duringrepeated exposures by measuring the position of first fixation, time totarget
fixation and sequence of fixations (scan path). In both the studies,the eye tracking
data issupported by qualitative data from pre- and postsessionquestions which elicited
the users' expectations and preferencesabout the target Web-link.
Keywords: Web design guidelines, eye tracking, visual search behaviour,
homepages, e-commerce, user-adaptation.
Ekaterini Tzanidou, Shailey Minocha, Marian Petre & Andrew Grayson
1 Introduction

When considering the design of websites for e-commerce, the HCI literature

proposesa variety of design guidelines for websites [Nielsen et al. 2001; van Duyne et
al.2003]. In addition, others have suggested that these guidelines have been derived
by personal reviews and anecdotes without reporting the involvement of human
participants [Ivory et al. 2001].

The homepage is viewed more than any other page on a website and first

impressions are important in attracting new users. Nielsen & Tahir propose that:
The homepage must communicate in a short glance where users are,what your
company does and what the users can do at your site. [Nielsen & Tahir 2002]van
Duyne et al. [2003] suggest that the first visit of the homepage is often the
sitebranding which must be initially clear to users. A detailed study has been
reportedin Nielsen & Tahir [2002] of heuristic evaluations of 50 homepages.Although
theguidelines proposed in these studies are straightforward to implement, it has been
stated that they could be too general to apply to a particular case, so a wide range of
websites are not supported [Beier & Vaughan 2003]. The first study reported in this
paper used eye tracking to investigate what users inspect on their first visit to a
homepage and how quickly can they establish the brand identity. The study recorded
what users looked at first (i.e. position of first fixation) and their initial scanning
behaviour. The eye movement data was compared to users' responses to questions
about company identity and services during the post-session interviews.Another set of
design guidelines is based on an assumption that visual search behaviour is shaped by
expectations, hence they suggest designing user interfaces that conform
toconventions. Nielsen et al. underline the importance of maintaining consistency
with other websites and webpages:
All webpages are much the same from the user's perspective, they share interaction
techniques, they are downloaded (slowly) from the Internet, and they have relatively
similar layouts. Those similarities are in fact good because they allow users a measure
of transfer of skill from one site to the next. Users complain bitterly when a site
doesn't try to use navigation from the majority of other sites. [Nielsen et al. 2001, p.
189] But how do users learn conventions and develop expectations? Ehret [2002]
suggests that when locations of design elements remain constant, performance
improves over exposures as users learn the placements of design elements. This
implies that consistency of placement of a design element influences visual search.
But a recent eye tracking study by McCarthy et al. [2003] investigated the impact of
changing the location of design elements and how users performed when viewing the
element in unexpected locations. They found that following conventions with other