Print this article | Return to Article | Return to CFO.com
A burst of innovation has made business-intelligence software more relevant to day-to-day operations.
John Verity, CFO IT
September 15, 2005
Cendant Hotel Group, one of the world's largest lodging franchisors, licenses more than 6,300 properties under Travelodge, Super 8, Ramada, and other brand names. To search for rooms and tour packages across this empire, customers can go to the company's brand Websites or to those of distribution partners such as Expedia, Orbitz, Hotels.com, and Travelocity, among others. Having plenty of options may be good for business, but until recently Cendant had only limited insight into how well it was being served by those channels, or how much and what kind of business any particular channel was bringing in minute by minute. If there was a surge of interest in triple rooms in San Diego, for example, Cendant wouldn't find out until it was far too late to adjust its display of properties or add inventory — its own or that of other chains.
And then there were the screen-scrapers. Every so often, unauthorized hotel reservation Websites would swoop in — all quite legally — and collect huge batches of data regarding room availability. "It's easy for them to screen-scrape us and stock their shelves," says Nick Forte, director of application architecture at Cendant. This activity might eventually lead to some bookings, but the massive spike in queries might also crush service levels for other, more important channels.
Enter software start-up Celequest Corp., one of many new players in the fast-growing field of business intelligence (BI). Cendant began using Celequest's Activity Suite software because the product provided visibility into and metrics about a range of operational issues. Now, Cendant can better maintain service levels across all channels and learn more about what its customers are looking for, minute by minute. "If the software's operational dashboard alerts us to a slowdown in a certain channel, we can drill down and see where the problem is," says Forte. And the company can keep the screen-scrapers at bay with "throttling" techniques (which block specific Web traffic), because it can now distinguish, in real time, between different types of requests. Activity Suite software can also compare its own real-time analyses against historic BI data (stored in a data warehouse, for example), which could help Cendant assess one marketing campaign versus another or understand current operations in a larger context.
Real-time BI, sometimes referred to as "business activity monitoring," "operational BI," or "operational performance management," is focused less on pulling reports from vast stores of data and more on pushing important findings to the right people, often before they even think to ask. As exciting as that is, it is just one of the many new ideas swirling about the BI scene right now, making this far-from-new sector more vibrant than it's been in some time. Innovation is everywhere, fueled by venture capital and by customers who realize that, as useful as traditional BI tools may be, they are in some respects simply the foundation on which whole new schemes for monitoring, measuring, and optimizing business performance can be built.
Indeed, BI technology is undergoing substantial change, both in the scope of the data it collects and analyzes and in the range of employees who use it. BI's reach is expanding because virtually every software application can now feed fresh data into a data warehouse or real-time data store. What's more, some new BI software can monitor events on a second-by-second basis. No longer is BI simply looking back at what took place 6 or 12 months ago. Its focus is more and more on the here and now. And, increasingly, its focus is on focus.
Many new products have been created not as generic BI tool kits but tailored to assess specific business processes. They offer deep domain knowledge out of the box. Software from Edge Dynamics Inc., for example, helps users monitor, manage, and optimize order management in the pharmaceutical industry, while a product from Virsa Systems Inc. does the same for real-time regulatory compliance. "BI is a fabulous opportunity for start-ups," says Josh Greenbaum, principal at Enterprise Applications Consulting. "[A BI vendor] can say to a user, 'Here is an out-of-the-box solution. I am the expert, and I am going to tell you what you need in order to be smart.' " Traditionally, Greenbaum points out, BI suppliers have left it largely up to customers to work out their own analytic strategies.
Innovation is visible everywhere. Tableau Software Inc., for example, is commercializing government-funded research, undertaken at Stanford University, in so-called visual-query languages that enable graphical exploration of complex data. And BI-related hardware is changing, too, as start-ups such as Calpont Corp., Datallegro Inc., and Netezza Corp. turbocharge the data warehouse's underlying database servers by harnessing low-cost, commodity processors.
Change is even afoot in BI business models, most notably with the rise of open-source software (see "Open Source: The Future of BI?" at the end of this story). And to make things even more interesting, Microsoft is mounting a push into this market, too, leveraging its SQL Server database and adding a dashboard-building tool, called Maestro, to its Office suite.
Keeping It Real Time
Real-time or operational BI has attracted the interest of many companies that, like Cendant, want to know about problems now. One customer uses Celequest's software to continuously monitor the quality of different parts it buys from outside suppliers. If the software sees parts quality dipping below preset thresholds, it can alert the appropriate managers via pager, E-mail, or cell phone and provide them with a Web-based dashboard to investigate the problem's causes and model the outcomes of proposed solutions. Monitoring ongoing events can also be useful in ensuring regulatory compliance and managing risk. Wall Street brokerages, for example, can analyze trades and other transactions as they are executed and identify processing bottlenecks or financially unsound risk positions. Software from Memento Inc. monitors individual employees' use of sensitive back-office applications. If the program sees one bank teller querying customer files much more often than other employees do, for example, it can alert managers to a possible identity theft or other insider fraud.
Companies should evaluate any claim of "real-time" BI carefully, experts say, because definitions can vary. Even the fastest systems have some latency, or time gap, between a given event (a transaction taking place or a quota not met, for example) and the system's ability to capture that and report it to the appropriate employees. Companies should also bear in mind that empowering people to make faster decisions, or redesigning a department or process to respond more quickly to changing conditions, are also important to the success of real-time BI.
Innovation Marches On
Technological innovations underlie many of these new products. Perry Mizota, vice president of products and marketing at Syndera Corp., explains that traditional approaches to BI often hinge on storing data in a database and running queries against it. In contrast, he says, newer software relies on "event-driven computing" techniques that can, on the fly, find correlations and trends in masses of data records that arrive from multiple sources in no set order.
One entrepreneur hot on this trail is database pioneer Michael Stonebraker, founder of Ingres Corp., whose latest start-up, StreamBase Systems Inc., is one of several firms working on so-called stream processing. Stonebraker says StreamBase's software can crank through 150,000 messages per second on an ordinary desktop PC. That's enough horsepower to help credit-card firms crack down on fraud, assist capital-markets traders in identifying fleeting arbitrage opportunities, and help manufacturers keep their factories running smoothly.
How far could real-time analysis go? The military envisions sprinkling thousands of tiny radio-equipped sensors — known as "smart dust" — across battlefields, each one reporting the movements of troops or vehicles that come its way. Individual tanks and personnel carriers would be chock-full of sensors, too, monitoring their locations on the ground and their fuel and ammunition levels, for instance. Analyzed in its totality, all of this data would supposedly make commanders better informed and better able to manage battles.
The business landscape may not be quite so complex, but it is moving in that direction as companies adopt what SAP calls real-time awareness. With most business processes and workflows now automated from end to end, a great deal of data is available about every phase of every transaction and work step. A huge number of devices, from computers and soda machines to automobiles and pacemakers, are now able to report automatically and continuously on their status. And as radio frequency identification tags show up on zillions of pallets and packages moving through retail and industrial supply chains, even more data will become available to help companies gauge and react to changes in customer demand.
None of this has escaped the notice of major BI and ERP players, all of which have recently expanded their BI offerings. As but one sign of the times, later this fall Cognos Inc. will provide what it calls "event management" as part of a major new release of its BI suite, Cognos 8. "We'll be able to look at things that change frequently and set triggers and rules," says Neal Hill, senior vice president of corporate development at Cognos. And this past July, executives from Dunkin' Brands revealed that they have created a BI system that uses technology from Oracle, Hyperion, and Blue Cube to provide a real-time look into how well products sell and under what conditions.
If the big players in BI are concerned about competition from this raft of newcomers, they aren't letting on. "Companies look to the big BI vendors for standardization," says Hill. "They come to companies like ours for worldwide support and a full set of capabilities. It's rare that a user would decide to go with a start-up for this kind of capability."
But he readily agrees that "there is a second wave in BI that is opening the door to lots of start-ups." This is being driven by Web-based technology that helps to lower the price of the software, and by a new focus on performance management that allows customers to monitor a huge range of data points minute by minute — everything from which customers are happy and which are the most profitable to which distributors are doing well or not so well.
"We have a technology partners program that was designed specifically for early-stage companies that have products complementary to our own," Hill says. "We work to build links between our products. If something really catches fire, we have the option of acquiring the partner company. And from that company's point of view, it gets access to customers that it wouldn't otherwise have."
Hill says that, especially for large companies, BI requires the muscle of a major vendor. "Delivering a single, unified view of a company's data is not that easy, particularly when the customer has two or three ERP systems, different database systems, different networks, and different desktop environments."
That may be true, but complexity at the high end does not seem to be stopping new software companies from coming to market with a burst of new offerings. They won't all survive, but customers seem likely to win either way.
Open Source: The Future of BI?
It changed the rules in operating systems (Linux), dominates the market for Web servers (Apache), and has won tens of millions of Websurfers to its cause (Firefox).
Can the open-source software model do the same in business intelligence? A handful of open-source companies, some with serious venture capital behind them, say yes. They are taking on BI's dominant players — Business Objects, Cognos, Hyperion, and Oracle — on the simple premise that those companies charge too much for annual software licenses.
The upstarts' primary pitch to customers: We've done away with the usual high-cost sales and marketing machines, so you save big bucks without sacrificing much, if anything, in the way of functionality. Development and testing, meanwhile, get handled largely by customers themselves. Business models vary, but an open-source company will typically offer at least a basic version of its software for production use at no charge, and customers, if they so choose, pay a yearly fee for support services and perhaps for a beefed-up version of the code.
JasperSoft, known until recently as Panscopic, provides a set of reporting tools designed to be embedded within homegrown or commercially developed enterprise applications, in much the same way that Business Objects' Crystal Reports product is used. Because the JasperSoft code costs virtually nothing to employ, and because it's rooted in Java, JasperSoft reckons it will have great appeal to corporations as they strive to get rich BI reporting into the hands of thousands of frontline employees. The company says its software, originally developed by a four-person team in Romania, has achieved around 10,000 installations worldwide and can save companies as much as $100,000 more than competing products. It recently raised an $8 million third round of venture money from investors, including Partech International, an early backer of Business Objects.
Greenplum is working the database angle, commercializing a 12-year-old open-source project called PostgreSQL. Its focus is online analytic processing, the core software in data warehouses and data marts. By running on clusters of cheap Linux servers, Greenplum's code enables customers to work with, say, one terabyte of data for between $15,000 and $150,000, versus the nearly $1 million that NCR's Teradata unit charges for a comparable solution.
Perhaps the most ambitious BI-focused start-up is Pentaho Corp., which has set out to create a full suite of software that would enable users to both monitor and control business processes and performance from dashboards. Pentaho doesn't expect to design all this from scratch but to harness other open-source projects wherever that makes sense — which is the open-source way. Pentaho's management team has worked together in BI for 15 years at Cognos, Hyperion, Oracle, and SAS Institute.
How vulnerable are BI's incumbents? Not terribly, as even the start-ups admit. "Nobody's going to go out of business because of open source," says Richard Daley, CEO and co-founder of Pentaho and a veteran of the BI market. "Hyperion and the others have revenue streams that will last a long time. But we'll be able to divert lots of dollars that go for new projects and additional features." Already, Business Objects has responded by offering some of its own software to the open-source community.
Eric Rogge, an analyst at Ventana Research, a performance-management research and advisory-services firm, says the true appeal of open source has been misunderstood. "The interesting thing for customers is not the low cost, it's that they can try before they buy," he says. For that reason, open-source companies typically target IT staff, while the big players in BI make their case to C-level executives in finance and operations. Says Rogge, "Open-source software will find a good niche, but will it take over? Unlikely."
In the near term, open source's main effect will be to alter the ecology and economics of BI development activities by providing low-cost alternatives for certain building blocks. But given the technology's continuing success at the foundational levels, experts say no one should discount its eventual ability to move "up the stack" to the applications level. —J.V.