Breakthrough Analysis: Will Search Deliver Better BI?Breakthrough Analysis: Will Search Deliver Better BI?

Here I thought we were doing a good job at business intelligence. A steady diet of presentation and packaging innovations--dashboards and portals, open source and suites, exploratory visualization and operational analytics--has brought BI out of the lab, providing tools for every information consumer. But now search has burst onto the BI scene.

Business Objects, Cognos, Information Builders and SAS recently announced partnerships with Google Enterprise, and Cognos additionally hooked up with Autonomy, FAST and IBM's OmniFind. Do these alliances portend a revolution in how we do BI? Or will the search box--omnipresent on the Web and indicative of Web-design shortcomings--merely be another easier access method? I can't think of questions that are more vital for BI.

Given the closely timed announcements of these alliances, my first thought was that the BI-search partnerships were simply about competitive positioning. Information providers that don't get on the Google bus risk being run over, and Google in turn must constantly expand its reach to stay ahead of its competition. Google's voraciousness seems to have motivated its partnership with Information Builders, which describes the iWay product as the widest-reaching adapter framework available, tapping into more than 300 different applications. IWay lets Google index--in addition to static documents--the deep web of content that is generated from databases.

Yet the BI vendors do recognize the benefit of providing access to reports and analyses using a search box. SAS Institute CTO Keith Collins says it will "change usage dynamics." James Thomas, director of product marketing at Business Objects, went so far as to tell me that "the concept of BI portals has always been a failure. Portals are about controlling information. Google is about embracing freedom." The issue is that portals and other BI interfaces incorrectly assume the user knows what information is available, how it's structured and where to find it.

The import of these vendor statements about search is weakened, however, by the singularity of allying exclusively with Google and the requirement to license Google's OneBox for Enterprise. Neither out-of-the-box search nor choice of search engine is built into most initial BI-search releases.

Cognos is the exception. Its strategy, as implemented in its Cognos Go Search Service--as explained to me by VP of product innovation and technology Don Campbell--seems well thought-out, not simply a matter of positioning. Cognos Go not only ties to multiple search engines, it also integrates with IBM's open-source Unstructured Information Management Architecture framework, which facilitates interplay with the spectrum of UIMA-compliant text-analytics solutions. Another differentiator is that Cognos' search work was driven by real-world need, by a pharmaceutical firm that wanted to construct a quality-management application linking customer contact reports with analysis of structured product data. The firm mines customer e-mail and phone transcripts using Autonomy for entity extraction, clustering by product category and trending. They merge results into an Oracle database for Cognos analysis. Cognos and Autonomy subsequently productized the application.

Cognos and Information Builders share another distinction, enabling what IBI VP for corporate strategy Michael Corcoran characterizes as "intelligent links" between search and BI. Users are not limited to accessing prerun reports. Instead, Cognos Go and IBI's WebFocus interpret search terms and extract what they recognize as parameters for template reports. The BI back ends generate new results on the fly and dynamically insert references into the search-engine indexes. These searches return analyses that are, if the technology works right, a large step closer to the answers we seek from search engines rather than the hit lists they most often produce.

My first impression that BI-search partnerships were motivated primarily by market positioning was only partly correct. Sure, the alliances will help search vendors tap the Deep Web and make BI access easier. What's most exciting is that we've moved a step closer to natural-language business intelligence.

Seth Grimes is a principal of Alta Plana Corp., a Washington, D.C.-based consultancy specializing in large-scale analytic computing systems. Write to him at [email protected].