Int. J. Internet Marketing and Advertising, Vol. 2, Nos. 1/2, 2005 143
Copyright © 2005 Inderscience Enterprises Ltd.
Personalisation of online avatars: is the messenger
as important as the message?
Natalie T. Wood*
Department of Marketing, The Haub School of Business,
St. Joseph’s University,
5600 City Avenue, Philadelphia, PA 19131
Fax: 610 660 3239 E-mail: firstname.lastname@example.org
Michael R. Solomon
Department of Consumer Affairs, College of Human Sciences,
308 Spidle Hall, Auburn University, Auburn, AL 36849
Fax: 334 821 1364 E-mail: email@example.com
Basil G. Englis
Campbell School of Business, Berry College,
Mount Berry, GA 30149
Fax: 706 238 7854 E-mail: firstname.lastname@example.org
Abstract: In traditional media, the selection of an appropriate source for
communication (such as a celebrity) is left in the hands of the advertiser who
generally adopts a standardised spokesperson for an entire target market.
Online, the opportunity now exists for this spokesperson to be personalised to
suit individual consumers, through the use of avatars. This exploratory study
investigates the potential functions and manifestations of avatars as well as
consumer preference for avatars as a direct source of communication. Using a
non-probability referral sample of college educated females (aged 18–30) from
Southern California preliminary findings revealed that empowering female
consumers to select their own source of communication for online apparel
shopping may offer a more effective may of developing persuasive messages.
Keywords: avatars; personalisation; communication source; spokesperson;
e-commerce; internet advertising; persuasive communication.
Reference to this paper should be made as follows: Wood, N.T.,
Solomon, M.R. and Englis, B.G. (2005) ‘Personalisation of online avatars: is
the messenger as important as the message?’, Int. J. Internet Marketing and
Advertising, Vol. 2, Nos. 1/2, pp.143–161.
Biographical notes: Natalie T. Wood is an Assistant Professor of Marketing at
St Joseph’s University, Philadelphia, Pennsylvania. Professor Wood received
her PhD from Auburn University, Alabama. Her dissertation research involved
examining the emerging use of icons and avatars as virtual consumer guides
and was the recipient of the American Academy of Advertising Doctoral
Dissertation Award. Professor Wood’s specialisation is consumer behaviour,
visual persuasion, product/service personalisation and customisation and
144 N.T. Wood, M.R. Solomon and B.G. Englis
themed environments. She has extensive international experience as a
marketing educator and practitioner with over a decade of experience
consulting to a variety of industries including retail, consumer products,
telecommunications and hospitality.
Michael R. Solomon is the Human Sciences Professor of Consumer Behavior
in the Department of Consumer Affairs, College of Human Sciences, at Auburn
University. His primary research interests include consumer behaviour and
lifestyle issues, branding strategy, the symbolic aspects of products, the
psychology of fashion, decoration, and image, services marketing and the
development of visually oriented online research methodologies. He currently
sits on the editorial boards of the Journal of Consumer Behaviour and the
Journal of Retailing, and he serves on the Board of Governors of the Academy
of Marketing Science. Professor Solomon’s textbooks, Consumer Behavior:
Buying, Having, and Being, and Marketing: Real People, Real Choices both
published by Prentice Hall, are widely used in universities throughout North
America, Europe, and Australasia. In 2003 his trade book, Conquering
Consumerspace: Marketing Strategies for a Branded World, was published by
AMACOM (American Management Association).
Basil G. Englis is the Richard Edgerton Professor of Business Administration
and Chair of the Marketing Department in the Campbell School of Business at
Berry College in Rome, Georgia. Professor Englis has experience working in
marketing research and strategic planning for such firms as E.I. DuPont de
Nemours, VISA/Moskowitz-Jacobs, Inc., Vanity Fair-Wrangler Division, and
Levi Strauss. His research interests include consumer behaviour, lifestyles and
values, mass media and consumer socialisation
The surging popularity of e-commerce forces marketers and advertisers to revisit many
existing theoretical frameworks that originated in the offline world to determine if or how
they apply in a virtual one. Until recently, the assumption has largely been that what
works offline will also work online. For instance, much has been written on the role of
the source as a vehicle for persuasion in traditional print and broadcast media, but little
work has addressed source effects for online communications.
In traditional media environments, the selection of an appropriate source (such as a
model or celebrity) is left in the hands of the advertiser who generally adopts a
standardised spokesperson for an entire target market. In the virtual world, the
opportunity now exists for this spokesperson to be personalised to suit individual
consumers. That is, there is more latitude to customise the message source by creating
one or more avatars who can convey the message.
The original meaning of the word ‘avatar’ is the manifestation of a Hindu deity in
superhuman or animal form. In the computing world, it has come to mean, “your
cyberspace presence represented by a three-dimensional character that you can move
around inside a visual, graphical world” (South China Morning Post, 2000). Thus, an
avatar is an online communication source, a sales associate or customer service
representative who assists internet users in the information search process, providing
product information and if requested, advice about selecting from a set of product
Personalisation of online avatars 145
The purpose of this paper is to investigate the use of avatars as online communication
sources. We begin by reviewing literature on the growth of internet advertising,
persuasive communication, and source selection both in offline and online environments.
We then provide some preliminary data on female consumers’ selection of avatars for
online apparel shopping and highlight the implications for marketers and advertisers.
2 E-commerce and online advertising
By 30th April 2004, more than 757 million people around the world had access to the
internet from home (Internet World Stats, 2004). Nielsen NetRating’s global internet
index revealed that during March 2004, the average internet user logged on 31 times,
spending 26 hours and 11 minutes online. Users visited 61 unique domains and screened
a total of 1,036 web pages, viewing on average 34 pages per surfing session but spending
only 16 seconds on each page (Netratings Inc., 2004a). The limited time spent on each
web page poses a challenge for advertisers to develop communication strategies that both
capture and maintain the attention of internet users.
To reach online consumers, internet advertisers use a variety of formats, including
sponsorship (web page or website) pop-up ads, interstitials, superstitials, and the most
common of all, banner ads. In 2002, the average internet user was exposed to
approximately 700 banner ads daily and this is predicted to increase to 950 by 2005
(Shimp, 2003). Despite their popularity, the inability of banner ads to capture attention
and their subsequently poor click-through rates have paved the way for more obtrusive
advertising in the form of pop-ups – ads that pop-up out of nowhere on a separate screen,
interstitials – ads that appear between two web pages, and superstitals – short animated
ads that appear over the top of a web page.
The constant bombardment and intrusive nature of these advertisements is often very
frustrating to users. They may view these advertisements as an annoying interruption, an
impediment to goal-oriented online activities. As more and more consumers turn to the
internet either to search for product information or to perform transactions online,
advertisers must revamp their efforts to harness the capabilities of the internet to their
advantage while overcoming the negative image of online advertising. The key to
successful advertising begins with an understanding of the online communication process
and how it can be employed to make advertising more persuasive.
3 Persuasive communications
Persuasive marketing communications play a fundamental role in helping consumers
make their decisions in both offline and online environments. The model adopted
(and adapted) by advertisers in traditional, offline settings emphasises a one-to-many
mass communication process that is linear, static and inflexible (Schram, 1955).
Advertising messages are conceived as one-way communications aimed at (hopefully)
the correct audience. In this model, interaction between the consumer and the company is
virtually non-existent. Interaction (if it occurs at all) only occurs at the feedback stage
once the message has been received and processed.
Unlike traditional communications media, the internet is an interactive medium that
presents the potential for consumers to interact in a variety of ways with the visual
146 N.T. Wood, M.R. Solomon and B.G. Englis
images to which they are exposed (Englis and Solomon, 2000). This interaction allows
the consumer greater flexibility, and in essence, greater control over what she views
(Rogers and Albritton, 1995). Unlike traditional media where the consumer plays a
passive role in the communication process, in an online environment her role is active
and at times even proactive. Online advertising that is perceived as an interruption or
obstruction to online activity can be removed with the click of a mouse. Up until recently,
efforts to improve the success rate of online advertising have largely focused on the
message and its format (banners, pop-ups, etc) with very little, if any, attention at all to
the source that delivers it.
3.1 The communication source in traditional environments
This neglect is ironic, since one of the most important components of the communication
model is the source. The source may be an individual or an organisation. In traditional
broadcast and print media, the selection of an appropriate source is central to the
communication process, but the optimal choice is a complex issue. Advertisers face
the challenge of selecting a source that is not only credible and attractive, but also
someone with whom the target audience can identify.
Credibility is “the extent to which the recipient sees the source as having relevant
knowledge, skill, or expertise and trusts the source to give unbiased, objective
information” (Belch and Belch, p.110). To convey expertise, advertisers will often select
a spokesperson because of his or her knowledge and experience with a particular product.
For example, Tiger Woods would be considered a credible spokesperson for golf shoes.
While it has been found that conformity with spokesperson recommendations varies both
with perceived level of source expertise and the strength of the advice offered,
researchers generally agree that expertise and trustworthiness are important constructs in
persuasion and both have a positive effect on attitude change (Ohanian, 1990).
The premise behind the attractiveness theory is that effectiveness of the
message depends on the source’s familiarity, likeability, similarity/identification and
attractiveness to the message recipient. Likeability refers to the “presence or absence of
feelings that the receiver of a message would have towards a source of product
information” (O’Mahony and Meenaghan, 1998, p.17). While it may be argued that
likeability is related to attractiveness – the more attractive the more likeable – there are
conflicting opinions on just how important likeability is as a persuasive characteristic
(Kahle and Homer, 1985).
As for similarity, it is argued that the more in common the receiver has with the
source, the greater the persuasiveness of the message. Similarity/identification in this
case includes factors such as attitudes, opinions, activities, background, and lifestyle
(O’Mahony and Meenaghan, 1998).
In many cases, it is the physical appearance of the source, and in particular, his or her
degree of attractiveness, which conveys social value to the consumer. Research has found
that attractive communicators are consistently liked more than non-attractive
communicators and as a result they exert a positive impact on products with which they
are associated (Ohanian, 1990).
Credibility and physical attractiveness of the source facilitate persuasion
(Stephens and Hill, 1994) and as such perceived attractiveness is said to be synonymous
with persuasiveness (Stephens and Hill, 1994; Mazis et al., 1992). However,
highly credible sources are not always more effective than less credible ones. Research
Personalisation of online avatars 147
has revealed that “… when the consumer is already favourably predisposed to the
message a less credible source can induce greater persuasion than a highly credible
source” (Ohanian, 1990, p.42).
To compound the process, a source can take the form of an expert, a glamorous
celebrity, a ‘typical’ consumer or even an animated character. The use of celebrities
including actors, fashion models and athletes is a common strategy. Celebrities represent
an aspirational lifestyle for many consumers and often act as a reference group for
purchase decisions (Kamins, 1990). The premise behind the use of celebrity endorsers are
that they will not only draw attention to the brand, but also that the image values
associated with the celebrity will be transferred to the product (O’Mahony and
Meenaghan, 1998; Englis et al., 1994; Till and Shimp, 1998). These image values can be
positive or negative. Just as positive source evaluations are expected to increase brand
evaluations, negative evaluations may lower them (Till and Shimp, 1998).
Despite the claim that positive and/or negative feelings towards the source will
transfer to the product, there is much debate as to the influence this will have on actual
purchase behaviour (Kamins, 1990; Caballero et al., 1989). Some argue though that
negative information/feelings really only pose a problem for unfamiliar products,
non-established brands or celebrities with which the consumer has limited knowledge.
When knowledge structures (for the brand and/or celebrity) are more fully developed, the
negative impact of the source is diminished (Till and Shimp, 1998).
To overcome the potential pitfalls associated with negative publicity surrounding
celebrities, advertisers are increasing the use of spokescharacters as product endorsers.
“A spokescharacter is an animate being or animated object that is used to promote a
product, service or idea” (Phillips, 1996, p.155). Spokescharacters have been used in
advertising since the late 1800s. Traditionally, characters have been associated with low
involvement products such as food items and cleaning supplies, but they are now being
used more extensively to promote high involvement purchases such as computer
hardware and software. For example, characters from the movie Shrek are being used in
advertising for Hewlett Packard computer technology (Marketing Vox News, 2003).
Until now, research in marketing and advertising has focused on finding the correct
match between the source of communication, the model or spokesperson, and the
product. The premise behind the match-up hypothesis is that the correct ‘match-up’
between the celebrity and the product may have a greater influence on consumer attitudes
towards the product and purchase intention than endorser likeability or product
involvement (Kamins, 1990; Solomon et al., 1992). Research by Kamins (1990) and
Solomon et al. (1992) revealed the need for congruency between the product image
and that of the celebrity when the product is attractiveness related. For products
that are attractiveness unrelated, the need for congruity was not found to be relevant
(Kamins, 1990). Additional research on the source selection for apparel advertising found
that whereas source trustworthiness, likeability and attractiveness were important in
attracting attention, credibility and expertise were more important in influencing purchase
intentions (O’Mahony and Meenaghan, 1998).
3.2 The communication source in online environments
While traditional media sources are often standardised in that the same source is utilised
to target an entire market segment, this is not necessarily the case online.
The interactive capabilities of the internet allow the source to be personalised to suit the
148 N.T. Wood, M.R. Solomon and B.G. Englis
purchasing situation or individual consumer. This interactivity raises the question: “Is it
possible to have a match-up between source and the consumer that will yield similar or
even better results than a match-up between the source and the product?”
While personalisation per se is by no means a new phenomenon to the
marketing world, personalisation of the source is. Personalisation evolved from the world
of one-to-one marketing where marketers believed the key to success lay in mass
customisation (segments of one) rather than mass marketing. Peppers and Rogers (1998)
claim that one-to-one marketing creates long-lasting, profitable relationships.
Traditionally, personalisation was associated with the purchasing situations in which the
company could justify the cost of personalising elements of the promotional mix
(e.g., business-to-business markets). However, low-cost computer assisted
personalisation makes it increasingly practical for a larger number of products and
services to use a more personalised approach (Peppers and Rogers, 1995).
The most common form of personalisation on the web today is personalisation of the
communication message through the use of ‘service scripts’. A simple example of this is
modifying a somewhat standardised message to include the consumer’s name. In the
offline space, this simplistic strategy has been termed ‘cosmetic personalization’
(Surprenant and Solomon, 1987).
However, a personalisation strategy obviously is not limited to the actual message; it
can also include the source that delivers it. While in traditional environments and
marketing communications personalisation of the source is limited for logistical reasons,
this is not necessarily the case online. The ability to personalise the source through the
use of avatars is rapidly becoming a reality.
4 Avatars as communications sources
The birth of avatars as we know them has been attributed to the success of computer
games. Over time, the use of avatars for recreational purposes has spread into virtual
communities; online environments in which internet users become ‘cytizens’ by adopting
another persona in the form of an avatar and communicating, interacting and shopping
with other ‘cytizens’ in virtual worlds. The latest version of Yahoo Messenger allows
users to create their own avatar to assist in communicating with others. Users customise
the avatar’s appearance by selecting its hair style, eye and hair colour and clothing
Avatars are now beginning to appear in online advertising, on e-commerce sites and
even on sophisticated vending machines as a mechanism for communicating advertising
messages and enhancing the purchase experience. For example, the artificial-life.com
website utilises Lucy, an animated avatar that can answer users’ questions and helps them
locate relevant information on the site. Brown & Williamson Tobacco Corp. developed a
high-tech vending machine featuring a cast of virtual characters on a video screen that
‘speak’ to customers as they are choosing a brand. When a smoker tries to buy a pack of
Marlboros made by rival Philip Morris Cos., a virtual vixen with a sultry voice and lurid
red lipstick entices him or her to switch to Lucky Strikes instead: “Toasted and delicious,
and I’ll give you a pack for 75 cents off ” (Tran and Regalado, 2002).
Now, rock bands, soft drink makers and other big-time marketers are using avatars as
well. Coca-Cola Co. recently launched an avatar-populated site for the Hong Kong
market where avatars mill around and chat in a Coke-sponsored world. British Telecom is
Personalisation of online avatars 149
also testing such products as avatar e-mail, software that makes the sender’s face appear
and speak the message aloud (Radcliffe, 2004). Avatars are even appearing on religious
websites – the Methodist Church of Great Britain is sponsoring the first virtual church
(churchoffools.com) where users choose an avatar that kneels in worship and listens to a
sermon (Lawton, 2002).
Avatars can provide a cost-effective way of personalising the communication source.
And, if combined with intelligent agent software, which creates a profile of the consumer
by recording her online behaviour (searches and purchases), they will even have the
ability to screen products and make recommendations providing a higher degree of
personalisation normally associated with high contact services (e.g., healthcare,
haircutting). To illustrate, Swiss watch manufacturer Swatch offers animated personal
shoppers that online shoppers can create, hire and fire. Once created, this personal avatar
asks the user three or four lifestyle questions in order to recommend styles that best suit
the shopper’s personality.
One important issue to consider in selecting a source is the wear out factor. ‘Wear
out’ is the term used to describe the decline in persuasiveness of an advertising message.
This generally occurs when a message has reached its target market, achieved its
objective and then loses its persuasiveness (Blair, 2000). The theory of wear out may also
apply to the source of the communication, in this case the avatar. After a period of time,
the avatar may become so annoying to the consumer that they fail to respond to it. To
overcome this, avatars will need to change overtime to minimise wear out.
An avatar that offers advice and recommendations regarding purchasing decisions
may in some cases result in the blurring of boundaries between spokesperson and opinion
leaders. For an avatar to become an opinion leader, first two conditions must exist:
• the information that they offer must be accepted frequently by message recipients
• their recommendations must be acted upon by the majority of those who receive
Second, the avatar must possess two qualities: expertise and trustworthiness. It may be
argued that trustworthiness implies impartiality, that the opinion leader should have no
vested interest in the outcome (the product or brand selection) of the purchase decision
(Sheth et al., 1999). An avatar that operates on a brand specific website (such as
www.swatch.com) may not satisfy these criteria. However, one that operates on a more
‘neutral’ website, such as www.epionions.com, a website dedicated to assisting
consumers make informed buying decisions may represent an opinion leader.
As Figure 1 shows, avatars can be classified along four dimensions:
• Function. An advertiser has the option of using the avatar as a decorative or
proactive element of the message. In other words, the avatar may simply be a
passive visible enhancement (see www.mysimon.com), one whose aim is to draw
attention to the product or site rather than deliver information, or it can be an active
component of the message that delivers (and perhaps collects) information to the
consumer and facilitates the online experience (see www.bonzi.com).
• Action. The avatar can appear motionless on the site, much like a still image
(www.mysimon.com), or it can be animated like BonziBuddy a purple gorilla who
moves around the web page, reads books and when not being used takes a nap.
150 N.T. Wood, M.R. Solomon and B.G. Englis
• Representation. The avatar may be presented as the real image of a person,
(photograph) (see www.pepsiworld.com/help/index.php), or it can be presented in
character form (cartoon or sketch) (see www2.coca-cola.com/contactus/).
• Classification. Finally, the avatar can be classified as an image of the actual site user
(www.ezface.com/static/customer/index.html), a real person such as a ‘typical’
consumer (www.customersaskford.com/kate/), or an idealised image like a fashion
model or a celebrity (www.time2bcool.com), representing a continuum from the
depiction of a realistic person to an idealised image.
Figure 1 Functions and manifestations of avatars
The creation of avatars for commercial formats is evolving into a cottage
industry as demand for compelling figures begins to grow. For example, noDNA, a
German based company (www.nodna.com) offers a variety of ‘virtualstars.’
These are computer-generated (non-humanoid) figures that appear as caricatures;
‘vuppets’ – cartoon type mascots and animals and ‘replicants’ – that are doubles of real
people (humanoids). Each of these actors can be developed to meet the needs of specific
target markets; the physical appearance, personality and communication message can be
calibrated to a particular group of consumers. The company’s contractual work is
evenly split between non-humanoid and humanoid avatars with female characters
representing 90% of their human characters (Schirm, 2002). Even Elite Model
Management, the agency that represents many of today’s supermodels, has created a
virtual digitally composed supermodel, Webbie Tookay. With a personality tailored to a
client’s needs, she is available for a licensing fee for all kinds of media related work
including internet advertising.
The advantages of virtual avatars compared to flesh and blood models include the
ability to change the avatar in real time to suit the needs of the target audience or
individual consumer. For example, in traditional advertising and shopping environments,
it is both unrealistic and technically impossible to change the model to satisfy the needs
of each individual consumer. However, in an online environment, this flexibility is quite
feasible. An infinite number of consumers may be logged onto the same website, at the
same time, viewing the same products, but each consumer could be viewing and
receiving information from a different source. From an advertising perspective, virtual
avatars are likely to be more cost effective than hiring a real person; from a personal
selling and customer service perspective, they have the ability to handle multiple
customers at any one time, are not geographically limited, and are operational 24 hours a
day, seven days a week freeing up company employees and sales personnel to perform
Personalisation of online avatars 151
On the majority of websites mentioned thus far, the avatar is standardised. That is, the
avatar is chosen for the site by the company, based on its perception of what is
appropriate. The user does not have the ability to select or personalise his or her avatar.
The selection of an avatar for any one site is normally made at the divisional level and is
based on a number of factors including the type of company (e.g., finance, travel) and
objective of the site (e.g., source of information, strengthening brand image, online sales)
rather than based on the personal needs and wants of the user (Fowler, 2001).
While companies may feel that their selection is appropriate, this may not always be
the case. For example, Eagle Star (www.eaglestar.co.uk) a British insurance company felt
that a professional looking male avatar was most appropriate for its site. However, focus
group findings revealed customers preferred an avatar of a young, somewhat casually
dressed female (Fowler, 2001). This implies that in online environments, the match-up
between the source and the user may be more appropriate than the match-up between the
source and the product.
The current investigation is a first attempt to explore consumers’ preferences for different
types of avatars. More specifically we aim to:
• determine what types of avatars consumers prefer to encounter on websites – if any
• compare and contrast consumer avatar preferences with those currently being utilised
• make recommendations to advertisers and marketers regarding the types of avatars
they need to adopt to satisfy the needs of the female online shopper.
To explore these issues a two-part study was conducted. Rather than attempting to
address all the functions and manifestation of avatars as presented in Figure 1, this study
explored the use of avatars as a direct source of communication. The avatars were real
and idealistic images in real/photographic and character form and all were motionless.
Animated avatars and those created in the consumer’s own image as direct and indirect
sources were excluded from the study.
Furthermore, this study focuses specifically on female shoppers. It is estimated that
women now represent 52% of the adult online population. Recent research on female US
internet users revealed that women are placing greater reliance on the internet as a form
of media. Thirty percent of working women surveyed stated that the internet was a ‘very
important’ source for making retail purchasing decisions compared to just 6% who said
the same thing about television. Forty-eight percent also indicated that they had increased
their internet usage over the previous year (eMarketer, 2004) and 60% of all US online
purchases made in 2003 were made by women (Rush, 2004). In terms of product
categories, apparel experienced the third highest revenue growth rate in 2003
(behind videos/DVDs and books) with an increase of 35% (Netratings Inc., 2004b). This
growth attests to the need for research that specifically explores elements of website
design (such as avatars) that may enhance the online apparel shopping experience for
152 N.T. Wood, M.R. Solomon and B.G. Englis
5.1 Study 1: avatar selection
The first component of this exploratory study involved 55 regular internet users
identifying images (avatars) they would like to see on an apparel website. The purpose of
this exercise was to identify types of images of avatars that users want to see online and
to compare them to those currently available/utilised online.
As an extra credit assignment, 55 undergraduate female students (aged 18–30) at a
large southwestern university were asked to each identify four images of potential avatars
for an apparel website. Subjects were instructed to identify an image of a person or
character they would like to have as an avatar to assist them (by making
recommendations) with the purchase of a bathrobe, lingerie, raincoat and dress. They
were asked to provide one image per product and were given complete latitude in the
source and selection of these images. The exercise provided a total of some 200 images
of potential avatars. There was a high degree of diversity in the selection of avatars, from
the character Miss Piggy to complete strangers, and from a variety of sources including
websites, books, magazines and photographs. After removing duplicate images, a total of
85 unique images were identified. Only 10 of these images were male. Avatars were next
categorised according to their manifestation (Table 1).
Table 1 Avatar manifestation
Realistic image/person: photographic form 21
Realistic image/person: character form 1
Idealistic image/person: photographic form 44
Idealistic image/person: character form 19
Our results clearly indicate that photographic images are more popular than those in
character form. Furthermore (at least in this study), the subjects preferred idealistic
images (celebrities, models) to realistic ones, and females over males. These findings
raise the question of whether our subjects’ preferred avatars are representative of the
kinds currently being used on e-commerce sites.
In fact, contrary to our subjects’ desire for photographic avatars, there is a distinct
lack of websites currently employing avatars in this format. As a result, it is difficult to
make such comparisons. However, comparisons can be made for the characters. On the
websites observed (and previously mentioned), character avatars currently being utilised
are both head and torso or full body shots. At the time of writing, these websites were,
and, to a large extent, still are the only ones utilising character avatars
(see www.swatch.com). A review of the character avatars submitted by subjects revealed
that only 3 of the 19 (idealistic/character avatars) submitted were full body images. Given
the ample supply of full body character images available on the web and in the print
media, could it be that subjects have a preference for head and torso images only? A large
number of the images provided appeared to have been ‘cut-off’ at the neck or chest.
Could it be that subjects only evaluated the facial features of the individual when making
their selection? Or that body shape/form was not important, or that subjects did not want
to see it?
Personalisation of online avatars 153
Similar to the character (cartoon or sketch) avatars provided, there is very little
evidence of exaggerated physical features (e.g., overstated lips) on commercially
used characters. But what is evident is that a large number of these commercially used
characters are very attractive (see www.nodna.com). They may not have exaggerated
physical features, and because of this the claim may be that they are more natural
looking, but they do largely represent an ideal image rather than a realistic one. They are
attractive both in facial features and physical stature. Their stance and posture is more
contrived. They are also all idealistic images of unknown individuals in comparison to
the subject-generated characters, which were a combination of the familiar (65%) and
unfamiliar (35%). It appears the users have a preference for idealistic avatars with which
they are familiar rather than those that have been merely created for the task at hand.
Avatar developers attempt to overcome this lack of familiarity by providing the avatars
with their own personality profile (see www.nodna.com). However, the ability to
humanise characteristics to the extent that consumers are as comfortable with them as
they are with avatars of real/familiar people is questionable and an issue that needs to be
further explored. User preference for an avatar in photographic and idealistic suggests a
potential opportunity for avatar developers. Companies such as Illusion2K may have
greater success with a photographic avatar of their current fashion models rather than
attempting to create new models with fabricated identities of which consumers have no
5.2 Study 2: avatar feedback
To explore further consumers’ attitudes towards and willingness to utilise avatars, a
sample of avatars was selected for further exploration. The first step involved the
selection of avatars. From the pool of 85 images previously collected, 10 were randomly
selected from each of the four categories (Table 1) for inclusion in the sorting task,
yielding a sample of 40 images. For the avatar category realistic/character for which only
one image was initially supplied, a random sample of nine images from a number of
websites both promoting and hosting avatars was selected.
A new sample of 24 females (18–30 years) from the same southwestern university
were then presented with the 40 images and asked to indicate if they would like to see
this avatar on an apparel website as a source of product communication. To avoid
extreme biases, the second most and the least popular from each category were
selected for further analysis. The end result was a total of eight avatars that would be
utilised for closer evaluation (Figure 2). Four of the avatars were in character form
(2 × real, 2 × ideal) and four in photographic form (2 × real, 2 × ideal).
154 N.T. Wood, M.R. Solomon and B.G. Englis
Figure 2 Avatar rankings based on willingness to adopt for each product
Next, a sample of 139 college-educated females (18–30 years) residing in Southern
California was recruited to participate in an online exercise to provide feedback on the
selection of eight avatars and their willingness to use avatars (in general) and the eight
avatars presented (specifically) as a source of communication for apparel related
purchases. In order to participate, subjects were required to have access to the internet
from home, to have used the internet at least once in the last month to search for a
product, and to have performed at least one transaction online in the last six months.
These basic parameters ensured that all subjects had the basic internet skills to participate
in an online study and the demographic description of age and education characterised
the average internet user. The subjects were recruited using a referral process by four
research assistants and were remunerated with two adult-priced movie tickets.
The images of the eight avatars were placed on a website, which the subjects were
asked to access from home. Once they entered the website, they were presented with a
series of basic demographic and internet usage questions. Next, the subjects were
Personalisation of online avatars 155
provided with an introduction to avatars and what they do on websites. They were
presented with four separate product purchasing scenarios, each emphasising the
purchase of a different product (see Table 2), for which they may enlist the aid of an
avatar to assist them in making their decisions. The subjects were asked to indicate
if they would be willing (yes/no) to use an avatar (in general) to assist in furnishing
information regarding appropriate products for each of the four products. They were then
asked to indicate on 7-point Likert scales the likelihood of their using each of the eight
specific avatars to assist them with their online transaction for each of the
product selection scenarios.
Table 2 Product/scenario
Dress Friend’s wedding
Lingerie To complement a new outfit
Bathrobe To use after a shower and to lounge around the house
Raincoat Prepare for wet weather/winter season
Each avatar was displayed with and referred to by its first name. Where appropriate real
names were used. For example, the name Meg was placed under the image of the avatar
Meg Ryan. In situations where an avatar did not have a name, one was provided for
him/her. In this case, a common everyday name that did not represent any specific
nationality was utilised (e.g., Dan and Amy). Avatars were then ranked from most
popular to least popular for each product. The subjects were also asked to select their
most and least preferred avatar for each of the four products. The final step involved the
subjects answering a series of questions on their attitude towards and willingness to use
avatars, again using 7-point Likert scales. Statistical analysis for this exploratory study
was limited to general measures of central tendency (frequencies, means, etc.).
Subjects were characterised as young (21–23 years of age: 60.4%), computer-savvy
women who are ‘comfortable’ with searching the internet for information and products
(75%). They were either completing (57%) or had obtained a college degree (43%)
In terms of willingness to utilise an avatar (in general) as a source of product information,
the results were favourable and somewhat similar for three of the four products.
Using basic frequency analysis, we found that more than half of the subjects were
willing to use an avatar for product information for raincoats (53.2%), lingerie (53.2%)
and bathrobes (56.8%). For dresses, however, only 27.3% of all subjects indicated that
they would utilise an avatar. This may be attributed to the high level of risk associated
with purchasing an item in which fit is extremely important and sizing can vary from
brand to brand. Given the subjects’ limited exposure to and lack of knowledge of avatars,
these preliminary results are promising.
We found that avatars in photographic/idealistic form were the most popular, taking
first and second choice for all four products (Figure 2). The least popular for all products
was the male avatar (character/real) and the second least favourite a female (photo/real).
Overall, avatar preferences were relatively consistent over all four products.
156 N.T. Wood, M.R. Solomon and B.G. Englis
Findings on attitudes towards avatars revealed that the subjects slightly agree with the
statement that avatars will make a website more visually appealing (M = 4.87,
SD = 1.42), and whereas they agree that avatars will be useful for communication product
information (M = 5.23, SD = 1.18), there is some uncertainty as to whether they would
use them to locate apparel items to purchase online (M = 4.45, SD = 1.53). This may
indicate that whereas avatars are a good source of product information, there is some
degree of uncertainty as to the expertise of the avatar to locate appropriate products.
What is interesting is that the subjects indicated a preference for selecting their own
avatar rather than being presented with one by the website (M = 5.94, SD = 1.16), which
is the current strategy employed by commercial websites.
Similar to the findings of the image generation exercise in Study 1, subjects in Study 2
showed a compelling preference for the avatars in photographic form as well as those
classified as idealised images. This now raises the question: what is it about
photographic/idealised avatars that make them more appealing than characters?
A number of observations can be made about the photographic avatars used in the
experiment. As previously mentioned, the experimental avatars are head-shot only
images. Unless subjects have prior knowledge of the avatar, it’s likely that they are
responding to the facial characteristics of the avatar. For example, subjects would be very
familiar with one of the avatars Meg (Ryan) and are likely to be using existing
knowledge of the actor when selecting an avatar. With the avatar Tanya (a photograph of
an unknown individual), they are more than likely evaluating her facial characteristics to
determine her appropriateness.
Unlike photographs, character avatars are not real. While the subjects may be able to
make a connection with a photographic avatar because of its realness (even when the
individual is a stranger), this may be more difficult to achieve with a character. This of
course is unless the character is well known. For example, the character Jessica Rabbit
has appeared in movies, cartoons and advertising and is a celebrity in her own right.
Similarly, there is numerous advertising ‘spokescharacters’ such as The Jolly Green
Giant, The Michelin Man, the Keebler Elves, etc., which could potentially be transferred
to online environments.
The experimental avatars also varied in terms of the degree to which their level of
attractiveness was realistic or idealistic. As in the offline world, advertisers can choose to
have ‘supermodels’ (whose features may be further enhanced by airbrushing and other
techniques) or ‘everyday’ people. When given the choice between real and ideal
avatars, the majority of subjects preferred the assistance of an idealised image, with an
idealised photographic avatar proving to be the most popular. Considering that the large
majority of subjects would have recognised at least one of these idealised images
(Meg Ryan – actor), and possibly the other (Laetitia – Victoria Secret lingerie model),
this pattern suggests once again that familiarity may have contributed to their selection.
If given the option between an unknown yet idealised image of beauty and a recognisable
idealised image, the subjects may well select the recognisable image. If this familiar
image were in character or photographic form, it appears the subjects would prefer the
Personalisation of online avatars 157
Based upon the rankings of avatars for each of the four products, it seems that if a
participant dislikes a particular avatar, that negative affect is potentially going to apply to
the products he or she is promoting. Forty-two percent (n = 59) of subjects chose the
same avatar as their least favourite for all of the four products. Of those subjects who
selected one avatar, the majority of them (59%) selected the male avatar. This finding is
consistent with previous research which found that if a product truly belongs to one
gender (such as apparel) it is advisable to keep the gender of the spokescharacter the
same as that of the target audience (Peirce and McBride, 1999).
According to balance theory (Heider, 1958), if a consumer is in a real purchasing
situation where a website is using an unfavourable avatar to promote a series of products,
a state of imbalance in consumers’ perceptions would occur. This imbalance would create
a state of tension that the consumer would be motivated to eliminate. To restore a state of
balance the consumer can choose to do a number of things. She can change her attitude
towards the avatar from negative to positive, or she can change her attitude towards the
product from positive to negative. She can attempt to negate the relationship between
the avatar and the product by ignoring the avatar. Alternatively, she may choose to
abandon the activity/transaction.
When promoting a product, advertisers strive to find the right balance between the
source of the communication, the product and the consumer. By choosing an appropriate
source, it is hoped that evaluations of the source will transfer to the product. However,
source evaluation is a two-edged sword. Just as positive evaluations of the source can
transfer to the product, so too can negative evaluations (Solomon, 2002). Consequently,
selecting appropriate avatars for a website is just as important as selecting an appropriate
model for an advertising campaign, particularly if the consumer has the ability to switch
to a competing website with the click of a mouse.
On the other hand, the identification of a preferred avatar does not imply there is only
one preferred avatar per product. We did observe some variation among subjects in terms
of their choice of a preferred avatar. Thirty six percent (n = 50) of the sample selected a
total of three different avatars for their preferred avatar with only 22% (n = 30) selecting
the same avatar for all four products. The ability for consumers to select a different avatar
for each product purchase (if that is what they desire) may be preferable to allowing the
selection of only one, even if it is a highly rated one.
Whether consumers are willing to change their avatar selection with the purchase of
each different product is questionable. The decision to alter avatars may depend upon
factors such as the number of items being purchased at one time. For example, if a
consumer logs onto an apparel website seeking three different items (dress, shoes and
purse), she may decide in the interest of simplicity to use one avatar. If, however, she
sought these items on three separate occasions each time she logged onto the site,
she may perhaps choose a different avatar. Although outside the scope of this research,
another factor worth considering is situations in which a consumer is shopping online for
other people. For example, a consumer purchasing apparel items as Christmas gifts for
family and friends may choose to alter the avatar for each purchase depending upon the
recipient of the gift.
158 N.T. Wood, M.R. Solomon and B.G. Englis
6 Strategic implications
The decision to integrate avatars onto e-commerce sites is not to be taken lightly.
The selection of an inappropriate avatar may have negative effects on factors such as
website traffic and brand image, not to mention the cost of creating and maintaining the
avatar. For example, the Scottish Government spent over £ 100,000 alone for the creation
of an avatar on its website (McGrath, 2002). From this preliminary study, a number of
observations and subsequent implications have emerged that an apparel e-tailer should
take into consideration should it choose to integrate avatars on its site.
Overall, the willingness of young female online apparel shoppers to adopt avatars
appears positive. As avatars continue to appear in other venues such as computer games,
online advertising and instant messaging, consumer familiarity and willingness to adopt
may increase. Integration and adoption of avatars on commercial e-tailing sites appears to
be a logical progression for avatar usage. However, careful consideration needs to be
given to the precise visual manifestation of these avatars.
From this study, it was revealed that when targeting the female apparel shopper,
developers and advertisers should perhaps be concentrating their efforts on offering
avatars in photographic as opposed to character form, which is largely the current
practice. This preference however may change. Over the last few years, we have seen an
increase in the number of blockbuster animated movies (Toy Story, Finding Nemo,
Shrek, etc). Given the growth of the animation industry and consumer familiarity with
this form of entertainment, avatars in character form may in time become more
acceptable. But as it stands now photographic avatars are the preferred choices.
Our findings suggest that photographic avatars need to be idealistic rather than
realistic in appearance. As a result, it may be more effective to adapt existing celebrities’
images to an online format than to invest the effort involved in creating new characters.
Familiarity may also assist in increasing credibility, and prove to be more persuasive
while offering the added advantage of maintaining a universal brand image across all
forms of advertising. When, and if, characters are to be offered, they should be ones that
the consumer is already familiar with. If the character is unfamiliar, then an identity must
be created for it. And, e-tailers need to provide greater selection of avatars on their
website. One or two avatars are not likely to satisfy the needs of all their target market.
Unlike the current practice of utilising the same avatar for all purchases, multiple avatars
should be offered and consumers should be presented with the option of changing their
avatar when they purchase different products.
When trying to select appropriate avatars for a given website, consideration needs to
be given to not only what avatar best fits the product but also what avatar best fits the
consumer. To date, marketers have concentrated their efforts on finding the correct
match-up between the message and the product, but these results indicate that in the
online space, the match-up between the source and the consumer may be a key
determinant of the virtual purchasing experience.
The sample population and more specifically its method of recruitment is the central
limitation of this study. Non-probabilistic sampling increases the likelihood of sample
bias. Individuals referred to the experiment may have similar characteristics as the
Personalisation of online avatars 159
individuals making the referral. This makes it difficult to draw conclusions and
inferences regarding the target population, as the sample may not be entirely
representative of the population of interest. The study also only focused on one product
category – female apparel. More conclusive experimental research is required to validate
8 Directions for future research
This research both answered and raised a number of questions, for which additional
research is needed. First, the current study needs to be extended to include a more
representative population (including male subjects) as well as a wider variety of products.
Factors such as the ability of avatars to cross geographic and cultural boundaries also
need to be considered: How accepting are other cultures of avatars and in what form
should they be created? For instance, whereas US users displayed a preference for
photographic avatars, in other countries such as Japan that tend to incorporate animated
figures into many facets of their popular culture (e.g., anime), a character avatar may
prove to be more popular and effective.
In terms of the actual avatars, additional research is needed to determine if the extent
of the image (i.e., head only, head and torso or full body) has any influence on adoption.
Future research should also investigate match-up between the product, the individual
consumer and the avatar to determine if, and when, such a relationship exists, and the
extent to which it can persuade consumers to purchase online. Finally, a series of
experimental studies exploring the persuasive ability of avatars are necessary to lend
support for expending the extra effort and money to develop technologies that will enable
websites to offer users the ability to customise and personalise their avatars.
Although beyond the scope of this study, it is also important to point out that in reality
advertisers have many options when it comes to employing avatars to execute their
message. For example, marketers may adopt an execution strategy in the form of a
testimonial, whereby a realistic avatar commends the product based upon his or her
personal experience with it. Alternatively, if an idealistic avatar such as a celebrity were
adopted, an endorsement may be used. Likewise a straight sell, factual message or
comparative format may be used.
Whereas personalisation is by no means a new phenomenon in offline environments,
its capabilities online are yet to be fully appreciated. What once was impractical offline
has become somewhat practical online. Empowering consumers to select their own
source of communication, as demonstrated in this preliminary research, may offer a more
effective may of developing persuasive messages.
160 N.T. Wood, M.R. Solomon and B.G. Englis
Belch, G.E. and Belch, M.A. (2001) Advertising and Promotion, Irwin McGraw Hill, New York.
Blair, M.H. (2000) ‘An empirical investigation of advertising wearin and wearout’, Journal of
Advertising Research, Vol. 10, No. 6, pp.95–100.
Caballero, M.J., Lumpkin, J.R. and Madden, C.S. (1989) ‘Using physical attractiveness as an
advertising tool: an empirical test of the attraction phenomenon’, Journal of Advertising,
Vol. 29, No. 4, pp.16–22.
eMarketer (2004) Busy Women Choose the Internet, March 31, http://www.emarketer.com
Englis, B.G. and Solomon, M.R. (2000) ‘Life/style online: a web-based methodology for
visually-oriented consumer research’, Journal of Interactive Marketing, Vol. 14, No. 1,
Englis, B.G., Solomon, M.R. and Ashmore R.D. (1994) ‘Beauty before the eyes of beholders: the
cultural encoding of beauty types in magazine advertising and music television’, Journal of
Advertising, Vol. 23, No. 2, pp.49–64.
Fowler, S. (2001) Personal Communication, 2nd March.
Hamilton, A. (2004) ‘Power chat’, Time, May 3, p.91.
Heider, F. (1958) The Psychology of Interpersonal Relations, Wiley, New York.
Internet World Stats (2004) Internet Usage by World Regions, May 14, http://
Kahle, L.R. and Homer, P.M. (1985) ‘Physical attractiveness of celebrity endorsers: a social;
adaptation perspective’, Journal of Consumer Research, Vol. 11, No. 4, pp.954–961.
Kamins, M.A. (1990) ‘An investigation into the ‘match-up’ hypothesis in celebrity advertising:
when beauty may be only skin deep’, Journal of Advertising, Vol. 19, No. 1, pp.4–13.
Lawton, C. (2002) ‘Virtual characters push cigarettes in new vending machine’, The Wall Street
Journal, August 6, p.B1.
Marketing Vox News (2003) HP to Spend Big on Multi-channel Campaign, November 18,
Mazis, M.B., Jones Ringold, D., Perry, E.S. and Denman, D.W. (1992) ‘Perceived age and
attractiveness of models in cigarette advertisements’, Journal of Marketing, Vol. 56, No. 1,
McGrath, D. (2002) This Avatar’s Got News for You, 9th October, http://www.wired.com/news/
Netratings Inc. (2004a) Global Internet Index, May 14, http://www.nielsen-netratings.com/
Netratings Inc. (2004b) Online Shoppers spend $2.95 billion in the Second Week of December,
May 25, http://www.netratings.com/pr/pr_031223_us.pdf.
O’Mahony, S. and Meenaghan, T. (1998) ‘The impact of celebrity endorsers on consumers’, Irish
Marketing Review, Vol. 10, No. 2, pp.15–24.
Ohanian, R. (1990) ‘Construction and validation of a scale to measure celebrity endorsers’
perceived expertise, trustworthiness, and attractiveness’, Journal of Advertising, Vol. 19,
No. 3, pp.39–52.
Peirce, K. and McBride, M. (1999) ‘Aunt Jemima isn’t keeping up with the energizer bunny:
stereotyping of animated spokes-characters in advertising’, Sex roles, Vol. 40, Nos. 11–12,
Peppers, D. and Rogers, M. (1995) ‘The end of mass marketing’, American Demographics,
Personalisation of online avatars 161
Peppers, D. and Rogers, M. (1998) ‘Better business–one customer at a time’, The Journal for
Quality and Participation, Vol. 21, No. 2, pp.30–37.
Phillips, B.J. (1996) ‘Defining trade characters and their role in popular culture’, Journal of
Popular Culture, Vol. 29, No. 4, pp.143–158.
Radcliffe, L. (2004) ‘Log on for salvation’, Newsweek, May 31, p.10.
Rogers, E.M. and Albritton, M.M. (1995) ‘Interactive communication technologies in business
organizations’, The Journal of Business Communication, Vol. 32, No. 2, pp.177–196.
Rush, L. (2004) Women, Comparison Shopping and help Boost E-commerce Holiday Revenues,
May 24, http://www.clickz.com/stats/markets/retailing/article.php/6061_3299531.
Schirm, O. (2002) Personal Communication, 13th August.
Schram, W. (1955) The Process and Effects of Mass Communication, University of Illinois Press,
Sheth, J.N., Mittal, B. and Newman, B.I. (1999) Customer Behavior, New York, Dryden.
Shimp, T.A. (2003) Advertising, Promotion & Supplemental Aspects of Integrated Marketing
Communications, South-Western, Ohio.
Solomon, M.R. (2002). Consumer Behavior, Prentice Hall, New Jersey.
Solomon, M.R., Ashmore, R. and Longo, L. (1992) ‘The beauty match-up hypothesis: congruence
between types of beauty and product images in advertising’, Journal of Advertising, Vol. 21,
No. 4, pp.23–34.
South China Morning Post (2000) ‘Avatar’s world puts reality behind mask’, South China Morning
Post, 15th August, p.6.
Stephens, D.L. and Hill, R.P. (1994) ‘The beauty myth and female consumers: the controversial
role of advertising’, Journal of Consumer Affairs, Vol. 28, No. 1, pp.137–154.
Surprenant, C.F. and Solomon, M.R. (1987) ‘Predictability and personalization in the service
encounter’, Journal of Marketing, Vol. 51, No. 2, pp.86–96.
Till, B.D. and Shimp, T.A. (1998) ‘Endorsers in advertising: the case of negative celebrity
information’, Journal of Advertising, Vol. 27, No. 1, pp.67–82.
Tran, K.T.L. and Regalado, A. (2002) ‘Web sites bet on attracting viewers with humanlike
presences of avatars’, The Wall Street Journal Interactive Edition, 24th January,