customer loyalty is
a competitive advantage for businesses (Adomavicius and Tuzhilin, 2005). CRM is made
possible by creating a customer profile for each customer interacting with the website.

E-commerce has made Internet based advertising that includes personalised recommendations
and customisation possible (Schafer, Konstan and Reidl, 2001). Businesses face many
challenges regarding the level of personalisation to provide as these relationships cannot be
too personal (Schafer et al., 2001). Research has shown that e-commerce is more competitive
than any other conventional commerce because of the low entry barriers (Chang, Changchein
and Huang, 2006).

The way in which advertising, product information presentation and screen layout are
implemented on an e-commerce system is an important factor leading to the website’s success
level and affects the buyer’s buying decision (Sharp, Rodgers and Preece, 2007). The
advertising technique could affect a visitor by changing a visit into a purchase increasing e-
commerce website success (Schafer et al., 2001).The way in which information is presented
can also affect customer buying behaviour to some extent therefore product information
displays should be used efficiently (Hong, Thong and Tam, 2004).

The product information displays are important factors’ influencing a customer’s buying
decision (Hofgesang, 2007). Product information presentation modes mostly contain
information such as the product name, description and price. These presentation modes
change based on the category of products being displayed (Bettman and Kakkar, 1977). The
product information display modes used to display product information about a certain
product might influence the customers buying decision (Hong et al., 2004).

In this study, an e-commerce website was implemented selling three product categories,
namely Grocery products, Wine and Electrical products. At the time of the research, wine,
groceries and entertainment products were some of the items that were most commonly
purchased on-line in South Africa (Goldstuck, 2007). The project utilised the customer
information and browsing behaviour patterns to maintain a customer profile. The information
captured is the customer’s personal information and the customer’s product knowledge levels
for each product category. The product information display mode then adapts the computer
user interface based on the customer’s product knowledge level for the product category.

The product categories used three different product information displays layouts (presentation
modes), namely list, matrix and commented list in order to display product information to
customers based on their product knowledge levels. Product information interfaces mostly
used on the Internet are the list and matrix information displays (Sharp et al., 2007) where the
image based presentation display is preferred above only textual displays.

The aim of the study was to implement a customer profiling system consisting of an e-
commerce website that could be utilised to present product information using different
product information displays and allowing customers to purchase products using a preferred
presentation display. The limitations to the study include the use of a convenience sample,
namely NMMU employees and a limited number of expert users per product category.

The three product information displays, namely list, matrix and commented list were used to
display product information to customers based on their product knowledge levels. The
research objective of this paper is to determine whether a relationship exists between product
information displays (presentation mode) and a customer’s product knowledge level.

The research objectives and methodology are presented in Section 2. The growth of e-
commerce, the implementation of the on-line store, the customer profiling module and the
information display of products are presented in Section 3. Section 4 discusses the research
study and results. Section 5 concludes and contextualises the findings and future research.
2. RESEARCH OBJECTIVES AND RESEARCH METHODOLOGY
The research questions addressed in this study were:
 Does a relationship exist between a customer’s product knowledge level per product
category and the product information display layout the customer prefers when
browsing products on a website?
 Do customers prefer the personalised product information displays the website
provides when viewing product information?

The primary objective of this research was to implement an e-commerce website that allows
customers to purchase products using preferred product information displays. The different
product information interfaces (presentation modes) used on the website contained different
levels of information about products, used different screen layouts which were linked to the
customer’s product category knowledge level. The presentation mode was adapted for the
customer based on his/her product knowledge level per product category. For example, a
novice customer would receive the detailed matrix layout which contains a high level of
product information.

Three different product information displays, namely list, matrix and commented list were
implemented per product category to display product information to customers based on their
product knowledge levels. Customers purchased products on-line and the system adapted the
product information presentation mode based on the customer’s product category knowledge
and the customer’s product information display layout preferences.

The specific research objectives of the study were to:

 Determine the product knowledge level (novice, intermediate or expert) per product
category (groceries, wine and electrical products) for a customer; and

 Determine if a relationship exists between a customer’s product knowledge level for a


specific product category (groceries, wine and electrical goods) and the product
information display layout (list, matrix and commented list) the customer prefers when
purchasing products.

The hypothesis investigated in the study was as follows:

 H0: No relationship exists between a customer’s product knowledge levels and product
information displays.
 H1: A relationship exists between a customer’s product knowledge levels and product
information displays.

The research methodology utilised in the study followed the traditional usability evaluations
approach. A CSIS on-line store was created implementing an e-commerce website selling
consumer products in the three categories, namely groceries, wine and electrical goods.
Customers purchased products on-line and the system presented different product information
displays (screens). Data gathering techniques included web-logging, pre- and post-
questionnaires, an on-line product knowledge survey and eye-tracking. The data were
statistically analysed and the results are presented in Section 4.

3. LITERATURE STUDY

Customer profiling

E-commerce has enabled on-line businesses to extend their customer base by obtaining access
to global markets. E-commerce is buying and selling products, services and information using
Internet technologies and electronic systems (Schneider, 2007). Businesses use different
techniques to increase customer loyalty and turn website visitors into customers (Kim et al.,
2006; Adomavicius and Tuzhilin, 2005). Customer profiling is a technique mostly used in
online businesses to gain a customer’s trust and loyalty. The main uses of customer profiling
are customisation, recommendation, personalisation and marketing.

The customer profile contains information obtained from the customer. This information
includes the customer’s personal data or the customer behaviour data when interacting with
the e-commerce website (Schafer et al., 2001). A customer profile can be divided into factual
information which is facts about the customer for example age, gender, name, geographical
and behavioural information where this information is captured when a customer performs
certain tasks on-line (Adomavius and Tuzhilin, 2005). The use of questionnaires is one
explicit method most commonly used by an on-line business to obtain customer information,
together with other techniques such as the use of cookies and web-log mining.

Customer profiles have been implemented in various e-commerce systems but mostly to
identify customer preferences and needs, for example, www.amazon.com (Amazon.com,
2010). A customer profile is established by implementing implicit and explicit feedback by
on-line businesses (Jokela, Turpeinen, Kurki, Savia and Sulonen, 2001). Explicit feedback is
simply determining factual information asking customers to register their details on the
website by using online questionnaires. Examples of on-line businesses that use explicit
feedback to determine a customer’s profile are Amazon.com and Kalahari.net (Amazon.com,
2010; Kalahari.net, 2010). Figure 1 is a screen shot of explicit methods to obtain profile
information from customers on Amazon.com.

Figure 1: Explicit feedback customer profiling method (Amazon.com, 2010).

Implicit feedback is the web based application captures information about the customer as he
or she interacts with the system. Web-log mining and cookies are techniques used by on-line
businesses to prevent users from entering their information every time they enter the website.
Web-log mining is the extraction of information from web-log files created and stored on the
web server. Cookies are web-log files created at the server side but kept on the client
machine (Nelte and Saul, 2000).

Table 1: Needed customer profile components.


Field Name Amazon.com Edgars Pick n Pay
Name Yes Yes Yes
Surname No Yes Yes
Title No Yes Yes
Gender No Yes Yes
DOB Yes Yes Yes
ID number No Yes Yes
Contact details Yes Yes Yes
Log on details Yes Yes Yes
Address details No Yes Yes
Table 1 indicates the common customer profile information required by online businesses
when opening a purchasing account. The table does not contain the customer product
knowledge levels as part of the customer profile which this project will use.

In this study, the following information for a customer is recorded:


 Customer Details: Records the customer details that include biographical details,
contact details and login information including a unique customer ID;
 Log Details: The log details of the customer’s activities during the interaction session
with the system. The log details will be used to update the customer profile;
 Customer purchase: Keeps information about items customers have purchased in the
past;
 Sale: The recording of information relating to a single product purchase, the user
interface settings, the product information and selected display mode. The information
is used to determine the product information display mode most frequently used; and
 Product: Product information and product categories.

Product information display interfaces


A customer needing to purchase a product has many external and internal factors which
impact on the customer’s acquisition decision (Hofgesang, 2007). The customer’s information
processing in memory plays a large role in the decision to acquire a product or not (Biehal and
Chakravaarti, 1983). Task structure differences then affect processing at acquisition.
Customer information processing differs in the way which product information is organised.
Product information can be represented by brand or by product attribute.

The product catalogues of e-commerce on-line stores should include more detailed
information about products (Callahan and Koenemann, 2000). Callahan and Koenemann
(2000) implemented a user interface called Info Zoom, a catalogue display that provides
customers with a wide range of products where the customer then filters and compares
products based on selected attributes. Early research has shown that customer choice is
influenced by processing attributes relating to the way in which information is displayed and
will be the key to future interface designs (Bettman and Kakkar, 1977). The reason is that
customers consume information as it is displayed to them (Callahan and Koenemann, 2000).

The way in which e-commerce websites display product information affects the buyer’s
buying decision (Sharp et al., 2007). Businesses use customer profile information for many
reasons, specifically for personalisation and recommendations. The customer’s profile can be
used to display products in a product information display layout the customer prefers. The
customer profile information can be used to acquire or predict the customer’s preferences and
a technique increasingly being utilised is the customisation of the user interface for a specific
customer.

E-commerce websites currently do not determine which presentation modes (product


information displays) to use for customers viewing different product categories based on the
customer’s product knowledge. Most e-commerce websites implement a single product
information display mode with a link to a more detailed page containing information about
the product being viewed. The product information displays (presentation modes) used on e-
commerce websites are factors which influence customers’ buying decisions and overall
satisfaction (Hong et al., 2004).

Research conducted by Hong et al. (2004) has shown that the product information display
mode has an effect on the customer’s browsing behaviour. The most common product
information interfaces used by e-commerce websites are the list, the extended list and the
matrix layouts. The image based presentation mode includes images on the website and this
mode is preferred over a text based presentation mode where no images are included.
Generally image based presentation modes outperform text based presentation modes (Hong
et al., 2004). Images and/or text are included in all product information interfaces.

The list product information display mode has been promoted to provide a more efficient
browsing behaviour due to the easier recall of product information. The list product
information display modes allow exploration of products and groups all products of a certain
category together. This allows users to browse a category of products more easily, which
positively affecting the customers browsing behaviour (Hong et al., 2004).

The list information format is mostly used for browsing tasks where the matrix format is used
for searching tasks (Hong et al., 2004). The matrix format is used when a customer
purchasing products requires a large amount of information about the product being
displayed. The detailed list information display mode consists of detailed descriptions of
products and allows customers to add and view detailed information about a product.

In summary, the three most commonly used product information displays (Hong et al., 2004)
used in e-commerce and implemented in this study are:
 List layout (Figure 2 left) - contains a list of products with little detail about a product
and mostly used by e-commerce website because it is easy for browsing;
 Matrix layout (Figure 2 middle) - contains products which are displayed in a grid
where this layout contains a fair amount of product information and mostly used to
display technologic products and for searching; and
 Commented list layout (Figure 2 right) - this is a layout which contains a list of
products with a brief of the product details and mostly used in e-commerce for
customers to add comments to products.

Figure 2: List layout, Matrix layout and Commented List layout implemented.

Figure 3 is an example of a matrix information display used by the Pick n Pay on-line
shopping store for grocery products.

Figure 3: Displaying grocery products in a matrix layout (Pick n Pay, 2010).

CSIS online store

The CSIS on-line store was re-developed based on the design by Ntawanga et al. (2009). The
three product categories implemented were groceries, wine and electrical goods. Electrical
goods were included as a number of users are generally unfamiliar with this product category
and might therefore have required additional product information before making a purchasing
decision. A number of changes were made on the user interface of the system to improve the
quality and appearance (Figure 4).

Figure 4: CSIS online store.

The user interface (Figure 5) for the creation of the initial customer profile was developed by
Ntawanga et al. (2009) and utilised in this study. Each question has five options that are
grouped per product category. Customer profiles that include static and dynamic information
were established in accordance to research conducted by Ntawanga et al. (2009). The
customer completed a questionnaire (Figure 5) and an initial profile consisting of all
demographic details and product knowledge levels for each of the product categories was
created. The product knowledge of a customer for each product category, namely groceries,
wine and electrical products were updated when the customer shopped on-line.

The customer profile is updated based on the dynamic information recorded; based on web-
site usage data, product information utilised and display mode per product category selected.
The customer’s profile is updated as the customer interacts with the website. An update to the
customer’s product knowledge levels occurs when the customer purchases a certain amount of
products from a certain category or when a customer selects a display mode explicitly. The
customer is then allowed to add products to a basket in one of three product information
interface layouts and check-out. The customer can also update generic information
accordingly.
Figure 5: Questionnaire for customer’s initial product knowledge profile.

The questionnaire (Figure 5) was utilised to establish the level of product category knowledge
contained specific questions on grocery products, electrical products and wine. Five questions
for each product category were utilised and each question had to be scored (1=Not at all to
5=Always). The system rates the customer according to the total score for each product
category. The system then uses the rating to determine the product information layout to
display. Table 2 summarises the score and rating process (Ntawanga et al., 2009).

Table 2: Initial customer product knowledge level rating.

Score Range Customer Rating Product Information Level


5-10 Novice Matrix
11-19 Intermediate Commented list
20-25 Advanced List

The content display used the three different product information levels depending on the
customer’s product category rating (Table 2). The three different information levels provide
different display modes and display product information in three different layouts. The list
layout information mode provides the least detail about a product where the matrix layouts
provide more detail and allows more products to be viewed in a single row, thus allowing the
customers to search for products information more effectively. The commented list layout
provides the most detail and a function to view additional information about products. Figure
6 shows a matrix product information display for a customer who has novice product
knowledge for groceries.

Figure 6: CSIS online store matrix product display for a grocery page.

4. THE RESEARCH STUDY AND RESULTS


The user evaluation for this project consisted of a pilot study to determine whether users use
different product information displays, followed by a main study to determine the relationship
between product information displays and customer product knowledge levels. The pilot
study consisted of a convenience sample (n=6) with staff and master students of the
Computing Sciences Department of NMMU. Respondents were provided with a task list and
a questionnaire. The evaluation consisted of a usability study which was used to obtain
feedback on the systems usability and the usefulness of the different product information
displays.

The user evaluation (n=31) conducted in the main study evaluated the goal of the study,
namely to determine whether a relationship exists between a customer’s product knowledge
level for a product category and the product information display customers prefer when
viewing the products. The users completed specified tasks and a number of metrics were
measured at various stages in evaluations. The evaluation procedure consisted of customers
completing an informed consent form, a demographical questionnaire and a pre and post task
questionnaire.
Customers completed the registration process on-line in order to determine the user’s initial
product knowledge levels based on the answers provided to on-line questionnaire (Figure 5).
A task list was provided which contained instructions to purchase specific products from
different product categories.

In order to compare the product information displays users preferred, the CSIS on-line store
was configured as two systems. The difference between the two systems was that system A
provided the user with a default product information display interface as follows:
 List layout – for the grocery product category;
 Detailed Matrix layout – for the electrical product category; and
 Commented List layout- for the wine product category.

These default interfaces were chosen randomly to ensure no relationship exists between the
interface and the customer’s product knowledge level for the product category. System B
provided the users with a more personalised product information display based on the
customer product knowledge level for a specific product category, as follows:
 Novice - Matrix layout for the selected product category;
 Intermediate - Commented List layout for the selected product category; and
 Expert - List layout for the selected product category.

The research assumption made was that if the user is presented with system A, which
provides a default product information display interface, the user would change the interface
to a personal preferred layout and when the user was presented with system B, the user would
not change the interface.

One half of the users were given System A first, followed by System B and the other half of
the group System B first, followed by System A. The task list contained 18 tasks where the
first 9 tasks were to purchase products on the first system presented to a participant. The
products the participant purchased and the layout preferred were captured by the system. In
the post questionnaire, participants were asked to rate the system according to various criteria
using a 5-point Likert scale, which allowed quantitative data to be tabulated and visualised
using different graphs.

Eye tracking evaluations


The results obtained from eye tracking provide evidence to support that the participants were
focused on the detail in this layout preferably reading the information. The Gaze plots in
Figure 7 show that the participants used the layout and the information provided. The layout
mostly preferred by participants when they viewed grocery products was the list layout and
the matrix layout for electrical and wine products. This showed that the product information
was usable and shows the participants used the layouts to support the results in the empirical
study. Figure 7 (left) shows a heat map of participants using the commented list layout and
Figure 7 (middle) the gaze plot of participants using the detailed matrix layout. The gaze plot
of participants using list layout are presented in Figure 7 (right).

Figure 7: Heat map (left), Gaze plot (middle) and a gaze plot (right).

Empirical evaluations
In the main study the gender of the participants (n=31) were male (n=18) and female (n=13).
All the participants were older than 19 years of age (reflecting a typical minimum age for
users of an online shopping website) where 23% were older than 36 years. There were five
different ethnicities/home languages and six different nationalities. The users had more than
six years computer experience, 70% has more than six years Internet experience and 90% of
the participants spend more than an hour on the Internet per day. This sample size (n=31)
allowed that statistics be calculated such as t-tests and Chi-squared analysis.

The participants (n=31) consisted mostly of participants with a novice product knowledge
level for wine (n=23), intermediates for grocery (n=19) and electrical (n=12). A few
participants were experts for the product categories (Figure 8). The users indicated that other
products bought online mostly are books, music, games, scientific papers, articles and flights.
Figure 8: Main study user product knowledge (n=31).

A demographic questionnaire was presented to each participant which consisted of five


sections A to E. Section A was basic demographical variables, Section B was based on
computer experience, Section C focused on on-line shopping experience, Section D included
the task lists for System A and B and Section E was the post questionnaire measuring the
systems metrics as well as the system usability and the participants presentation layout
preferences.

The post questionnaire which measured the various criteria used a 5 point Likert scale. This
scale was represented as: 1 - Strongly disagree and 5 - Strongly agree. The metrics which
were captured in this evaluation were chosen to meet the study goal, which was to determine
whether there is a relationship between the customer product knowledge levels and the
product information display layouts they prefer. The metrics captured were system
effectiveness, performance metrics and task success. The evaluation also contained a
comparison of the two systems to determine which one was mostly preferred.

Task list results


The results tabulated are task success and task time for the activities for system A and B.

Table 3: Task success information (n=31)


Task success
System Observed count Expected count Chi-squared p-value
A 229 237,5 0.44
B 246 237,5
The results tabulated in Table 3 were calculated using the Chi-Squared test where the value
obtained (t30 = 0.44, p < 0.05) using a 95% confidence interval shows that there was not a
significant difference between the task successes for the two Systems. In Table 4 the task
times were measured as interval data in seconds. The numbers 1-5 recorded as 1 is < 30 sec, 2
is 31-60 sec, 3 is 61-90 sec, 4 is 90-120 sec and 5 is > 120 sec. The average time range overall
for System A was 2,63 seconds and 1,61 seconds for System B.

The t-test on the task times (t30 = 14.66, p < 0.01) indicated that the task times were
significantly different. It is clear that there was a slight decrease in task time between the
systems, this could be due to the learnability of the system or that the users just performed
better on System B overall. The average task time for system A’s task 2 was the highest
overall where this task was to purchase an electrical product. This can be due to the users not
changing the layout to the appropriate one in time.

Table 4: Task times for System A and B (n=31)


Task Mean A Mean B Median A Median B Std.Dev A Std.Dev B
1 2.81 1.97 3.00 2.00 0.91 0.80
2 2.90 1.77 3.00 2.00 0.79 0.80
3 2.71 1.42 3.00 1.00 0.78 0.62
4 2.71 1.42 3.00 1.00 0.69 0.67
5 2.61 1.65 3.00 1.00 0.92 0.75
6 2.68 1.58 3.00 1.00 0.75 0.67
7 2.26 1.42 2.00 1.00 0.73 0.62
8 2.45 1.87 2.00 2.00 0.77 0.72
9 2.81 1.45 3.00 1.00 0.98 0.72

Post-questionnaire results
The questionnaires were based on the evaluation of the systems metrics. In Table 4 it was
identified that participants were satisfied with the system and the product information displays
they were presented with. The direct comparison of System A and B illustrates that
participants preferred using System B which provided a product information display
according to their product knowledge level. This illustrated that participants like a
presentation which is personalised.
Table 5: System metrics for System A and B (n=31)

Question (1- Strongly disagrees and 5- strongly agree) Mean Median Std.Dev

1. The system speed was slow/fast? 4.68 5.00 0.48


2. I had enough information available to decide my
purchase? 4.45 5.00 0.81
3. The layout of the product information matched my
product knowledge level in System A? 3.42 3.00 1.03
4. The layout of the product information matched my
product knowledge level in System B? 4.13 4.00 0.76
5. I prefer System A which had default displays or System B
which had a display according to my product knowledge
level? 4.45 5.00 0.85
6. I enjoyed using System A or System B because it was
easier to find the product I had to purchase? 4.26 4.00 0.77

The first question in Table 5 obtained a high mean indicating that users found the system fast
and question 2’s mean illustrates that there was enough information available to make
purchase decisions. Question 3 presented a high standard deviation, proving that users’
opinions regarding whether layout matched their knowledge levels varied widely. Median
values indicated that participants felt that the layouts of product information in System B
provided a closer match to their preferences than the layouts provided by System A.

The Likert scales for Question 5 and 6 had system A as a 1 rating and system B at rating 5.
The means for question 5 and 6 were statistically significantly different from the intermediate
value of 3 (Table 4), illustrating that those participants preferred using System B. This could
be due to the system providing the product information displays according to the customer
product knowledge level. This illustrates that participants preferred the system to provided
personalised product information displays.

In the final evaluation it has been identified that participants mostly preferred the list layout
when viewing grocery products. This could be due to the fact that most participants have an
expert or intermediate product knowledge level for groceries. Comments received from
participants stating that when purchasing groceries they know what they need to purchase and
do not require detailed product information.
Participant interface interaction results
Figure 8: Preferred layout for product category (n=31).

The above figure represents the participant’s product knowledge levels and layouts they
purchased products in a particular category. The first graph in Figure 8 (left) shows that of the
19% participants who were classified as novices for groceries and of this group over 70% of
the sales for groceries were conducted using the matrix layout. Figure 8 (left) also indicates
that the participants who were classified as intermediates and experts for groceries, also
mostly preferred the matrix layout for their grocery purchases.

Figure 8 (middle) shows that all the users classified as novices for electrical goods (32%) used
the matrix layout for all electrical product purchases. The majority of the participants who
were classified as intermediates and experts for electrical products also preferred the matrix
layout. However, of the 39% of participants who were classified as intermediates for electrical
products, a quarter of this group completed their purchases using the commented list layout.

Figure 8 (right) indicates that 74% of the participants (n=31) in this study were classified as
novices for the wine product category. 60% of the novice group completed the wine purchases
using the matrix layout. The participants with an intermediate product knowledge level for
wine generally preferred the commented list layout. It is clear from the graphs above that
experts also had a greater percentage of their sales completed using the matrix layout all the
time.

The results from the Chi-squared test with a confidence interval of 95% are presented in
Tables 6 to 8. The tables below show a statistically significant difference between the layout
preferences for each class of participants as indicated by the p-values for the Chi-square tests
being less than 0.05 for all classes of participants. Participants’ preferred layout was not
equally split between the three layouts.
The results presented in Tables 6 to 8 show that participants preferred the matrix layout. The
user interaction results section show that participants with any product knowledge level
preferred using the matrix layout when purchasing a product from any category except for
wine. The wine category participants with intermediate product knowledge levels used the
commented list layout for wine. This can be due to the participants with an intermediate
product knowledge level having a clear idea of the wine they needed to be purchase.

The Chi-squared tests used further showed that a significant difference existed between layout
choices and participants preferred the matrix layout overall. It can be deduced that there was a
relationship between participants with a novice product knowledge level and the matrix
layout. No statistically significant difference has been identified between the other product
knowledge levels and product information displays where all participants with the other
product knowledge levels have shown a preference for the matrix layout. The interpretation of
the results concludes that the null hypothesis H0 cannot be accepted.

Table 6: Chi-squared test for grocery product category (n=31)


Layout Novice (n=6) Intermediate (n=19) Expert (n=6)
Observed Expected Observed Expected Observed Expected
List 33 33 35 35 17 33
List(com) 17 33 24 35 33 33
Matrix 50 33 47 35 50 33
p-value 0.000259287 0.022692648 0.000259287

Table 7: Chi-squared test for electrical product category (n=31)


Layout Novice (n=10) Intermediate (n=12) Expert (n=9)
Observed Expected Observed Expected Observed Expected
List 10 33 11 35 33 33
List(com) 0 33 16 35 0 33
Matrix 90 33 78 35 67 33
p-value 9.42046E-33 5.18895E-18 1.68814E-15

Table 8: Chi-squared test for wine product category (n=31)


Layout Novice (n=23) Intermediate (n=6) Expert (n=2)
Observed Expected Observed Expected Observed Expected
List 2 37 17 36 0 33
List(com) 17 37 8 36 0 33
Matrix 87 37 83 36 100 33
p-value 6.18041E-25 5.87645E-21 1.34796E-44
In some cases, however, there is evidence suggesting a link between product knowledge and
preferred layout. For example, a relationship clearly existed between participants with a
novice product knowledge level and the matrix layout. Furthermore, there were indications
that participants preferred System B, which provided personalised levels of product
information. Comments participants made were however captured as these comments affect
the goal of this evaluation, where participants stated that:
 “Even if I am an Expert I would still like to see details especially for electrical products.”
 “I know which grocery products I want to purchase before hand and nutritional
information is irrelevant to me.”
Therefore this supports the reason for the non existence of relationships between the other
product knowledge levels and product information displays.

5. CONCLUSIONS AND FUTURE RESEARCH


E-commerce and customer profiling have become an integral part of modern businesses
today. The use of customer profiling in e-commerce has brought about personalisation which
improved customer satisfaction levels and loyalty (Adomavicius and Tuzhilin, 2005). The
goals of this project were to successfully implement an e-commerce system including a
customer profile, recording the customer’s product knowledge levels and purchasing
behaviour and display settings. The CSIS on-line store presented three different product
information displays depending on a customer’s product knowledge level.

The interfaces were evaluated and the user testing consisted of two tests, the pilot study to
determine whether the product information displays were useful and a main empirical
evaluation to determine whether a relationship exists between a specific customer product
knowledge level and a specific product information display.

The results obtained from the empirical evaluation were that a statistical significant
relationship exists between a novice product knowledge level and the matrix product
information display mode presenting a large amount of detail about a product. There is no
statistical significant relationship between other product knowledge levels interfaces and
specific information display modes. No relationship exists between the intermediate and
expert product knowledge levels and their related product information displays. The users in
this study did however indicate that they would still want to see more detailed product
information for the product categories for which they are classified as intermediates or
experts.

Participants in the study found the product information displays useful and agreed with the
way in which the product knowledge levels were updated. The participants were highly
satisfied with the personalised product information displays and preferred the system
providing personalised product information displays. The researchers further concluded that
the reason there is no relationship between certain product knowledge levels and product
information displays could be due to customers knowing which products they wanted to
purchase. The user sample did also not include a large number of expert participants in
specific product categories, for example electrical goods and therefore the results of this
research cannot be generalised.

Future research will test these findings on other systems and further investigate guidelines for
systems to accommodate users with different product knowledge. Management need to take
the research findings into consideration when developing new e-commerce applications or
evaluating existing websites and specifically noting Internet customer’s product knowledge
levels and product information display preferences. The use of product information displays
should include enough information where this affects a customer’s buying decision and can
lead to increases in sales and profitability.

REFERENCES
Adomavicius, G. and Tuzhilin., A., 2005. Personalisation Technologies: A Process-
Oriented Perspective. Communications of the ACM , 48 (10), pp. 83-90. October 2005.
Amazon.com, 2010. Amazon web services. [Online] Available at:
www.amazon.com/gp/browse.html?node=3435361. [Accessed 20 February 2010].
Bettman, J. R. and Kakkar, P., 1977. Effects of Information Presentation Format on
Consumer Information Acquisition Strategies, Journal of Consumer Research, 3, pp. 233-
240.
Biehal, G. and Dipankar C., 1983. Information Accessibility as a Moderator of Consumer
Choice. Journal of Consumer Research, 10 (1), pp. 1-14.
Callaham, E. and Koenemann, J., 2000. A Comparative Usability Evaluation of User
Interfaces for Online Product Catelogues. ACM EC'00, 17-20 October 2000, Minnesota,
USA.
Chang, S. E., Changchein, S.W. and Huang, R., 2006. Assessing users’ product specific
knowledge for personalisation in e-commerce. Expert Systems with Applications 30, pp.
682-693.
Eirinaki, M. and Vazirgiannis M., 2003. Web mining for web Personalisation. ACM
transactions on Internet Technology 3 (1), pp. 1-27. February 2003.
Goldstuck, A., 2007. Dreaming of an e-Christmas. [Online] Available at:
http://www.worldwideworx.com/archives/59#more-59. [Accessed 30 July 2008].
Hofgesang, P.I., 2007. Web personalization through incremental individual profiling and
support based user segmentation. In Proceedings of the IEEE/WIC/ACM international
conference on web intelligence. USA, 2007. Silicon Valley.
Hong, W., Thong, J.Y. and Tam, K.Y., 2004. Designing product listing pages on e-commerce
websites: and examination of presentation mode and information format. International
Journal of HCS, 1 (1), pp. 26-28.
Jokela, S., Turpein, M., Kurki, T., Savia, E. and Sulonen, R., 2001. The role of structured
content in a personalised news service. Proceedings of the 34th annual Hawaii
international conference on systems science (HICSS-34). Washington DC.IEEE Computer
Society: 7 January 2001.
Kalahari.net, 2010. [Online] Available at: http://www.kalahari.net. [Accessed 28 March
2010].
Kim, J. W., Lee, K. M., Shaw, M. J., Chang, H., Nelson, M. and Easley, R. F., 2006.
A preference scoring technique for personalised advertisements on Internet
storefronts. Mathematical and Computer Modelling, 44 (1-2), pp. 3-15. July 2006.
Liu, D., Lin, Y., Chen, C. and Huang, Y., 2001. Deployment of personalised
e-catalogues. An agent-based framework integrated with XML metadata and
user models. Journal of Network and Computer Applications, 24, pp. 201-228.
Nelte , M and Saul, E., 2000. Cookies. Weaving the web into a state. The ACM Student
Magazine. 7 (1), pp. 10-13.
Ntawanga, F., Calitz A.P. and Barnard L., 2009. A Model for Dynamic Product Information
Display as an Internet Service, Proceedings of the 3rd International Business Conference
(IBC 2009), Zanzibar, 15-17 September 2009.
Schafer, A.I., Konstan, J. and Reidl, J., 2001. E-commerce recommendation applications.
Data Mining and Knowledge Discovery, 5 (1-2), pp. 115-113.
Schneider, G.P., 2007. Electronic Commerce. 7th Annual edition. Boston: Thomson.
Sharp, H., Rogers, Y. and Preece, J., 2007. Interaction Design, beyond human-computer
interaction. 2nd Ed, John Wiley & Sons.

View publication stats

You might also like