Search Realities: No Fun for Sure
November 22, 2011
Our Overflight service pointed us to a write up called “Enterprise Search Explained. What about Next Steps?” We agree in principle with the article because enterprise search is convoluted. We don’t think the write up goes far enough. In fact, we think that the approach is part of the “search problem,” not part of the solution. There is a sharp distinction about the use of clicks, links, and popularity via user scoring with thumbs up, like buttons, and happy faces and the enterprise.
First, most enterprise content gets one or two clicks. Once in a while a document will generate a large number of clicks, usually for the World Cup pool or a change in the benefits program. The routine enterprise content is of interest to a small number of users. Popularity is zilch. There are few inbound and outbound links in most enterprise content. Finally, the notion that “big data” will unfailingly point the user to a hot trend is silly.
Second, in an organization the notion that “all” content is indexed is also wrong and, in many cases, illegal. It would be great to peer into the employment applications to find the colleague with the exact experience one needs for a proposal. The challenge is to lever out that information without dragging salary, employment reviews, and other “personal” data along for the ride. In some firms with government contracts, colleagues are not permitted to know about the existence of a project. I have worked at one firm where the president was not cleared to review the details of a major government contract.
Third, the marketing baloney that says, “Our system can index enterprise content” is an invitation to a cost overrun. How does a “free”, “low cost”, or over-hyped search system handle drawings from an AutoCAD system, pluck data from a legacy Ironside application running on “frozen” AS/400s, and tap into price changes in a traditional database system. The fact is that transforming and processing content is an expensive task. The phrase “you don’t know what you don’t know” applies to much in the enterprise search sector.
We are okay with systems from such firms as IBM, Microsoft, Oracle, and Polyspot when there are sufficient resources available. We know that established vendors like IBM can make “anything” work. IBM is a consulting firm as you may know. Even promising vendors like Polyspot can work wonders in an organization unable to locate information in a timely manner.
The trick is to keep ones feet on the ground and the realities firmly in mind. We know that’s not as much fun as making up crazy assertions. But a system which works is the objective in my opinion.
Stephen E Arnold, November 22, 2011
Sponsored by Pandia.com
Comments
2 Responses to “Search Realities: No Fun for Sure”
-
- Subscribe to Beyond Search
Feature archive
News archive
-
[…] Enterprise Search Realities: No Fun for Sure Beyond Search: Most enterprise content gets one or two clicks. Once in a while a document will generate a large number of clicks, usually for the World Cup pool or a change in the benefits program. The routine enterprise content is of interest to a small number of users. […]
[…] Enterprise Search Realities: No Fun for Sure Beyond Search: Most enterprise content gets one or two clicks. Once in a while a document will generate a large number of clicks, usually for the World Cup pool or a change in the benefits program. The routine enterprise content is of interest to a small number of users. […]