In September, 2000 Razorfish, Germany was charged with the task of
Rather than describe the project from beginning to end, this case study focuses on three aspects of particular interest:
- Razorfish’s approach to schematics (i.e., wireframes).
- An automated page layout technique referred to as “jumping boxes.”
- A user test that compared the performance of a left-hand navigation to a right-hand navigation.
Many web projects suffer from a lack of “traceability.” By this I mean the ability to trace a concept, idea, element, or artefact across a set of documents.
Unless a project employs all-encompassing document management tools, documents tend to end up separate and independent from one another. They are often owned by different people, reside in different locations, and are created in different formats. It is not uncommon that, by the end of a project, updating something as simple as a navigation label requires updating half a dozen documents or more. This is inefficient and leads to version control problems.
To address this problem, Razorfish, Germany turned to Adobe GoLive 5.0 in hopes of achieving a true convergence of documents. The plan was to integrate a range of deliverables, including sitemaps, schematics, text content, and screen designs. We even wanted to create functional specifications directly in GoLive in HTML format.
Information was shared between the sitemap and schematics. Updating the page name in the sitemap, for example, updated the page name for the schematic.
Page schematics were created using components. This allowed for the definition of global elements, such as the main navigation. Changes were made across the entire set of schematics very easily.
- File Sharing
Working with a WebDAV server, IAs could check schematics in and out, thus offering version control. Audi was also able to see the schematics “live” online in HTML format through the project extranet.
GoLive is available for the PC and the Macintosh, and the output is simple HTML. Conversions to Adobe PDF, for example, were not necessary.
There were, of course, disadvantages to GoLive:
- File Size
Even without text content and screen designs, the site file for the Audi schematics grew to 30 MB and became unwieldy.
We experienced some crashes and loss of work with GoLive 5.0, which had just been released before the Audi project began.
The sitemap tool is primitive and doesn’t allow a great deal of control over appearance.
- Team Buy-in
The use of GoLive didn’t get the buy-in from the whole Razorfish-Audi team and ended up being used primarily by IAs. In the end, the idea of true document convergence across skill groups never happened.
Overall, GoLive worked well and met most of our expectations, particularly from an IA standpoint. But it still isn’t the right tool for the job and our experience underscores the need for a program that meets all information architecture needs. Though no single technology will solve the problems of site conception and planning, a more appropriate tool would help.
Razorfish, Germany wanted to address the fact that users surf with different
There are many ways to achieve flexible page layouts, but we developed what can be called an automated layout solution. Essentially, the Audi sites have “smart” pages that detect browser size and serve up the right layout automatically. Entire content areas of a page appear in different locations depending on the user’s resolution. These content boxes appear to “jump” around in the layout, hence the phrase “jumping boxes.” Three sizes are offered on the Audi sites —small (640×480), medium (800×600) and large (1024×768+).
There were at least two reasons for this approach. First, it fulfilled corporate design constraints. All page elements are aligned horizontally and vertically on a grid. Automated layout allowed us to better control alignment. Second, the solution is highly technical and speaks to the Audi slogan “Vorsprung durch Technik” (“Advancement Through Technology”). The site is based on JSP modules which are arranged to form a template. A style sheet (XSLT) controls the three possible arrangements of modules for a given template depending on the user’s browser size. This all happens in the front end and does not require extra server requests. In a sense, the layouts were supporting the brand with this technical solution.
An automated layout solution can be complicated to implement depending on the technology involved. For us, it proved to be more challenging than initially thought. Further, it is still unknown if there are any usability implications. We don’t believe so, but to date have no proof. Finally, the automated layout solution is not necessary for all page types.
With an increase of alternative browsing devices on the horizon, the continuum of viewable browsing sizes will continue to expand. Never before has the demand for flexible layouts been greater. Since the web stands at the center of our collective digital attention, solutions developed there can drive solutions in other formats and media. The Razorfish, Germany “jumping box” technique is an innovative technique, and we learned a great deal about page behavior from it.
Try resizing this screensaver download page on Audi.com with an Internet Explorer browser to see the jumping boxes in action.
BMW, Mercedes and other car manufacturers generally have conservative page layouts with the navigation on the left or top. To set Audi apart from its competitors, we placed the navigation on the right side of the page. This solution addresses a core Audi brand value: innovation.
We tested the right-hand navigation extensively with our external partner, SirValuse. Two clickable prototypes, of about 10 pages each, were constructed – one with a left navigation and the other with a right navigation. 64 users were split into two groups of 32 each. This was a very large sample and not a sample of convenience: participants were recruited based on our user profiles and to fit Audi’s target group.
|Prototypes used to test the Audi website.|
The test consisted of three parts:
Part 1: Completion times for six tasks were timed with a stopwatch.
Part 2: Eye movements were analyzed to see where participants tend to look on the page.
Part 3: Users were directly asked what they thought about the right-hand navigation
Our hypothesis for Part 1 was that there would be a significant difference in task completion time for the first task and that by the last task there would be no significant difference in task completion time. We expected that users would need to use the site a couple of times to learn the uncommon pattern of interaction (i.e., a right-hand navigation), but that the learning curve would be very steep.
Part 2 looked at eye movement patterns. Instead of relying on traditional eye-tracking methods that make use of expensive equipment and headgear, we used a new method developed by an agency in Hamburg called Media Analyzer. This technique asks users to rapidly coordinate mouse clicks with where they look on the screen. Each click then represents a focal point of visual attention. A software program captures user interactions for later analysis.
We found that people tended to focus more on the content side of the page with a right navigation than with a left navigation.
In the final part of the test (Part 3), we asked several questions that addressed the central issue, “Do you like the right-hand navigation?” Overall, users were apathetic towards the navigation position. Most didn’t notice that the navigation was on the right and, when directly asked, they didn’t seem to care. However, seven people actually preferred the right navigation to a left navigation, while only two disliked it.
Subsequent usability tests and post-launch user feedback corroborate these findings: there is no apparent difficulty using a right-hand menu to navigate the Audi.com and Audi.de sites.
Though there is research about expectations of the location of page elements in a layout, such research does not correlate breaking these expectations with actual usability (see: Michael Bernard, http://www.internettg.org/newsletter/dec00/article_bernard.html and Jakob Nielsen, http://www.useit.com/alertbox/991114.html). That is, while users normally anticipate a left-hand navigation, positioning the navigation elsewhere does not necessarily result in usability problems.
Don Norman’s concept of affordance —the perceived properties of a thing that determine how it is to be used —seems to be a better predictor of usability than conforming to standards or matching patterns to user expectations. With the Audi site, it is clear what is navigation and what is not. Users can build a pattern of interaction with the site immediately. Our findings show users have no problem distinguishing a right-justified navigation and tend to make generalizations about its function.
This does not mean that all sites should have a right-hand navigation. Indeed, a left-hand navigation may work best in most situations. However, for sites with particularly long texts that require scrolling, for example, a right-justified navigation might be beneficial.
The bottom line is that placing a navigation scheme elsewhere than on the left is not a taboo, contrary to “standards” professed by usability gurus. Without sacrificing usability, Razorfish, Germany was able to leverage a deviation in so-called standards to set Audi apart from its competitors and project an innovative brand image.
|For more information:|
|James Kalbach is currently head of Information Architecture at Razorfish, Germany and has a masters degree in library and information science. Previously he established a usability lab at I-D Media, a large German digital agency. .|