Business intelligence (BI) and search convergence has already resulted in a cutesy new term -- as well as some market misconceptions.
Stamford, Conn.-based analyst firm Gartner has coined "biggle" -- that is, when BI meets Google -- to describe efforts to bring BI and search together. In recent months, many BI vendors have made search-related
"Many companies are recognizing now that they need to provide business users with more access to BI information. Just the training costs [for commercial BI systems] are often quite expensive. Organizations are looking for easy and simple interfaces," said Dan Vesset, research director at Framingham, Mass.-based IDC.
While there are clear business drivers for search and BI convergence, it's still an evolving market that could change with time, said Jim Murphy, research director with Boston-based AMR Research. Step 1 is understanding the possibilities -- and the myths.
Myth #1: There's one correct way to approach business intelligence search.
Search technologies vary greatly -- from simple keyword searching to relevancy ranking (like Google), to pure-play vendors' unique algorithms and text mining technology. Text mining, also called text analytics, analyzes unstructured content to better determine the context and meaning of the information in relation to the search terms, while relevancy ranking looks at popularity and linkages to other documents.
Companies use very different methods to search BI and other data sources. New York-based Information Builders Inc. leverages data transaction adapters to discover and index data en route to applications and databases, so users can search information recently entered into enterprise systems. Ottawa, Ontario-based Cognos Inc. touts its mechanism for exposing BI report metadata to search engines, which it says enables better performance and scalability. And, Cambridge, Mass.-based Endeca Technologies Inc. said its homegrown search algorithm can analyze structured and unstructured data with no help at all from a BI system.
Myth #2: It's just about using search to access existing business intelligence information.
Beyond returning information to users, some new search technologies act as "ETL [extract, transform and load] for unstructured content," Murphy said. Some pure-play vendors and IBM can search unstructured data sources, use text mining to extract relevant information, and load appropriate content back into data warehouses. Then BI systems can apply their own analytics, he explained.
Since unstructured information represents about 80% of the data in an enterprise, according to IDC, there is a huge opportunity to enhance BI systems, not just search existing BI information, according to Stouffer Egan, chief executive officer of Autonomy Inc.
"All of these unstructured information assets have meaning that needs to get into the BI system. It's the integration of those assets into the BI paradigm that's been missing," Egan said.
Myth #3: Demand for business intelligence search has exploded.
Using search technology for BI has been possible for some time. Fast could have already connected its search technology to a BI application before the formal partnership it announced with Cognos, said Rob Lancaster, vice president of channel development with Fast. But customer demand hasn't been there because, he believes, BI applications typically have been an "island," accessed by power users and holding sensitive -- often financial -- information. But now that BI's role is changing, search is an attractive way to make information available to non-power users.
Market maturity and new business drivers are the main reasons BI and search convergence is happening now, AMR's Murphy agreed, and formal integration partnerships between vendors are important because BI systems maintain key information security and access controls, he added.
Myth #4: Choosing the right business intelligence search technology is the hard part.
Search must be integrated with processes, as well as technology, Murphy said. Integrating search technology can also take time, depending on the method and number of data sources.
For example, Information Builder's Intelligent Search requires integration time, to set up the appropriate hooks into all of the back-end systems, acknowledged Michael Corcoran, the BI vendor's vice president of corporate strategy and chief communications officer. During product demonstrations with prospective customers, companies often wanted to search all structured and unstructured information throughout the company, he reported. But Information Builders advises customers to prioritize the systems in the interest of implementation speed and scale. Customers will be able to get to everything eventually, but it will take time, Corcoran said.
Myth #5: Business intelligence search automatically offers immediate, enterprise-wide benefits.
BI search must be deployed strategically if it's going to be useful, according to IDC analyst Vesset.
"Just slapping in the search technology doesn't make sense," Vesset said. "Companies need to figure out which applications and processes can benefit from search."
Companies might deploy search for users to find reports within a BI system, he added. This could save on training and deployment costs and speed adoption by end users.
"A lot of companies have hundreds or even thousands of reports developed over time. There's a productivity value in users finding something faster," Vesset said.
Another potential application is analyzing warranty claims, Murphy said. Reports by mechanics repairing automobiles, often in free text form, might contain a goldmine of information -- but only if a company has a way to effectively parse through all of the reports and find patterns. Other applications might be customer support functions or fraud detection.
Companies investing in search should deploy the technology with a specific problem and user community in mind, Vesset and Murphy agreed.