TAXONOMY OF APPAREL WEBSITES

ASSOCIATED PROFESSOR HRISTO KATRANDJIEV, PH. D.1

IVO VELINOV, PH. D., STUDENT2

Abstract:.
The paper presents a typology of apparel websites. This typology includes
three types of websites which are identified by applying the method of cluster
analysis. The profiles of the three major types of apparel websites are
described. The authors outline the guidelines for future research and explain
that the present research is the first step of a larger project.

Key words: cluster analysis, apparel websites, visual merchandising, typlogy

1. INTRODUCTION

During the last decade we observe a huge increase of internet trade, especially in
the sector of clothes and shoes. The internet space bristles with apparel online stores and
their number rise constantly. The science and practice of visual merchandising develop fast
nowadays and move toward a new dimension – that of the digital world. The visual
merchandising principles as well as the elements of visual merchandising change rapidly in
order to response to the changes. These elements need to be studied and analyzed. An
important part of the analysis of visual merchandising elements of apparel (fashion goods)
in digital environment is to classify the websites on the basis of these elements.
The objective of the present paper is to classify the apparel websites (online shops
for fashion goods) on the basis of visual merchandising elements. We applied the method
of cluster analysis and discovered three major types of apparel websites from the point view
of their visual merchandising elements. We have analyzed hundreds of apparel websites
and have determined the newly emerged elements of apparel presentation in digital
environment or the so called visual merchandising elements. These elements concern the
visual presentation of fashion goods within the online stores. We have identified tents of
visual merchandising elements and used them for the sake of forming a typology of apparel
websites.

1
UNWE - Sofia, Bulgaria, katrandjiev@unwe.acad.bg
2
New Bulgarian University – Sofia, Bulgaria, i.velinov@nbu.bg

1
 2. DEFINITION OF CLASSIFICATION DIMENSIONS

The list of classification dimensions are presented in Table 1. This list consists of
17 dimensions (variables). All of the variables are measured on the dichotomous scale (1 –
yes; 0- no). If we point our attention to the first dimension which (“Site map”) we might
say that a specific website could propose a site map (which is coded by 1) or could not
(which is coded by 0). The same coding is done for the rest of the variables. The next 5
dimensions in the classification list are grouped together because they concern the
searching options. Some apparel websites offer a searching option while others do not offer
such an option. The different variants of searching options are included in the list – by
brand, by item, by target, by price, and by style.

Table 1 List of classification dimensions

Classification dimensions Scale
1 Site map Dichotomous 0,808 0
2.1 By brand Dichotomous 0,802 0
2.2 By item Dichotomous 0,849 0
Searching Dichotomous
2 options
2.3 By target 0,853 0
2.4 By style Dichotomous 0,849 0
2.5 By price Dichotomous 0,884 0
3.1 Horizontal Dichotomous 0,866 0
3.2 Horizontal place/ appear Dichotomous 0,866 0
Presentation Dichotomous
3 style
3.3 Vertical 0,825 0
3.4 Vertical place/appear Dichotomous 0,825 0
3.5 Intro-page Dichotomous 0,932 0
4.1 Intro-music Dichotomous 0,308 0
Atmospheric 4.2 Music during browsing Dichotomous 0,952 0
4 features 4.3 Video Dichotomous 0,952 0
4.4 Simple click-on Dichotomous 0,429 0
Sales/ 5.1 Automatically-moving advertisments Dichotomous 0,264 0,041
5 Promotions
Dichotomous
signages 5.2 Non-clickable advertisments 0,922 0

The third group of dimensions includes decisions concerning the presentation style
accepted within a specific website. An apparel website can be organized by a vertical or a
horizontal order. The main presentation styles within this 3rd group of dimensions are the
following: horizontal, horizontal place/appear, vertical, vertical place/appear and
presence/absence of an intro-page. The fourth group of variables cover the so called
atmospheric features – intro-music, music while browsing the website, vide, and simple
click on. The last set of dimensions presents the Sales/Promotions signages. This set
includes two options - Automatically-moving advertisments and Non-clickable
advertisements.

2
 It is important to point out that we used only a part of all dimensions in order to
run the procedure of cluster analysis. The rest of the dimensions (the so called external
dimensions) were used later for the sake of assessing the external validity of the defined
clusters. This approach has applied in other classification studies and has proved its validity
and appropriateness (Katrandjiev, 2011).

3. DEFINITION OF CLUSTERS’ NUMBER

Figure 1: Dendogram of 3 cluster solution In the process of cluster

analysis we applied the method of
hierarchical clustering. Within the
hierarchical methods we chose the
method of Ward. The reasons we
preferred this approach are the
following: first, the hierarchical
approach graphs the clustering
process in clear and easy-to-interpret
manner; second, the visual
presentation of the clustering process
eases the process of defining
clusters’ number. The dendogram of
the clustering process is shown at
Figure 1. As can be seen we chose a
3 cluster solution (the clusters are
surrounded by ellipses).
The decision for the final
Cluster 1 Cluster 2 Cluster 3 number of clusters was taken on the
basis of some analysis and
consideration.
First of all we applied a widespread approach for defining the number of the clusters – the
“scree test”. According to Aldenderfer and Blashfield the practical application of this
method means to “graph the number of clusters implied by a hierarchical tree against the
fusion or amalgamation coefficient, which is the numerical value at which various cases
merge to form a cluster” (Aldendefer,
Figure 2: Defining the number of clusters Blashfield, 1984, p.66). The scree
diagram (called also the elbow
diagram) is presented at Figure 2.
According the rule of the “scree test”
one must identity the point at which
the amalgamation levels (Y-axis)
drop suddenly and line becomes flat
and not so steep. This point shows on
the X-axis the number of clusters.

3
As shown on the diagram of agglomeration levels (Figure 2) the zone at which the
flattering if the curve starts is corresponding to 3 or 4 cluster solution. The final decision of
defining the clusters’ number was taken after a thorough analysis of the two options.

4. THE CLUSTERS‘ PROFILES

As stated earlier by the help of cluster analysis we identified three types of apparel
websites. The websites were profiled on the basis of their characteristics (or elements)
listed in Table 1.
From the "presence" or "absence" of these elements we analyzed the clusters’ profiles.
Applying this method we identified three major types of apparel websites: (1) simple, (2)
practical and (3) convenient online stores for fashion goods.
The first type is characterized by its simplicity of implementation. It contains some
basic elements such as sitemap (without an option for searching products). The positioning
of the menu buttons is mainly vertical. The products are presented by a human model in 2-
D version and there is a colour sample as well.
The second type of online stores is distinguished by its richer performance. It has a
search criterion in the online stores - search. It has to look as option product and style of the
garment. Site menus are vertical positioning. Advertising here is presented by static
advertising brochures. Monochrome background is again presented in the cluster. When
choosing the font color white was chosen. Products are offered in this kind of online stores
and usually are represented by 2-D version accompanied by front and rear views of the
product. Magnifying certain parts of the apparel is an extra option here.
Offering these products at the site also includes a model that provides an option for
offering additional suggestions for each item (the so called cross-selling). Registration is
mandatory.
The third type of clusters is “the richest” among all. There is a sitemap, search
engine submission style menu on the site in a horizontal position, introductory page in the
online store. Atmospheric features are characterized by music on demand and video. The
Background color and text color is white. Also this type of cluster also has other
predominant colors - black, red, etc. The type view of the product is 2-D with three options
- front, rear and sides. And there is also a color sample. Featured products are represented
by a model, and there are proposals related to the desired product called cross selling.
Registration shopping here is mandatory.
To be more clear and summarized the status of the various clusters can be produced
visually as shown on Fig. 3. As can be seen, all clusters of online stores can be
characterized on the basis of the following characteristics:
 Sitemap
 Search engine
 Presentation styles
 Atmospheric features

4
  Promotion signage
 Background color
 Text color
 Type of product view
 Product view presentation methods
 Product display methods
 Swatch
 Mix and Match
 and Registration
In order to enable easier and more compact visualization of the three types of online
stores for fashion goods, the authors of this paper combine the elements into three main
groups: Online path finding assistance (sitemap, search engine, presentation styles);
Environment (Atmospheric features, promotion signage, color, background color, color
surrounding product, text color); and manner of product presentation (Type of product
view, product view presentation methods, product display methods, swatch, mix and match
and registration).
Our method of cluster profiling is based on the idea that all clusters possess all these
characteristics. The differences among clusters are sought from the point of view of the
“degree” of the attributes (characteristics). For example some apparel websites include a
rich product presentation (which is market by “+” at Fig. 3). Other websites are relatively
simple or poor from the point of view of the mane of product presentation (which is marked
by “-“at Fig. 3). The same logic of cluster profiling is applied for the rest of the attributes.

Figure 3 Classification of apparel websites (online stores)

Simple Practical Convenient

5
 5. CONCLUSIONS AND FUTURE RESEARCH

The present study attempts to provide a clearer guidance for online retailers of
fashion goods in Bulgaria. The results of this study could provide useful information by
giving a detailed list of visual merchandising’s elements of apparel websites. This will help
internet traders to adjust their online shops depending on regions and countries of residence
of the target group of the style of clothing. This is especially important for companies
offering products in several countries worldwide.
Our research is just the first step of a bigger project. Here we revealed the
typology of apparel websites. The next step will be to conduct a survey and gather data
about the specific requirements of Bulgarian online shoppers. The research design will
include the application the method of conjoint analysis. This methodological approach is
suitable for defining the optimal combination of visual merchandising elements from the
point of view of Bulgarian consumers. The results of this stage of our research project will
be published in a separate article.

References:

1. Aldendefer, M.S., Blashfield, R.K., Cluster Analysis, London, Sage Publications,

1984.
2. Катранджиев, Хр., Сегментиране на телевизионната аудитория на основа на
зрителските навици, УНСС, докторска дисертация, 2005.