Faceted Approach to Web Redesign

A site´s information architecture is not usually the first thing that comes to the minds of web developers. Most of the time it is only after a site is built that the deficiencies are noted and solutions are sought. At that point an assessment of the information architecture becomes a task that should be considered.

The discipline of information architecture is relatively new, but it draws on many age-old concepts from library science, multimedia application construction, and information design. The idea of facets as a way of classifying information is an example of a very old concept from library science, which is more relevant than ever in an automated environment like the Web.

Facets are properties of a subject that are clearly defined and mutually exclusive. They can be contrasted to hierarchical parent-child ways of categorizing information; for example the way files are typically organized in a computer file system. Facets have more in common with the way a relational database treats its data.

Naturally, some problems are best solved by a hierarchical approach to navigation. In fact, without the power of database-driven technology, facets are difficult to implement as web site navigation. But in many cases, especially as web-publishing systems get more sophisticated and data-driven, facets are an invaluable tool.

The web content becomes more scalable and manageable by using facets and storing data in a relational database. Users of the site benefit from this approach, since it is a more friendly-intuitive approach to navigating content. Data driven aspect also helps the site designers and managers in maintaining a fresh content that is easily updateable without impacting the user interface.

Redesign Scenario
To illustrate the use of facets, we will draw upon our experiences developing a portal for a state government. A section of the site devoted to information about starting a business in that state needed re-organization to be more citizen-centric than bureaucracy-centric.

Several state agencies, and departments within those agencies were involved in helping citizens start a business, and they all were content contributors. It made sense to re-organize the information from the citizen´s point-of-view and store it in a database, to make the diverse information sources easier to maintain and manage. The old content had been gathered and presented with no eye towards the overall information architecture–instead it was based on good sense following web standards.

The challenges we were facing could be characterized as the following:

Labeling

There was no consistent way of referring to common concepts in the site. Labels were created on an ad hoc basis, however the content experts were accustomed to labeling a concept. This led to many labels being inconsistent and not particularly geared towards a citizenï¿½s point of view. There was also a problem matching page titles to the text used to link to those pages.
Navigation

The general navigation scheme of the overall citizen portal dictated the first level of sub-site´s navigation, as well as the general navigational style we had to work with.

On one hand, this led to a good consistent navigation scheme throughout the portal. On the other hand, it was a constraint for our process. There was little room for introducing new navigational techniques.

Analysis of what the navigation was able to deliver to users revealed that there were many shortcomings. Users were not able to tell what page they came from after their second “click” on the site. Navigation within sub-pages and major portal pages was intermixed, making going back to the sub-topic navigation a confusing ordeal with very little intuitiveness.

Document Structure

Content analysis showed us how some pages were aggregates: containing related but different types of links, topics, and information. We called these types of pages “Frankenstein pages”. They were born out of pure expediency, basically just to stuff the information somewhere in a web page it most related.

Our basic approach to solving these problems revolved around a series of abstractions that allowed us to rearrange our content while leaving behind troublesome labels and document structures. These abstractions also helped us make specific recommendations about how to reform elements of the whole portal´s navigation system.

1.1. Information models & content maps
We conducted a content inventory, where we went through each page and recorded some metadata about the content, audience, and structure of the page. The inventory showed the siteï¿½s poor organization, unnecessary duplication of information, missing information as well as some disconnection between relevant information. Yet, by analyzing the results of our content inventory, were able to develop an information model.

The granularity and logical abstraction of a content inventory allows us to ignore navigation, redundancy in content, and problems with document structure (“Frankenstein pages”). An information model is created by organizing the various topics, content, and audiences discovered in the inventory into types. We decided to document this model in a diagram, although it can also be captured in an outline.

Once an information model has been discovered, it can transform the content inventory into a content map. The content map represents the whole of the site´s implicit organization. It is basically the underlying information architecture of the current site, however haphazard, which may or may not have been reflected in the site´s original navigation and labeling. This map can be compared to business goals and users Mental Models to discover defects and deficits in a site content.

1.2. Mental Models and Business Needs
The contextual and cognitive situation that frames a userï¿½s interaction with a web site is what we call their MentalModelï?

In our case, because time and money prevented us from doing proper usability tests, we relied on keyword analysis as a substitute. Independent of the content mapping process, we grouped keywords around a set of tasks users might be trying to accomplish by searching on those words. We combined this analysis with previous focus group data to decide on the fundamental tasks that our users were trying to undertake on our site. Such an analysis also reveals something about the way user´s organize terms and concepts in their heads, helping us decide on how to label certain ideas clearly.

The Mental Model analysis revealed a deficit in the content (also known as a content gap). We realized that there were topics and information that was not covered adequately. Although not significant ones, some tasks that users wanted to accomplish were missing as well. This discovery led us to examine the business needs and main goal of the site. By revisiting these needs and comparing them to the Mental Model analysis, we came up with a more solid set of requirements. Once we understood the users’ Mental Model we were able to determine tasks and facets that would fulfill the business requirements.

1.3. Information Architecture & Facets
Information architecture, at a minimum, is used in many different ways, but as we understand it, it can be taken to mean the organization of labels, hyperlinked navigation, and content types in a web-publishing environment.

With a better understanding of both the business requirements of our site and our user´s Mental Model, we could better understand how users related topics together and what labels they might easily recognize. In addition, by articulating the information model and then comparing the current model to business requirements and user feedback, opportunities for navigation became evident which were previously unclear.

We quickly noticed a clear parallel between the topics covered by the site and the types of content clustered under the topics. Each topic had instances of most of the content types, and vice versa. On that basis, we decided to break the content types out as facets of the topics and use these facets as our primary navigation.

This process also manifested in a new and improved navigation scheme that was more intuitive for users. In the end, we accomplished our goal of overcoming the challenges we had identified with the site. Faceted approach has been the answer to our labeling, navigation and document structure problems. We do believe that these are common challenges experienced throughout the Web. Quality user experience is significantly diminished by these factors that can be eradicated by proper information architecture methods such as “Facets.”

Conclusion

In short, “Facets” are not a new idea, and it is getting more prominent use by information architects that discover the value of using facets. We hope that our Redesign Scenario gave you an idea of how facets can help when redesigning web sites. Here are some successful examples of facets in action -in variety of industries-:

[Mass.gov]
[Wine.com]
[Epicurious.com]
[Kohler.com]
[Tower Records.com]

We have found facets to fit very nicely to the scenario we have shared with you. However, not only government but any industry content can utilize facets to the multi-source web content they need to present on a web site. Using facets is particularly useful on database driven on the fly content. Existing web sites that are redesigning or reorganizing their content can truly benefit from the use of facets.