Business intelligence program needs long-term tending to fully blossom

Quicken Loans has a good story to tell about its self-service business intelligence program.

About 2,000 business users -- one-quarter of the company's workforce -- access its self-service BI system on a daily basis. The BI team at the online mortgage lender has built 40 dashboards tailored for different groups of users, as well as 400 reports; in addition, end users have created more than 15,000 customized data views and made 4,000 of them available to fellow workers through an information sharing feature. The system also operates in near real time: BI data typically is updated five to 30 minutes after transactions are processed in source systems.

BI managers from Quicken Loans have given presentations about the self-service initiative at several conferences, including The Data Warehousing Institute's TDWI BI Executive Summit in Las Vegas last month. And the Detroit-based company won TDWI Best Practices awards in both 2010 and 2011; it was recognized for its operational BI capabilities and its data warehouse architecture, respectively.

But the current system didn't come fully formed when Quicken Loans began developing BI applications five years ago -- far from it, in fact. At first, the company built a portal for delivering canned reports to business users. "Over time, we started realizing that wasn't scalable," said Dan Jones, director of BI at the lender. "We wanted to be oriented more around data discovery, as opposed to just building hundreds of static reports." The BI strategy shifted, and the focus has been on dashboards and self-service tools for the past three years.

Jones, who spoke at the TDWI summit, along with co-worker Nuverre Naami, said in a follow-up interview that the company has also upgraded its BI processes significantly since getting started, following a playbook that includes these elements:

Top-down leadership. When the BI team built an initial dashboard showing an end-to-end view of the loan approval and funding process, it demonstrated the tool to Dan Gilbert, the company's founder and chairman. Gilbert liked what he saw and became "a great evangelist for us," Jones said. "He wants his leadership to be able to dive deep and leverage both their business expertise and what the data is telling them."

Agile development. In 2011, Quicken Loans adopted an Agile development methodology in which new BI functionality is rolled out in increments every three weeks. That approach, which includes frequent planning meetings with business users, has paid big dividends, according to Jones. "In the past, we would maybe get overwhelmed by the scope [of projects] at times," he said. "The three-week windows help us to better manage what we're working on."

Dashboard diversity. The BI dashboards available to end users range from simple tools that provide basic views of key business-performance metrics to a power-user dashboard called QSlice that supports data slicing and dicing in hundreds of different permutations. "What we're trying to do here is build tools where maybe it's the same data being presented to users, but in different ways and formats," Jones said.

Visual appeal and usability. Over the past year, the BI team has started paying more attention to the visual aspects of dashboard design in an effort to further boost adoption and use. In addition, BI developers sit in on training sessions to watch how users interact with the self-service BI tools and listen to their questions. "You get a lot of usability insights that you wouldn't get otherwise," Jones said. For example, if users instinctively move their mouse to a spot on a dashboard screen where there's nothing for them to click on, that could drive a design change.

Quicken Loans could be a good model for organizations that are just beginning their own BI programs -- and its experiences certainly should boost their confidence that there's light at the end of the BI tunnel. At the TDWI conference, I talked with several attendees whose companies are in the early days on BI deployments. Chris Wade, manager of technical services at Halogen Software, was one: He said the Ottawa-based vendor of cloud-based talent management tools currently uses "brute force" coding to pull data out of its customer-activity logs. "It's kind of kludged-together," he said. "We're trying to get to where [the data] is more presentable."

For Wade and other BI and IT pros in similar positions, the immediate priority is nailing down the basics of a business intelligence program. Once that's accomplished, the real fun begins. And if all goes well, so do the real business benefits.

Craig Stedman is SearchBusinessAnalytics's executive editor. Email him at [email protected].

Follow SearchBusinessAnalytics on Twitter: @BizAnalyticsTT.