October 30, 2003

The Composite View

How Documentum allayed its data management problem and cleared the way for future growth

by Amit Asaravala

Roy Gum knew that Documentum Inc. was facing a tricky problem. 2001 was fast coming to a close and, by most indications, the content management market was set to boom over the next few years. As a leading solutions vendor in the space, Documentum had already been reaping the benefits. The company's revenues were nearing $250 million annually despite a downturn in the economy, and new leads were coming in everyday it seemed. In the Pleasanton, Calif. office, company executives had even begun to discuss their plans to take revenues past the $500 million "magic number" — a critical step that would signal Documentum's transition from young startup to mature industry leader. On the surface, everything seemed to be on the way up.

Yet, lying underneath all this was an ironic twist: The rise in business was creating a data management nightmare for the IT department. Because the company had no convenient way to synchronize data between its sales and marketing systems, the staff was cleaning and loading the contact data for all leads by hand — a process that could take days, depending on the number of records. The sales team would often have to wait until long after a marketing event to access the information, and some leads were even getting lost due to problems with the way they were routed.

For Gum, who had just joined Documentum as the director of e-business strategy, the problem was clear: If the company didn't figure out how to scale its information systems quickly, its growth would be stunted. It was time to get to work on his first major project for the company.

Weighing the Options

To an outsider, the solution to Documentum's woes might seem simple: Just set up a nightly export-import cycle between MarketFirst and Siebel so that the data would be synchronized each morning. Unfortunately, differing data models made the process more difficult than it sounds. In addition to mapping one set of fields to another, the systems needed to know how to handle duplicate or missing information. To make matters worse, the Documentum staff found that the way Siebel 6.2 handled updates to existing contact data was unacceptable for its needs. "If there's an existing record, any updates will overwrite the old information," explains Gum. "You can't merge the data."

Of course, even if Gum and his team could have found a way to overcome all these obstacles, they would still need to construct a method to weed out inappropriate data. Anyone who has ever filled out an online form knows how easy it is to enter false or incorrect data. "Weeding out the bad records is important," says Gum. "If I hand off bad data to the sales team, I'm wasting their time."

Gum and his team first approached Siebel Systems to see if the software vendor had an existing solution that they could use. The response wasn't what they'd hoped for. According to Gum, the primary option was to use Siebel's Universal Application Network. But so-called extract, transform, and load (ETL) solutions don't traditionally offer the capability to evaluate data in real time — a requirement that Gum felt was important in helping Documentum grow its customer management capabilities.

The second option was to migrate out of MarketFirst and into Siebel's marketing automation suite. Again, the solution had its drawbacks. For one, the process of importing MarketFirst data into a Siebel product seemed too similar to the very problem they were trying to solve. Additionally, the solution would require Documentum to train its marketing staff how to use the new technology. Gum knew there had to be a better solution.

"We spent a year looking at the various options," recalls Gum. "We looked into building a solution ourselves, but we knew we'd be writing code for a while and that it would be a nightmare to manage. So we said, 'Nah, we don't want to do that.' We thought about doing an all-out EAI implementation with all our systems, but we didn't really want to ship data around the enterprise. We wanted a real-time view of our data, and we knew that we didn't want to spend $5 million to get it."

A Solution at Last

When Gum finally found the solution he was looking for, it was in an unlikely place. "I actually ran into a former colleague at Oakland International Airport who was working for Siperian Corp.," recalls Gum. "He knew that I was trying to solve this same problem when I was at KANA Inc.," (a CRM solutions firm where Gum had worked previously). "He told me that I should look into Siperian."

Founded in 2001, Siperian is a relatively new face in the data integration industry. The company focuses exclusively on solving the customer data integration problem, which it believes will plague more and more companies in the coming years. Siperian's Activity Server solution helps companies organize all their data across the enterprise and then act on that data according to explicit business rules. The solution manages to combine some of the best features of the various existing data integration techniques, and it does so in a way that has analysts watching the company closely.

• Privacy
• Your California Privacy Rights
• Copyright © 2008 United Business Media LLC