Taking the Time to Get it Right

by Katie Haegele

From Internet Publishing, February 2001

Potentially expensive and time-consuming, implementing content management technology is a big step for any company. Nowhere was this more true than at Congressional Quarterly, a multi-million-dollar Web service and publisher.

Congressional Quarterly (CQ) is a 55-year-old "legislative tracking service" that provides news and information from Capitol Hill to trade associations, lawyers, lobbyists, the government itself, and industries affected by legislation. Subscribers rely on CQ's 15 print and online publications for all information from the Hill—from article-length analyses of bills and laws to fielded or "relational" content, such as databases with voting tallies of all senators and representatives.

Its first Web product went live in 1995, and CQ immediately saw tremendous business opportunities to provide even deeper information to subscribers. "With the Web's ability to do technical things like cross-referencing, we needed a much more robust system, " said Larry Tunks, CQ's chief information officer.

The publisher's primary goal, however, was to create a central repository of news and content that would be multi-purposed for both print and online publications—without interrupting business as usual in the editorial department.

An off-the-shelf solution would not have been comprehensive enough for CQ's complex needs. Today, all CQ content is published from a custom solution—the Unified Data Repository (UDR)—devised by a team led by Lisa Bos, executive vice president of consulting services and co-founder of Really Strategies. Bos was with Reed Technologies (now Thomas Technology Solutions) when she helped CQ implement the project.

CQ faced several issues before building a solution. The existing production system was not Y2K-compliant and actually supplied data to an online service that predated the Web. It was easier for the company to scrap the VAX-based system and start over. But before CQ could think about giving its editors new ways to prepare content, it had to organize the content that already existed.

"CQ had data in multiple databases and various desktop formats. In some cases data overlapped and had to be maintained in more than one place. All of this reduced the ability to create new products quickly for the Web and print," explained Bos. "Lots and lots of content was stored and managed in lots and lots of different ways."

The team used this to the publisher's advantage. The new system's functions were divided along those same editorial lines and developed for one group at a time. Although the system took 18 months to fully implement, CQ was using portions of it after a year.

The first step—a pretty enormous task," recalled Bos—was to take all existing content and dump it into one central repository (UDR), removing any redundancies that existed between databases. This task was handled entirely by Bos' team.

The next hurdle was to ensure that existing systems would work with the UDR. This was especially critical for CQ.com, a subscriber-based online product updated with real-time Congressional data. Reporters worked on the Quark Publishing System in a Mac environment, and CQ was wary of pulling them off a system they were already comfortable using.

"We decided to keep them on QPS, with just a few modifications that limited interaction with the new system and required some training," Bos said.

CQ editors still create stories in QPS, but they also define certain QPS fields that are useful for searching and linking in the UDR and on the Web. When copy is ready to go online, it is marked as ready to publish, and the system saves that file to a directory where another process automatically converts it to XML (Extensible Mark-up Language).

The XML content, which is then loaded into the UDR, includes the stories, complex in-line data such as links to a particular bill or committee, and information that relates the story to other information in the UDR. XML enabled all this information to be captured together in a structured, consistent format, thereby increasing the system's reliability and accuracy.

In a similar way, the system automatically converts UDR data to XML for deliver to the Web. Once stories or other content are published to the Web, the links automatically go live, and the relationships to other content can be used for sophisticated searching, navigation, and automatic e-mail notifications. The combination of XML and databased content enables site users "to do all sorts of complex queries," explained Bos.

The UDR is sophisticated, yet streamlined into a centralized interface. "There are many different modules to it, but they all operate on the same paradigm and they all go through the same front porch," Tunks said. The friendly interface "looks very much like a Microsoft Outlook application, with icons down the left side that you simply punch to go into a specific work area," he said.

Where does the client/vendor relationship stand now? Although CQ initially intended to build an IT staff solely of in-house development of the system, Tunks said his priority is now Web site development. He handles some time-critical production issues internally and leaves the rest up to Really Strategies.

"As a CIO in today's world, you absolutely cannot function by doing either everything out of house or everything in-house," he explained. "It has to be a combination of your team and partnering vendors who can integrate into your environment."