Sorting Out the BI Players

Computerworld |

Analytic technology is on the rise. Business intelligence software expenditures are a growing share of the IT budget. Data collection and storage costs are increasingly driven by analytic uses. DBMS and enterprise application vendors are focusing their product differentiation efforts largely on analytics.

I wrote those lines in November 2004 (see "Playing Catch-up on Analytic Technology"), but the message would be equally true in 2002, 2005 or 2008. Applications that record data got pretty mature before the Y2k deadline. The action now is in applications that help use information to make actual business decisions. As a result, enterprises look at analytic technology vendors very differently from how they used to. They aren't just buying tactical solutions for departmental application needs or enterprise data consolidation; they're hoping to pick serious enterprise-standard analytic technology "partners."

As desirable as vendor consolidation is, however, these standardization decisions are a little premature. Sure, in many cases it would be reasonable to pick a single enterprise reporting/query/analysis/dashboard/scorecard vendor, use its platform for your new BI apps and perhaps even migrate some old ones. But nobody's product is fully baked yet, the industry has a lot of consolidation ahead, and so the vendor you pick today might not look like the best choice three to five years down the road. I'm not trying to warn you off of vendor standardization. Metadata integration is a good thing, enterprise price breaks are a very good thing, and even if you choose badly, the future migration away probably won't be too painful anyway. I'm just serving notice that whichever vendor(s) you do pick will have serious deficiencies, and you'll just have to keep your fingers crossed that they eventually fill in the gaps.

The biggest trade-off is between user interfaces and operational BI. On the one hand, only two vendors have convincing analytic/transactional application integration stories, namely app powerhouses SAP and Oracle-PeopleSoft-Siebel-WhoeverIsNext. Cognos suggests that a modern service-oriented architecture equals easy integration, which equals excellent future integration, but that's still highly speculative. On the other hand, neither SAP nor Oracle offers a really competitive user interface -- although both are, of course, working feverishly to close that gap.

The other big trade-off is among product breadth, integration and proven scalability. The major contenders in the category of "broad, integrated, scalable analytics suite" are Cognos, Business Objects, Information Builders, Oracle, SAP and Microstrategy; the second tier probably starts with Microsoft, Hyperion and SAS. Cognos, Information Builders and Microstrategy have better claims to streamlined architectures, but I don't think the others are at a significant practical disadvantage. All have broad product lines, but all also have areas where product is missing, weak, unproven or just plain bolted on.

The best solution at this time is probably to select two enterprise BI vendors, one of which is your dominant application vendor, and the other your favorite independent. As third-generation analytic business processes (download PDF) become more prevalent, your app vendor may have a huge advantage in meeting some of your more process-oriented analytic needs. On the other hand, for at least the next few years, the slickest decision-support technology will probably still come from the independents. All this could be changed, of course, by yet another big merger (and successful product line integration). But if you have to decide right now, my recommendation is that you hedge your bets. (Besides, are you ever really going to have just one BI vendor? Yeah, right, you will -- just like you're going to have just one app vendor, or just one brand of DBMS.) And what if you must pick just one? Then go with an app vendor. That analytic application integration story is really compelling, and the app vendors have strong infrastructure offerings as well.

Even if you go with a tandem of vendors, whomever you pick will probably have holes in their product lines. So besides the big guys, you might want to look at one or more specialist vendors as well, or big vendors that you cast in a specialty role. Cognos, Business Objects, Hyperion and the app vendors all have planning technology, but if you're not satisfied, consider another vendor on that list. For really serious optimization, you might look up Aspen Technology or even little Riverlogic (and if neither of them does the job, it's probably time to roll out a few copies of Mathematica). If enterprise reporting is an issue, look at Cognos, Business Objects, Information Builders, Actuate or the Actuate-related open-source BIRT. For a user interface, Spotfire has cool analytical widgets not offered by the bigger vendors, and there are always a bunch of struggling data visualization companies. If you need predictive analytics tools, SAS or SPSS are the obvious choices, unless the simpler KXEN offering suffices. Fair Isaac is obviously a major supplier to industries with fraud concerns, and for revenue management apps you might want to look at Zilliant or Vendavo. As for text mining -- well, the candidates are too numerous to list.

Ultimately, which BI vendors you should choose depends on a lot of factors, starting with the diversity of your computing environment, the applications you need, the analytic culture of your knowledge workers, and, of course, the details of your software portfolio, analytic and otherwise. And as with most software purchases, less important than the actual technology you buy is the process change you undergo in connection with the purchase (see "Not So Fast for Enterprisewide Analytics"). So no simple guide or quadrant graph is going to tell you the right choice. This is one of the few technology areas where there really are still a lot of viable options, and you have to pick the one that seems best for your particular situation.

See Curt Monash's blog.

Smarter BI

Stories in this report:

Business Intelligence

It’s time to break the ChatGPT habit