The Right BI Tool for the Right User

When too many users have too many tools, the result is faulty BI. Heres how to restore order.

Nearly anyone involved in business intelligence technology can recall a horror story that involved giving users too many freedoms with BI tools. Indeed, of any technology that IT is involved with, BI might be the most difficult to deploy when it comes to putting the right tools in the right hands so that the right data is delivered at just the right time.

For instance, three years ago, San Antonio-based Valero Energy Corp. had an array of BI tools because it had gone through multiple mergers and acquisitions. The options included systems from Cognos Inc. and Hyperion Solutions Corp., as well as Crystal Reports, which is now owned by Business Objects SA. A pitfall, says Kirk Hewitt, director of reporting and financial systems at the $82 billion refiner, was that users were running reports off different versions of the data.

Somebody might take data from the data warehouse, load it onto a Crystal Reports server and do their reporting off the server, he says. But they might not do an update every morning, so all the data in that report was not consistent with what was in the warehouse the previous night.

Valero took measures to solve that problem in 2004, when it began a two-year project to consolidate its BI architecture. That effort included standardizing on SAP Business Information Warehouse (BW), SAP ERP software and an Oracle data warehouse. Valero also moved all report- and query-building functions into the IT department and chose one front-end tool Information Builders Inc.s WebFocus as well as the online analytical processing (OLAP) capability in SAP BW.

Now, although users can slice and dice and drill down into data using WebFocus, theyre using reports built by the IT group and comparing data that has been qualified and structured by IT. All reporting originates from information services; we no longer have reporting pockets in the business, Hewitt says. That ensures that were aware of what the requirements are, that the requests make sense within our infrastructure and that the data is available on a timely and accurate basis.

Some of the People, Some of the Time

Valero isnt the only company thats moving from what might be considered the Wild West era of BI to a more controlled and sophisticated approach. As companies realize the potential of BI implementations, theyre also learning how many things can go wrong if they dont standardize tools, work with users on how they want their business intelligence delivered, coordinate with them on creating queries and reports, and ensure that everyone is working off the same set of correctly structured data.

At Del Monte Foods Co., before the company standardized on one set of BI tools, users were free to design their own reports with one of several query and reporting tools, says Andy Woje­wodka, director of business systems and decision support. I found different departments sending performance results to management on fill rate, but each used a different set of business rules and filtering, he says.

For instance, there was no agreement on which types of orders or businesses should be included or excluded. Consequently, all reports were correct, but their definitions of fill rate were completely different, Wojewodka says.

Too much flexibility and ad hoc capabilities in the hands of the wrong person can result in islands of autonomy, homegrown subsystem processes and the proliferation of multiple versions of the truth, Wojewodka concludes. Now, BI analysts in Del Montes IT group work with end users and developers to determine how best to present information in a meaningful form for business owners consumption.

Edward Smith, chief information director at Utz Quality Foods Inc., a snack foods manufacturer in Hanover, Pa., had a similar experience. Weve gone down the road of letting users build their own queries, and it didnt work too well, he says. Inefficient queries were bringing the companys main business systems platform, an IBM iSeries server, to a screeching halt.

Theyre savvy users, but they didnt understand data structures, and they dont know the data well enough to get the right answer, Smith says. Since 2002, he says, the IT group has taken over designing queries and reports using WebFocus, so users can slice and dice data safely, using efficient queries and correct data structures.

Part of the problem, says Bill Hostmann, an analyst at Gartner Inc., is that the reports and dashboards that get delivered to users are just the tip of the iceberg, and whats underneath the data integration, qualification, analysis and formatting is often not adequately funded. Most people dont appreciate how hard it is to get data in an accessible, controlled, qualified form, he says. Those are important and difficult pieces of the puzzle.

Know Your Audience

Reining in report- and query-building is just the beginning of the job, however. Getting the right BI tools into the right hands also requires IT to really get to know the user population and all the different needs to be served.

At Del Monte, some managers prefer performance indicators to be delivered by hand at regularly scheduled meetings with face-to-face dialogue, while other users want the freedom to keep a pulse on whats going on, Wojewodka says. I used to have the misconception that by supplying interactive analytics and reporting and making it intuitive in nature, that it would be embraced by all, he says. But that isnt the case.

Thats why BI analysts not only need to focus on the metrics, reports and analytics, but also need to devote time to understanding the process within a given role, Wojewodka says. This entails spending time with users and understanding the key drivers for their roles. Delivery should be tailored to the users area of responsibility in a format thats actionable for him or her, he says.

Thats exactly how Cindi Howson spends a lot of time: helping companies match tools to individual user segments. Many users want to be empowered and have faster response time to their questions, but they dont want to become experts in sophisticated tools, says Howson, founder of BI system evaluation site BIScorecard.com and author of Successful Business Intelligence: Secrets to Making BI a Killer App (McGraw-Hill Osborne Media, 2007).

So while IT report developers need tools such as Microsoft Reporting Services and Crystal Reports, business power users would get more benefit from business query tools, she says. Sophisticated Excel users might want BI delivered in an Excel interface, while casual users might want portal-based BI or data delivered via e-mail or in interactive reports.

Higher up in the organization, executives and other managers most likely want a BI dashboard or scorecard. The information has to be relevant and personalized to the person accessing it, Howson says.

At Valero, some users want their BI delivered in PDF files, while those who want to manipulate the data need Excel spreadsheets, Hewitt says. Still others want reports to be accessible through a Web portal or delivered via e-mail.

In IS, we understand what the problems and issues are because, in many cases, we came from the business, Hewitt says. For instance, most of the service managers who collect reporting requirements originally worked on the business side of Valero.

We Have the Data, Now What?

Still another challenge is ensuring that people know what to do with the data once they can access it. Dont assume, If we build it, they will come, says Wojewodka. People need to understand how to utilize the information.

People might have access to data, but they still dont act on it, Howson says. Companies need to get away from gut-feel decision-making and foster a culture that supports What is the data telling me? versus What is my bias telling me?

Utz has successfully established a data-driven culture. Sales managers, for instance, are highly aware of current sales in comparison with same-week, year-ago sales a key retail metric. So when a sales manager queries the data and sees that regional sales are down 10%, he can quickly find out why, Smith says.

It generates a whole host of questions and gets people to react maybe a route salesperson was out sick, but why didnt someone pick up that route? he says. Theyre accustomed to looking at the data every day, so we can react to business failures and fix them.

But whether youre transitioning to a data-driven culture or moving away from chaotic and uncontrolled query- and report-building, you cant overlook the difficulty of change management, Hewitt says. At Valero, it couldnt have happened without the support of top executives like the comptroller and the CIO.

They really drove the change management of the project, getting us the support we needed so that people would participate, Hewitt says. Obviously, a lot of end users were affected, and not everyone wants to change.

Hewitt also made sure that end users were involved with the final selection of the BI tool. They got to try it out and asked a lot of questions: Can you get it in Excel? Deliver it through e-mail? Go to the portal? Do OLAP? Everyone had different requirements, he says.

Making the commitment in time, money and change management necessary to get reporting right is one of the biggest problems in BI today, according to Hewitt. But its worth it, he says. At Valero, the change has resulted in users being more confident in what the tools are telling them. By having a centralized and consistent data source, Hewitt says, were able to provide more people with access to the tools.

Brandel is a Computerworld contributing writer. Contact her at marybrandel@verizon.net.

Copyright © 2007 IDG Communications, Inc.

7 inconvenient truths about the hybrid work trend
Shop Tech Products at Amazon