Instead, the integrated systems automatically resolve the problem and process the claim in about two minutes without staff involvement. Since July of 2008, the department has saved approximately 1,750 hours of staff time, says Pat Lashore, administrator for the department's technology services division, and taxpayers who are due a refund now receive it more quickly.
In a similar vein, Allstate has had success pushing report-creation and customization capabilities out to end users through the deployment of dashboards. Previously, the company had a centralized report-writing function within IT, and "it took a lot longer to get answers into the hands of business people," Abbattista says. Now, his team creates dashboards, walks users through the basics of using the tool, and lets them do the rest.
One goal of Allstate's is to build models very quickly with the tools they already have to prove or disprove whether a given hypothesis is any good, he says. If it is, the model will be incorporated in a dashboard and disseminated to staff.
Allstate has institutionalized dashboards to monitor its speed in closing claims and the time to resolve those claims. It wants to drive those numbers down to the point where its performance in these areas becomes a competitive advantage. The self-service environment provided by the dashboards allows managers to monitor their department's performance. They also can add or subtract factors or combine data in different ways to achieve those goals.
Back in the IT department, the self-service BI tools helped Abbattista's team get out of the report-building business and clear out a long backlog of report requests. Through the self-service initiative and a data-warehouse consolidation, he's reduced headcount by two thirds while expanding access to self-service BI tools to 25,000 users within the organization.
Keep your models clean
Make sure you have a clear and consistent data model before you bring new data into your data warehouse or bring in data from another part of the business, and then ensure the new data conforms to your model.
Too often, says Millman, information from different sources or sections of the business gets added to the data warehouse without enough attention to how the existing data is modeled. The result: "It's hard to make sense out of business reports or queries that go across more than one section." For example, financial and customer service data might be modeled in entirely different ways. Accenture, he says, spends a lot of time helping its clients re-architect the way their data is stored.
For Abbattista, re-architecting data also meant rationalizing it for different business uses as new data sources were added. For example, at Allstate, different definitions of the "policy effective date" had to be reconciled before data from different departments could be combined into a single data warehouse for analysis.
IT may also need to allay fears about data quality. If managers don't trust the quality of the underlying data, that can derail their interest in business intelligence projects before they even get started. Mistrust "is like a silent cancer in organizations," says Milley. "Companies are at the mercy of their data quality." Users may suspect that the data is old, or that too many records may be missing data in a given field, such as birth date, she says.
"A common complaint is that the quality of the data isn't good enough," even when it is, agrees Millman. To drive up use of existing BI tools, he recommends producing data-quality dashboards that show just how timely the data is. "It's about crystallizing how good the data quality is and making that visible to the business."
Help users understand the data, not just the tools
Scaling up the number of users who have access to BI tools won't help unless people know how to use them. But that's not the biggest issue when it comes to educating the user. "The trend has been for the front end to get simpler and more intuitive," Millman says. And certainly dashboards have helped in that regard.
"What's often missing is the explanation of where the data comes from and how you can use it to derive some insight," Millman says. For example, the data generated by Creativity's CubiScan system was foreign to business people in the back office. "We have to explain what the data points are and what the data points mean," Mulholland says.
Abbattista focuses on building that knowledge one user at a time. "We build out initial capabilities with front-line managers and people in the trenches. They then become the consultant to people around them."
"It's really [about] teaching people to mine for value," Abbattista sums up. In that respect, he says, "I don't think we'll ever be done with our BI efforts."