Start small.” It’s the standard advice offered for big IT initiatives, the assumption being that incremental successes will lead more business units to buy into technologies that have proved wise investments for other departments.

What these advisers often fail to mention is that enterprises may already have several groups running around making overlapping “small investments.” That’s how many business intelligence deployments gain steam: A competitor launches a loyalty campaign, so your sales group reacts by commencing a BI project to initiate its own program. Next, manufacturing catches wind of a competitor’s supply-chain project, and before long it’s purchased a data warehouse and optimization application. Toss in an acquisition or two, and the collection of “small” BI approaches and technologies can get really messy.

It’s encouraging that more than one-quarter of the 358 business technology professionals we polled for this InformationWeek Special Report say they’ve managed to centralize BI planning and standardize practices, capabilities, and technol ogies. However, a significant number of respondents, 48%, report their companies are just starting to centralize and standardize BI, and 20% say they’re still stuck with “decentralized deployments with little or no coordination of practices and investments.”

What’s holding these companies back, and how can they mature toward strategic use of business intelligence? Our survey shows that better business performance goes hand in hand with pervasive use of BI. At the cutting edge, companies are getting more competitive with the aid of low-latency data access and near-real-time analysis, but today you need more than broad deployments or innovative technologies. Our survey showed what successful companies are doing differently (or more of). The consistent theme? BI must bring insight to the right people at the right time, but it can’t overwhelm, it can’t be too open to interpretation, and it has to help employees do the right thing.

Broader Is Better
The most telling analysis from our survey takes the 19% of respondents who rank their companies as “very successful” in supporting improved business performance with BI and compares them with the 81% who report less success. Just under two-thirds of the successful respondents classify their BI deployments as “mature” or “centralized,” whereas three-quarters of less-successful respondents report their deployments are either decentralized or just moving toward centralization. Successful companies are also far more likely to report “pervasive” or “fairly broad” BI adoption.

The differences are dramatic, but we weren’t surprised to see success correlate with maturity and breadth of deployment. BI benefits compound as orgs gain experience and spread better decision-making enterprise-wide. That’s been the experience at the Transportation Security Administration, a very distributed organization that’s responsible for passenger screening and security operations at 450 airports.

If you traveled by air during the early days of TSA, which was quickly formed in the wake of the 9/11 terrorist attacks, you likely experienced the long lines that led the agency to seek a better way to measure and manage passenger screening. The Performance Information Management System (PIMS) project, announced in late 2004, was TSA’s plan to help thousands of airport managers anticipate passenger volumes while also tracking trends related to screeners (attrition, absenteeism, overtime, injuries), dangerous items, and total passenger throughput. Built on BI software from MicroStrategy, PIMS was rolled out in 2005 and gradually expanded to reach more than 12,000 employees.

Some 2,500 power users check Web-based dashboards every day, gauging key metrics and drilling down into detailed data in the PIMS warehouse. More than 9,500 casual users receive weekly reports in PDF, Microsoft Excel, or HTML formats sent by e-mail or downloaded from an internal Microsoft SharePoint site. PIMS tracks more than 1,000 metrics in total, but TSA has customized dashboards and reports for approximately 30 roles within the agency, including top executives, optimization teams (they’re the BI power users), and front-line security personnel.

TSA puts the cost reductions and avoidance at $100 million for fiscal year 2007 and 2008, most of it coming from year-over-year decreases in overtime, severity of injuries, attrition, and unscheduled absences.
Passenger wait times also are down. There’s even a public-facing Web site (waittime.tsa.dhs.gov) designed to let passengers check the latest statistic trends if they want to anticipate screening wait times by airport, day of the week, hour, and gate.

Avoid Spreadsheet Overload
To promote the sort of broad adoption that benefitted TSA, BI vendors have been talking up “pervasive BI” and “operational BI” in recent years. Yet these buzz phrases fly in the face of ongoing complaints about their software; in fact, “complexity of tools and interfaces” and “cost of software and user licenses” were ranked first and second by survey respondents as impediments to success.

Vendors have responded to demands to make BI more accessible and user-friendly with Web-based options, including visual dashboards and key performance indicators, that can be embedded within portals or applications. But when it comes to grassroots use of BI, no option has been more deployed than integration with desktop applications, with Microsoft Excel usually the tool of choice. More than half of all respondents, 57%, say they’ve implemented desktop integration, with 70% of “very successful” companies having done so.

Desktop integration projects are typically one-way, supporting automated and on-demand delivery of pre-set metrics, alerts, and reports. Those can even include raw data extracts from a central repository, generally a data mart or warehouse but sometimes an operational data store or system of record, such as ERP.
Truly interactive tools have been slow to come. Within the last year, vendors including Microsoft and SAP’s Business Objects unit have added two-way integration functions that let authorized users edit or revise information in Excel and then update the central repository. The latest integration advances also allow for refreshing data in linked Excel-, Word-, and PowerPoint-based reports and presentations when they’re opened, so users don’t have to manually update their documents. Oracle became the latest to add this capability with a BI platform upgrade announced in July.

But the real art in exploiting desktop integration, as well as dashboards, is coming up with user-friendly presentations that help people make better decisions. At Esselte, a $1 billion-a-year manufacturer of office supplies, that meant creating visual, dashboard-like presentations formatted for Excel. “The last thing we wanted was for people to see spreadsheets with tons of data,” says Luis Becerril, a director who oversees reporting at the company’s plants and distribution centers.

Esselte pulls data from more than 300 legacy-system reports using mining software from DataWatch that then delivers, via e-mail and the Web, customized Excel-based charts and dashboards geared to specific roles. Becerril says standardization eliminates the problem of “people developing unique reports and coming up with different conclusions in different plants.”

It’s The Decisions, Stupid
Complicated interfaces and data latency be damned—visionaries, and some vendors, increasingly are saying we should look beyond BI to automating decisions.
That’s an ambitious vision, but since last year’s mega acquisitions of BI vendors Hyperion, Business Objects, and Cognos, acquirers Oracle, SAP, and IBM, respectively, have been tossing around buzz phrases like “contextual BI,” which translates to embedding intelligence into day-to-day workspaces and applications. Most large BI vendors have introduced services-based approaches to delivering alerts, metrics, and key performance indicators within business applications and portals, but our survey shows that only 32% of all respondents are using them, versus 62% of “very successful” respondents.

The most sophisticated services-based deployments are going so far as to automate decisions, but more often the advance is the development of a single, intelligent interface through which employees understand what’s going on and can use that to decide the appropriate action. Contextual BI has been around for years in the form of “smart” or “analytic” business applications, such as customer relationship and supply chain management tools with embedded reporting and analysis capabilities. Now the approach is showing up in SaaS applications from providers including Salesforce.com, Workday, Oco, and LucidEra.

One appeal of these smarter applications is that they cut the information management and data warehousing burden, so look for these options to multiply. “Within three to five years, there will not be an application on the face of the planet that does not have embedded BI,” predicts Marge Breya, executive VP and general manager at SAP’s Business Objects unit. “The question is, what do you do when you have to look at information outside of the application?”

That’s a not-so-veiled call for a centralized warehouse, but even the largest companies with extensive IT capabilities sometimes favor a simpler, application-oriented approach. That’s the case for one survey respondent, an enterprise architect working for a major defense contractor who says his company has embedded analytic capabilities within a CRM application.

“The classic BI tools demand data integration, aggregation, and mining capabilities that we’d love to have once our organization is ready to spend that kind of money,” he says. “But we’ve looked at the requirements people have for data analysis, and we’ve implemented those in custom applications tailored from [off-the-shelf] software.”

The obstacles to conventional BI at this company are more than just money; the architect also points to cultural challenges: “Folks want to spend money within their own departments, rather than a cross-departmental warehouse.” And there are also timing problems. “Our CRM is at a particular stage in its life cycle, and that bears no resemblance to the maturity of our finance and supply chain systems,” he says.

Manage Business Performance
Performance management has long been closely associated with BI, and it’s getting even more attention in the wake of last year’s BI megadeals. This summer Oracle and SAP both made major announcements about their progress in integrating Hyperion and Business Objects, respectively, and much of their attention is focused on new or upgraded performance management applications, including strategy management; profitability and cost management; and governance, risk, and compliance applications.

Performance management applications are designed to not just provide insight—the part powered by conventional BI—but also to help people take action to improve the performance of the business. Finance departments usually take the lead in implementing apps around planning, budgeting, and consolidation, also known as financial performance management, but as in BI, many companies are pushing into operational areas. Whether in the financial realm or otherwise, the objective is to connect people, business processes, and technologies to big-picture strategic goals.

At HSBC Hong Kong, the Asia-Pacific hub of the global banking company, a BI and performance management implementation has managed to get more than 5,000 employees in some 200 branches aligned and motivated to meet group-wide sales performance targets. Previously, the bank had some 30 full-time employees dedicated just to the task of collecting and reporting information on customer visits, product sales, leads, and follow-up activity in each branch. Reporting amounted to keying data into isolated Excel spreadsheets or, worse, handwriting and faxing data up to division level.

In 2005, HSBC implemented a new CRM-like process and standardized interface for capturing detailed sales and lead information across the division. Integrated with half a dozen legacy banking systems, the accompanying data warehouse and Cognos BI deployment can serve up granular detail ranging from total results for Hong Kong down to sales results and leads by product, branch, and individual salesperson. There’s no ad hoc reporting or downloading into Excel, however, at least at the branch level.

“We give the branches a standard set of reports, a standard way of looking at those reports, and a standard way of reacting so we don’t operate in a different way in different branches,” says David Campbell, head of BI for personal financial services in HSBC’s Asia-Pacific region. Rolled out in 2006, the deployment paid for itself within one year through data-collection and reporting efficiencies alone, HSBC says, but that doesn’t begin to calculate the benefit of the performance management aspects of the project.

For example, dashboard-style, Web-based reports are tied to a sales performance management plan. Managers and salespeople see branch and individual progress toward incentive compensation targets. The goals are tied to strategic objectives around, say, up-selling and cross-selling mortgages and insurance, and incentives can add as much as 30% to an individual’s salary.
Talk about a smart way to get employees excited about BI.