In-Depth
BI Trends to Watch in 2005
The coming year should look a lot like 2004, although new several trends will loom large.
Compared to the acquisition tumult of 2003, 2004 was a year of relative tranquility in the BI marketplace. A host of small vendors were snapped up, but outside of analytics powerhouse AlphaBlox Corp. (which IBM Corp. acquired this summer) no big name BI brands changed hands.
Instead, BI trends took center stage, with dashboards finally coming into their own. BI vendors such as Business Objects SA, Cognos Inc., and Hyperion Solutions Inc. took the fight to the big ERP vendors, and enterprise reporting went mainstream—thanks to the efforts of Microsoft Corp., among others.
According to many industry watchers, the coming year should bring more of the same, although several trends—including an expected uptick in the offshoring of BI tasks, a surge in uptake of data mining solutions, the continued commoditization of ETL technologies, and a long-expected backlash against ad hoc BI—could loom large in 2005.
The Year in Review
Last year at this time, Wayne Eckerson, education director of The Data Warehousing Institute, predicted that dashboards would be prominent in 2004. Looking back, even he admits to being surprised by the accuracy of this prediction. “Who did not build a dashboard in 2004?” he asks. “People are calling whatever they build a dashboard, as long as it displays a few key metrics on the opening screen. Nonetheless, the business community has finally discovered the medium by which it wants to consume analytical information.”
What accounted for the runaway success of dashboards this year? In part, users realized just how valuable they can be.
“A dashboard gives them a quick glance at the key measures that they need to track and lets them drill down into more detail if something is awry,” Eckerson notes. “In essence, dashboards are graphical exception reports. The message from users is, ‘Don’t bother me if nothing is wrong.’”
For some time now, ERP powerhouses such as SAP AG and Oracle Corp. have pushed into the BI space, stealing potential market share from BI vendors by promising better integration with their ERP systems. To a large extent, this continued in 2004, although Cognos and Hyperion also demonstrated some backbone, pushing back with business performance management analytics of their own design.
“ERP customers who’ve made substantial commitments to the ERP software find it a no-brainer to use the vendor’s analytical software for all their BI needs,” says Eckerson. “Rather than pursue acquisitions, BI vendors seem to believe that the best defense is a good offense. Thus, Cognos and Hyperion, in particular, are working hard to etch themselves as the leading providers of business performance management applications.”
What’s In Store for 2005
Eckerson and other industry watchers have long predicted a rise in BI outsourcing, but to date there’s been more than a trickle (and considerably less than a torrent) of such activity. Many firms have outsourced their ETL development to the Indian subcontinent, for example. In this respect, 2005 could be a make-or-break year for BI outsourcing, some experts believe.
“While many BI shops take it for granted that they will be using offshore help, I’ve also encountered a number of BI managers who say that they tried offshoring and it doesn’t work very well in BI, which requires close communication with users and iterative development,” Eckerson comments. “I suspect that in 2005 we’ll begin to really discover what jobs and tasks make sense to offshore and which don’t. At a macro level, any analysis and project management must stay on-shore, but pure development based on well-defined specifications can go abroad.
Peter Nolan, an independent data-warehousing consultant with more than 20 years of experience, agrees—to a certain extent. “I also think 2005 could also be the year that more ‘outsourced’ BI starts happening. That is, more and more customers allowing the BI solution to be run from a data center ‘somewhere else,’” he indicates. “Very large numbers of companies have their data processing outsourced now, and it makes sense that they will outsource BI. But that does not necessarily mean they will outsource their BI to the same vendor that the OLTP is outsourced to.”
The Reemergence of Data Mining
Ever since the passage of the Sarbanes-Oxley Act of 2002, Hyperion, SAS, and other vendors have predicted a compliance-driven increase in demand for data-mining solutions. Don’t look now, but there’s some evidence that demand for data mining tools did surge last year, although 2005 could be the year in which this prediction is put to the test.
“Anecdotally, I can say with great confidence that this has happened. At TDWI, we see greater interest in data-mining classes. We think the time is right for wholesale adoption of data mining because the BI market has matured and organizations are looking to extract more value from their investments, which sometimes reach into the tens of millions of dollars,” says Eckerson. “Of course, not all organizations have a need for data mining. It’s not a cheap thing to get into.”
Ted Ledman, director of information management with a financial services institution based in the Midwest, allows that his employer is one such organization. “Our experience with [OLAP] cubes is that they are inefficient, and the overhead of creating the cubes provides no benefit to our users. We are a Kimball shop and we do everything with star-schema data marts,” he comments. “We have no data-mining tools. Hopefully, some day we will, but there are other tools that have a higher urgency and priority than data mining.”
ETL Blues
Over the last 24 months, many traditional ETL powers have sought to recast themselves as data integration specialists, a move that some industry watchers is a reflection of the ongoing commoditization of ETL. In this respect, notes TDWI’s Eckerson, 2004 was a disappointing year for Ascential Software Corp. and Informatica Corp., the two most prominent data integration players, as both companies notched less-than-overwhelming financial results.
Unfortunately, 2005 could bring more of the same. “Although I believe the data integration market is real, I also know from experience how hard it is to sell middleware, which is what data integration is,” he says. Given the lure of India as a destination for ETL and other data integration work, Eckerson points out, Ascential, Informatica, and other ETL vendors have their work cut out for them. “Many organizations don’t think about using a tool to move data from place to place, unless it must happen on a regular basis to support operational processes. And why use an ETL tool for that?”
Data warehousing consultant Nolan concurs, adding, “There are too many ETL vendors chasing too few ETL dollars.” As a result, he says, what happens in 2005 could be the prelude to a thinning of the ETL herd. “Of the independent ETL/integration vendors, only [Informatica and Ascential] have the ability to survive over the longer period. Even vendors like Ab-Initio are going to struggle against these two and the big vendors,” Nolan speculates. Elsewhere, he suggests, Business Objects, Cognos, Microsoft, and Oracle all field ETL tools. These vendors could increasingly compete with Ascential and Informatica for share: “These are major vendors and they will fight with [Informatica and Ascential] for license sales in 2005. They will not give up easily.”
Like Eckerson, Nolan believes that too few customers understand the value of data integration solutions. “Customers are still not showing much sign of really understanding the ‘data integration’ problem and are far more likely to sign the order for more BI software than more ‘infrastructure’ software,” he notes. “I am unsure of the reasons behind this, but the revenue growth of the BI tools vendors in the face of tough times for ETL vendors tells a pretty clear story.”
Ad Hoc BI Backlash
Finally, 2005 could be the year in which organizations finally reject what has traditionally been called “ad hoc” BI. Ad hoc BI tools, explains Eckerson, let users graphically define queries using predefined query objects. There’s one big problem with this approach, he says: “Our industry has finally discovered that graphically creating queries is really not much better than handwriting them,” he comments.
“In 2004, I discovered that BI managers who’ve deployed ‘ad hoc’ capabilities to their users have really deployed something other than ad hoc query tools. In most cases, they’ve deployed parameterized reports where users can filter predefined reports using a series of predefined drop-down lists. Or they’ve deployed OLAP views into an OLAP database, which itself creates a predefined sandbox in which users can navigate.”