What Ever Happened to “Eureka!”?


      Bookmark and Share

This year marks the 10th anniversary of the launch of Alta Vista in December 1995 and more than 40 years since the first full-text search engine with Boolean operators became operational. Indeed, most of the features now commonplace on enterprise search engines had been developed and commercialized by the mid-1970s. So you would think that the use of enterprise search technology would be well-established and reliable by now.

In July 2005, the Quarterly Survey from McKinsey Consulting reported on the Global Executive Survey that the company had conducted among 7,800 executives in 132 countries, a fifth of them at CEO or CIO level. Overall, 29% of C-level respondents and 40% of other senior managers reported that it was difficult to find information on which to base company-wide decisions. The inability to find information from within the enterprise almost certainly will be decreasing the productivity of employees and increasing business risks.

This is born out by a survey among some leading UK financial sector institutions that was carried out under the sponsorship of Inxight Software. According to this survey, 73% of respondents reported that the main barrier knowledge workers face in sharing corporate information is not being able to use one information retrieval tool to capture data across several repositories. Fifty-eight percent of respondents said that their company's search tools were ineffective at sourcing needed information quickly and efficiently, and 66% of the companies interviewed said employees were re-generating information simply because they were unaware whether the documents already existed in the business. Figures from Sue Feldman's group at IDC suggest that the average time spent searching is around 9.5 hours a week, of which 3.5 are wasted finding nothing of relevance.

The overall awareness of search by many managers has been limited to the experience of using Google and other Web search engines, and there seems to be little understanding of the differences between Web search and intranet/enterprise search. This situation may improve now that desktop searching is rapidly increasing in performance.

The fundamental issue about search is this: if a user does not trust a search engine to deliver information that they can use immediately for making a business decision, then any metrics such as relevance and speed are irrelevant. The moment that level of trust is broken because of poor indexing, search results lists that cannot be interpreted easily, or documents that still cannot be obtained, then the investment has been totally wasted. There also has to be the trust that accepts that no relevant documents have been found because there are none in the index. Null results are always difficult to accept.

The challenges of specifying, selecting, and implementing an enterprise-level search engine can be daunting. Until the search engine is fully implemented and has indexed all the designated content, the performance of the software cannot be assessed, and there is also a continuing need to "tune" the search engine as new content is added and new types of user requirements arise.

One of my passions (I do have some outside the confines of the firewall) is motor sport, and car manufacturers are always trying to show that their technology not only is the best there is but is being used to build better cars for the public. In a similar vein, you would assume that the search vendors would demonstrate the very best they could offer clients by the search functionality on their own sites. After all, they have the software, and they are in control of every page/document/metadata tag. Try this test. Just search for the term intranet in the Web sites of the major search vendors and look at the results you get. I dare not name and shame because by the time you read this column, one can only hope they'll have made some significant improvements.

I also despair of the lack of any serious guidance from the vendors on how to make a business case for implementing enterprise search. One of the benefits cited currently by a leading vendor is "reduced time to convert data to knowledge discovery and action improves business process efficiency and increases return on information assets." Just run that past me again, would you? Making a business case for a search engine probably is harder than for a CMS. I am very skeptical about cases built on productivity gains, because the numbers are not credible. Having a better search engine may not reduce the amount of time spent searching for information, as the users may now spend longer on search tasks because they feel that in doing so they will identify all the relevant information that the organization possesses or purchases from third parties.

Over the last year or so I sense there has been a significant improvement in the level of understanding of the benefits, implications, and challenges of implementing a CMS. We have come so far in search and yet so often remain lost.