Predictive analytics, agile development, user-centric business intelligence and improvements in visualization are giving new life to this mature technology।
How does your organization extract true value from its business information? Answering this question has been a persistent challenge facing technology and line-of-business executives for decades.
While business intelligence (BI) has evolved since the days of “green bar” reports, the industry still has a long way to go to offer companies business information that can be translated into actionable steps that drive business results, says Joe Bugajski, senior analyst in business intelligence for the Midvale, Utah-based Burton Group.
“There is a sea change coming in business intelligence,” he says. “The existing tool sets have been out there since the early ’90s—some of them before that. And we still have tools that are too complicated for most folks. We’re pushing too much of the technical mumbo jumbo behind BI into the faces of users, and we’re still not giving access to valuable information in a simple fashion to the majority of the business population.”
As a result, users are increasingly asking their BI units, “What have you done for me lately?”
The best BI teams answer that question with a bevy of new capabilities based on five trends that experts say are changing the face of business intelligence.
Trend 1: Predictive Analytics
If Ram Nagappan had to name one critical area where he thinks business intelligence has the most potential to completely transform his enterprise, predictive analytics would get the nod.
“If you look at it, everyone supplies records, everyone has dashboards—or they're planning on doing it,” says Nagappan, managing director for Pershing LLC, a Jersey City, N.J.-based financial services affiliate of The Bank of New York Mellon. “But in these economic times, the information that I know beforehand is what will help me save money and steer the ship in the right direction.”
As he puts it, the BI industry is just “scratching the surface” of predictive analytics. This is partially because analytics in general has lagged so far behind the rest of the more traditional reporting functions of BI.
“Analytics has been the last to the party in the BI space,” Burton Group’s Bugajski says. “All the easy stuff has been done. We can make very pretty charts and graphs, but it's not the [same as an] interaction with the core information of the business so I can understand what’s going on. That’s still missing.”
Right now, most organizations are pushing the boundaries of current tool set capabilities. “What we’re seeing are mostly in the research and university areas,” Pershing’s Nagappan says of current predictive analytic tool development. “I know that people can take their current analytical data models and other things that they’ve created and do a prediction on them, but the tools are not quite there yet.”
Clearly, Nagappan’s expectation for better tools tracks well with recent rumblings in the business intelligence marketplace. According to IDC, the analytics market is expected to grow about 4 percent this year.
In late July, IBM banked more than $1.2 billion on a bet that predictive analytics is the key to BI’s future. The investment was made in the acquisition of SPSS, an analytics firm well-known for its predictive analytics technology.
“With this acquisition, we are extending our capabilities around a new level of analytics that provides clients not only with greater insight, but also with true foresight,” Ambuj Goyal, general manager of information management for IBM, said in a statement about the acquisition. “Predictive analytics can help clients move beyond the ‘sense and respond’ mode—which can leave blind spots for strategic information in today's fast-paced environment—to ‘predict and act’ for improved business outcomes.”
While organizations wait for the market to shake out, Pershing’s Nagappan believes that those who prepare their subject-matter experts will be best prepared to take advantage of new technology innovations. “I think the challenge in predictive analytics is going to be building the subject-matter expertise within the analysts,” he says. “In order to predict systematically an analyst needs to know the subject matter well.”
Trend 2: Agile Development
The economic downturn is forcing BI departments to rethink the way they develop their solution sets, according to Wayne Eckerson, director of research and services for The Data Warehousing Institute, a Renton, Wash.-based analyst firm. With users crying for more capabilities and management demanding faster deployments, he believes more enterprises will start to port the agile development philosophies that have permeated the developer community to the more traditional BI development teams.
“With the down economy, there is a lot of movement to come up with lower-cost models and faster deployment to keep up with the business,” says Eckerson. “Organizations are exploring agile BI because the business doesn't want to wait around for even three months.”
That golden three-month period used to be the perfect milestone that BI teams would shoot for to satisfy the business with new innovations. “Now it’s more like a couple of weeks to a month,” Eckerson says.
The key to the agile approach to BI is that it “rolls out business intelligence in an incremental evolutionary way with a lot of involvement and participation from end users or customers,” says Ken Collier, senior consultant in business intelligence and agile product and project management for Cutter Consortium, an Arlington, Mass.-based IT advisory firm.
In the enterprise projects he leads, Collier targets a two-week iteration of new functionality releases. He says the factors most critical to meeting this demanding schedule are to keep milestones small and targeted; to foster a highly collaborative environment between analysts, developers and users; and to implement test automation for databases.
“That’s an entirely new concept for database folks who have been manually testing for years,” he says. “The problem is that if you work in two-week iterations and you’re trying to manually test these new features, you quickly get buried under the weight of your testing processes.” If you do it right, Collier adds, agile BI can deliver value in a number of ways. The most important is responsiveness to user needs.
“In a relatively medium-size data warehouse or BI system, it typically could take eight or 12 months of requirements analysis and development and testing before users get to see working [betas] on their desktops,” Collier says. “What is really key is being able to show users features within the first few weeks of a project when new data is trickling in every night—even if you don’t roll these things live into production—and being able to show users working features and get feedback so you can quickly adapt.”
Agile development can also cut down on function overkill. Collier cites industry statistics that show the typical user of any given system uses only about 50 percent of the features included in the application.
“Simply by virtue of the fact that we focus on the highest value things first, we can complete projects faster and at less expense,” Collier says. “We can converge more quickly on a system that’s ready to go into production, and the user can say, ‘This is good enough; I don't need that other 40 percent in there.’”
Even if an organization isn’t gung-ho about developing on a two-week schedule, the lesson to take away about the agile movement is its bite-size mentality of incrementalism. Pershing’s Nagappan says this is for organizations seeking to ramp up their intelligence maturity.
“Many organizations try to do an enterprisewide solution on Day 1, and that is a huge elephant to move,” he says. “That is not going to be a success. Any time you wait a year for a product to show up, it’s not going to be easy.”
Trend 3: User-Centric BI
Since the last time the economy took a nosedive in 2001, Nagappan has shifted his department’s focus to better customize the information he delivers to different user segments based on their roles within the organization.
“We have noticed that our customer segments—we call them personas—that use our platform are all different,” Nagappan says, explaining that personas can range from financial reps to marketing and sales folks to business advisers. “The key thing we recognized was that one size was not going to fit them all.”
As he helps the businesses come out the other end of the recession on a strong note, Nagappan’s top priorities include a shift to offering user-centric analytics based on role.
“We need to take the same data and create analytical models that satisfy the various personas that are going to look at the information so they have better decision-making ability from it,” he says. “We take the same transactional data and create various functional areas so these different consumers can come and take what they need.”
According to Burton Group’s Bugajski, it is this kind of user-centric focus that more BI departments must develop to enable the business to drive true value from the information it is analyzing. He says the typical organization too often faces the prospect of gathering BI information from what he calls the “human GUIs” of the enterprise: users who know how to extract data and end up interfacing with BI systems for colleagues who either don’t know how or don’t want to learn.
“There’s value there,” Bugajski says. “There are reasonable and responsible behaviors there. But that’s not the original vision for BI.
“Where is the tooling that my CEO could use? Where is the tooling that my business analysts who are not experts in data, but who are experts in marketing, could use? Where is that stuff? They want something that is as simple to use as a Google search. And that’s fair to ask.”
Trend 4: Visualization Improvements
Bugajski believes the only way organizations are going to extract the full value from their BI endeavors is if they redesign their visualization philosophies and designs. “Business intelligence as we know it is just about dead,” he says. “We need a new paradigm, and I think visualization is the key.”
The visual ways in which users collaborate and analyze information through Web 2.0 tools are setting the bar high for BI deliverables, which Bugajski says are sometimes stuck in form factors from the 1990s and even the 1980s.
Nagappan couldn’t agree more. He says Pershing is focusing on Web 2.0 technology in order to innovate better ways to pump up visualization and improve the way users interact with data.
“If you look at the traditional ways, you just put out a spreadsheet of information, and people might pivot that spreadsheet, but that's about as far as they used to go,” he says. “Now, with Web 2.0, we can create geographic mapping using Google; we can do Flash-based animation. We have AJAX technology and sharing.”
Nagappan says the cross-section of Web 2.0 and BI enables organizations such as Pershing to take the same information and make it consumable in a number of ways. This dovetails nicely into the user-centric model that his organization is striving for, he says.
In fact, Nagappan and his organization are so passionate about how Web 2.0 can change the face of visualization BI that they recently used these technology philosophies in creating a new flagship software platform for the company’s financial planning customers. Launched in July, the Pershing NetX360 puts the power of customizable data dashboards in the hands of users, allowing them to visually see real-time financial numbers and data crunched on-demand from a smorgasbord of data sources.
Trend 5: Operationalization of IT
At the advent of BI, the idea was to put the right data and analytics in the hands of people who could make actionable changes that would improve the way business is done. Somewhere along the line, that simple idea grew muddled.
BI systems grew up to be scattered across enterprises, complicated and difficult to use—even by business analysts.
As enterprises assess how to move forward with their BI efforts, one of the driving forces of these initiatives should be to make BI simpler and easier to access by a wide range of workers. In short, organizations want to bring BI back to its philosophical roots.
“When I started, one of the promises of BI was empowering decision-makers and knowledge workers,” says PivotLink’s chief marketing officer, Dyke Hensen, who calls himself an old BI “oak tree” after 20 years in the space. “It was to create pervasive BI and leverage BI for everyone. The problem is that over the years, a lot of these offerings became very complex, very bloated and expensive.”
Eckerson of TDWI agrees, saying that it is odd that so many enterprises’ operational workers have to switch gears between a business intelligence application and an operational application in order to open a dashboard or a report to see the impact of an action taken based on business intelligence.
Over the last two years, the market has seen a drive to consolidate these tasks with a push by major ERP vendors to help bring BI under a larger operational umbrella. Acquisitions such as the SAP pickup of Business Objects last year are a sign of where the BI space is headed.
Eckerson believes the recent shift to bring together business intelligence systems and operational systems such as ERP makes sense. “There is definitely an opportunity for vendors to embed BI right into operational applications,” he says.
Pershing’s Nagappan also thinks that operationalizing BI is a no-brainer. He says his organization worked to do so years ago, leveraging tool sets from Information Builders and in-house work.
“It’s a very key area for us, and we have done this for many years,” he says. “I know many other people in business intelligence who focus on finance [intelligence] and a few other [intelligence areas], but at Pershing, we focused on operational and compliance [intelligence] so many years ago because compliance is a key aspect of our business.”
Brian Kilcourse, managing partner for Retail Systems Research, a Miami-based retail IT analyst firm, agrees that this “operationalization” of BI is currently one of the most significant intelligence trends sweeping through enterprises. He’s seen a lot of anecdotal evidence illustrating how a shift to embedding BI within operations gives workers out in the field better tools to drive day-to-day operations and gives customers better ways to make informed purchase decisions.
“We’re seeing a lot of companies injecting actionable information into operational processes in just-in-time fashion,” Kilcourse says. For example, in one case study he analyzed, Kilcourse witnessed Virgin Megastores offer its store managers an effective way to improve sales. BI systems were integrated with up-to-the-minute in-store sales so that managers could see how hit titles were selling in comparison to other hits with similar sales.
The intelligence match-up compared the first few days of release of one title with other releases that had similar sales starts, giving managers the ability to project sales going forward. It also offered actionable analysis that enabled workers to pair up overstocked albums with hot sellers in endcaps to move otherwise stationary products.
Even though Virgin closed its retail stores for other reasons, Kilcourse says this application of operational BI is too good to be ignored. “They were basically doing a kind of a product mashup on the sales floor, in more or less real time, based on the signals they were getting from sales as they were occurring,” he says. “They were basically doing shelf resets based on the fact that one title was flying off the shelves, and they wanted the other one to fly with it.”
Kilcourse says these kinds of initiatives help organizations better adopt a sense-and-respond mentality. He also believes that embedding BI into operations provides very good back-end benefits.
“One of the big values is that the operational systems or processes can deliver to the business intelligence system some information that says, ‘This is what happened after you responded.’”
No comments:
Post a Comment