of the American Society for Information Science and Technology          Vol. 28, No. 5         June / July 2002


Go to
Bulletin Index

bookstore2Go to the ASIST Bookstore



Redesigning an E-Business Taxonomy: Egreetings Project Case Study

by Chris Farnum

Chris Farnum is with Compuware Corporation and can be reached by e-mail at crfarnum@yahoo.com

This case study was presented at the ASIST Information Architecture Summit 2002 in Baltimore. Please see http://home.earthlink.net/~crfarnum if you wish to download the full presentation. The following article focuses on the first part of the project, which involved user research to make recommendations on revising the site's controlled vocabularies.

As a consultant with Argus Associates in 2000, months before the full impact of the dotcom avalanche was felt, I had the chance to work on a project that would make any information architect salivate. Besides the fun of visiting San Francisco and getting to know a truly hip e-commerce company from the inside, I got to help Egreetings (http://www.Egreetings.com) with a redesign effort to improve the number of online greetings sent by users. At the time, Egreetings was generally ranked third among online greeting sites (behind Blue Mountain and American Greetings). The company's core content was its collection of online greeting cards comprising flash animations, animated GIF images and still images. Some were created in-house by a highly talented team of graphic design artists, and the rest, such as cartoons from the New York Times, were licensed art from outside sources. Anywhere from 40,000 to 200,000 cards were sent each day depending on the season. Argus was called in by Tim Scheele, the senior director of publishing, to help with a number of goals:

Increase card sending statistics

Reorganize the card collection (taxonomy/controlled vocabulary)

Improve navigation and searching

Suggest key places for ads and promotions (need to "monetize" )

Find an approach for music greeting collection

Improve the checkout process

Challenges and Approach

The team consisted of four Argonauts: a lead information architect (myself), an assisting information architect (Michele de la Iglesia), a project manager (Shawn Stemen) and a usability specialist who advised us (Keith Instone). We faced a number of challenges during the project. We needed to influence a very fast moving organization. Egreetings truly operated on "Internet time," and could implement suggestions very quickly. On the other hand, it was a challenge to keep up with them and to ask them to take time to wait for the results of our research and design efforts before jumping to a decision. The project was deadline driven. We began work on our strategy and recommendations in Spring 2000 with the target of a Fall 2000 re-launch. Another issue was that it was important for us to reconcile need to "merchandise" with need to organize and index content from an information science perspective.

An information architect's approach should involve an investigation of the content, the organizational context and the users. Often the user research part of the methodology gets less emphasis than it deserves because of time and budget constraints. However, this project incorporated user testing and research during each phase. Therefore, it was a wonderful opportunity for me as an information architect to learn how to include usability testing practices such as card sorting into my methodology.

Investigating the Taxonomy

The Egreetings site had always relied on a taxonomy of categories for its online greetings. The term taxonomy is used somewhat loosely here, because the site does not obey the rules of a classic taxonomy, which places all items into one and only one node. Egreetings approached the arrangement of cards as merchandising, rather than classification, so they often placed items in more than one place. The categories and sub-categories of cards on the site are analogous to the aisles and placards in a traditional paper greeting card store. The organization scheme is based on a combination of logical arrangement and marketing goals. For example, both Egreetings and a traditional store would need to feature seasonal holiday cards. One of the main problems with the site, which the company identified by analyzing the server logs, was that many users abandoned the site without selecting a card to send. They decided to call in the information architects to help make it easier for users to find cards by improving the way the content would be organized by categories.

There were a number of steps we took before any kind of face-to-face user research. We examined the existing taxonomy and explored the site. Next we did a comparison with competitor sites' categories. At this point we already knew that there was too much overlap among the meanings of the categories on the Egreetings site and also that many cards were being placed into too many categories. The meanings of the categories in the taxonomy were diluted. We saw that the taxonomy was a combination of many different types of organization schemes with no coherent overall theme.

We also performed content analysis by closely examining a representative sample of about 100 cards sent to us on CD-ROM. For each we assigned relevant indexing keyword terms that came to mind such as birthday, dog or humorous. The goal in this exercise was to identify as many possible distinct facets of the cards as we could. Some of the ones that we found included "reason to send" (or occasion), recipient, sender, gender, format (animated vs. still image), card text message, card image content (birthday cakes, babies, baseball) and emotions. At the end of this round of research we knew that we had several key design questions:

What facets are most important to users as they browse to find the category containing the card that matches their needs?

Which overall organization scheme should be presented to users on the main page of the site?

What kinds of terms and labels are most meaningful for users?

Card Sorting to Learn about Facets

To find answers to these questions we arranged to have one-on-one sessions with approximately 24 users in Michigan and California. We had the luxury of being able to hire a market research firm to provide the testing facility complete with one-way mirror and video recording. This firm also helped us by preparing user screening guidelines based on information about Egreetings' target demographics and by recruiting the users via phone. We chose to include a card sorting activity as part of the interviews. Card sorting is a tried and traditional user-testing method to explore ways to organize a set of content. It's also helpful for determining which facet or organization method should be the primary scheme for a site. In our version of this test we first asked users to look at a sample of 10 to 12 cards on a computer screen. It was crucial to show them the actual electronic versions prior to sorting because we wanted users to get the full audiovisual effect of the cards, especially the animated ones. As a warm-up we showed a sample card to the person and briefly pointed out that each card could be described according to many different facets.

Then as we showed each card on the screen to the person we set the following tasks:

Pick three keywords you'd attach to this card that describe important features - the ones that stand out most to you.

Describe the tone of this card.

Tell us what kinds of recipients you think this card was intended for.

Tell us for what reasons someone would send this card.

Our plan was to make the test subjects familiar with the cards and to get them to apply indexing keyword terms. We already had a strong hunch that the tone/emotion, recipient and reason to send would be important facets, which is why we followed up the open-ended questions with the three directed ones. Next, we handed the subjects color printouts of the greeting cards and told them to place them on the table. After viewing all of the cards we asked the users to arrange the paper copies into groups, and then we provided Post-It notes so that they could give each group a title. After the sessions were complete we analyzed the keywords respondents used to describe the cards. We examined each keyword and tried to determine which of the facets it represented. Then we made a tally of the facets most commonly used. We also analyzed the pairings of cards within the groups that users created by counting the number of times that any two cards appeared in the same group.

Our final step was to represent graphically the strongest associations among the cards by creating an affinity diagram and then examining how our clusters compared with the group names assigned by the users.

Our findings provided us with the confidence to make some important design recommendations. For example, one of our biggest surprises was that the image content facet was less important than anticipated, while emotion or tone was one of the first things that users noticed about a card. When users first saw cards the first words out of their mouths involved emotions the majority of the time. Most of the time they seemed to add up all of the audiovisual components and immediately assign concepts such as "happy" or "funny" or "sincere." However, when they had difficulty determining the emotion conveyed by a card, they often commented that they did not like it. We concluded that the emotion facet is one of the most important factors for users when they are deciding whether or not to send a particular card. In addition we found that the format of the card was also an important part of people's card selection decision. They also alerted us to some distinctions in format that we had not previously considered. For instance they differentiated between still image cards that contained cartoon illustrations versus those that contained photography. Despite the finding from the keyword exercise, we concluded that the site's taxonomy should not be arranged from the top down by emotion or format. The results of the card sorting analysis showed us that users have a mental model of the organization of cards that is strongly tied to the "reason to send," whether it is a birthday or just to say "hello" to a friend. This "reason to send" facet manifested clearly in our cluster analysis. The Egreetings site was already largely organized according to this scheme, but our findings helped to validate the approach and gave us the insights we needed to make improvements.

Based on these findings we made a number of recommendations to conclude the strategy and recommendations phase. We drafted a revision of the top level for the site's taxonomy and recommended that they make it more consistent by focusing on "reason to send." We suggested that the facets "recipient" and "card image content" be used as a means to systematically subdivide lower level categories in the site. The "emotion" and "format" were to be used as additional metadata indexing elements separate from the "reason to send" taxonomy so that users could filter, narrow and search according to these secondary facets.

Furthermore, we encouraged them to create controlled vocabularies for these facets so the cards could be consistently indexed. We also delivered wireframes at this point, including one for the new main page of the site to show how to integrate our taxonomy suggestions into the site.

Unfortunately the filtering and searching recommendations were never implemented, but our recommendations led directly to a next phase of our project in which we completely reworked the Egreetings taxonomy and provided indexing and classification guidelines. Our work was really just beginning when we delivered our initial recommendations to Egreetings in the spring of 2000. Eventually they resulted in the re-launch of the site in October 2000 in time for Halloween. Whenever a website has a major makeover, it is normal to expect that users will encounter difficulties and some traffic loss will happen. In this case, traffic statistics actually improved as well as the overall ratio of cards sent per visitor. I am certain that the time and effort we spent in learning about how the site's users think about online greetings was a key factor in this successful outcome.

Further Reading

Bailey, S. (2002). Do you need a taxonomy strategy? Knowledge Management, 5(5).

Toub, S. (2000). Evaluating information architecture. Argus Center for Information Architecture. Viewed 4/10/02. http://argus-acia.com/white_papers/evaluating_ia.html.

IBM. (2002). EZSort and EZCalc  (Software for automating card sorting. Download includes white paper on prototyping.) www-3.ibm.com/ibm/easy/eou_ext.nsf/Publish/410

Rosenfeld, L. & Morville, P. (1998). Information architecture for the World Wide Web. Sebastopol, CA:  O'Reilly.

How to Order

ASIST Home Page

American Society for Information Science and Technology
8555 16th Street, Suite 850, Silver Spring, Maryland 20910, USA
Tel. 301-495-0900, Fax: 301-495-0810 | E-mail:

Copyright © 2002, American Society for Information Science and Technology