One goal of the case study was to establish an estimate of baseline costs for Web site operations. This was complicated by the fact that the CTG Web site is continuously growing with the addition of new and different types of information. The estimated costs for maintenance of the current site (at the start of the study period) were approximately 75 percent of one full-time staff person’s effort per week. This was treated as stable over the term of the case study.

The case data also included estimates of the effort devoted by staff members to researching XML (dynamic Web page design), which began several months prior to its initial use by the Technology Unit staff. The director of the Technology Unit had been contemplating moving to XML for quite awhile. This early deliberation was prompted by the rate of growth for maintenance and development efforts on the CTG Web site. The effort devoted to this period of informal research and deliberation was considered part of the initial investment.

Baseline costs. The Center’s Web site is very extensive, consisting of approximately 3500 Web pages and receiving on average over 1000 visitors per day. Information is continuously updated and new products and reports are introduced frequently. Maintaining the Center’s Web site is quite an undertaking. Under the old HTML architecture, a Webmaster was responsible for maintaining site links, updating current materials, adding announcements and new materials, and related tasks. A full-time staff person devoted approximately 75 percent of their effort to maintaining that site (considered nine person-months per year.)

Investment costs.The primary investment cost for this project was staff time to learn the software and develop the skills necessary to build the pilot projects. The software costs for the conversion might have been quite high, since initial research showed most commercial products to be very expensive. However, after a review of available tools the Technology Unit decided on the open source (free) application, Cocoon. This tool seemed to fit the needs of CTG and the staff began working with it.

While the software was free, it required the staff to learn new skills. The underlying logic of how the Web site operates was also considerably different under the new architecture. So the development team had to learn to think about its tasks in a new way as well. By contrast, there were zero learning costs for maintaining the Web site with the old architecture and methods.

The initial learning process was a slow one. It took approximately three months for the staff to reach a level of skill necessary to move from using HTML to using XML. At first, as much as 80 percent of their time was spent on learning the XML application and related skills, and 20 percent on production. After a 10-week period, the time devoted to learning dropped to approximately 40 percent of the total, with the remaining 60 percent used in production. Three months into the effort, 20 percent of staff time went to learning and 80 percent to production. The staff found that this latter 1:4 learning-to- production ratio is consistent with the learning-to-production ratio under HTML (prior to the switch to the new technology).

The learning took place in two stages. In the first stage, the team shifted its way of thinking about Web development from the HTML to the XML architecture. The second stage was to learn the technical details of working in XML. After the initial three-month learning stage, staff members considered themselves knowledgeable enough about the language of XML to deal with problems quickly. The second stage included learning to use the Cocoon development tool. Since documentation was limited, they used the Cocoon listserv for questioning others who use the program. Much of this learning came from applying the new tools to the tasks specific to the pilot projects. This included creating interactive tools that allow users to work with applications on the Web site. The team’s learning process was focused and task-specific, the results of which could be applied immediately. The staff avoided exploring unneeded features and functionality of Cocoon.

Additional learning costs are reflected in the different amounts of time needed to revise the printed version versus the electronic version of the Gateways decision making guide. The first pilot application developed with the XML architecture was based on that guide. Initial changes to the electronic version of the guide took up to 50 percent longer than print-only revisions because of the time spent learning to apply the new technology in the Web application. Initially the staff found it difficult to use XML data structures. As a result it took considerable time to understand basic concepts and acquire basic skills for solutions that at first often appeared to be quite simple.