Search in the Enterprise: Silly Putty
October 20, 2008
Reading the summaries of what people said or did not say at the Carnegie Mellon conference about search on October 17, 2008, left me confused. I am in London, and I had to rely on write ups by the capable David Needle for Internet News here, a couple of emails, and one hurried phone call to keep tabs on the program speakers’ thoughts.
What I took from these inputs was:
- Enterprise search is not yet too good. I like having my conclusions validated when I am thousands of miles away working with my co author Martin White. Our new study “Managing Successful Enterprise Search”, forthcoming from Galatea in November 2008 accepts this point of “not too good” as a given. We focus on what must be done to address the management shortcomings that are often more responsible than technology for search problems.
- Search continued to be used in different ways and with different meanings. The Googlers talk about finding information. Other speakers talked about locating experts or an individual who worked on a specific project. One word “search” is left fuzzy. Little wonder then that it was, based on the information I received in London, to know exactly what “flavor” of search is the one the speaker is using to explain a particular approach to these issues. It was clear to me that when high profile experts get tangled in definitional issues that users are left at out in the cold.
- Enterprise search, regardless of how one defines it, remains a tough problem. As many entrepreneurs and their sources of funding have discovered, enterprise search is leaving many users dissatisfied. An employee needs information to perform work. Problem is that “work” is somewhat fluid. Some employees need to answer a customer’s question about an invoice. That’s one type of search. Other employees need to get a document signed and want a police style “where is she” service. That’s another type of search. Other employees don’t search at all. These folks are sitting in a meeting and need information pushed to them germane to the task at hand. One system performing all of these functions quickly finds itself expensive to deploy, expensive to maintain, and expensive to support.
One attendee suggested to me that the answer to these problems was more sophisticated systems. My reaction to this is, “Maybe.” In tough economic times, organizations want systems that solve mission critical information problems in ways that make benefits clear, show payoffs from the money poured into search.
Good enough is no longer acceptable. The woes of Fast Search & Transfer in the 18 months leading up to its acquisition by Microsoft are a grim reminder that “traditional” approaches to enterprise information may not be sustainable. I will have to wait until I return to the US to gather more details of this meeting. As it stands on Saturday morning, October 18, 2008, enterprise search is moving quickly to an embedded function, wrapped in more useful enterprise applications, and destined to become one component in different types of information access options. In short, the take away for me from the conference inputs I have is that enterprise search is a bit of a dog’s breakfast.
Stephen Arnold, October 20, 2008
Comments
4 Responses to “Search in the Enterprise: Silly Putty”
-
- Subscribe to Beyond Search
Feature archive
News archive
-
[…] The first reaction to Manbar’s unhappiness was from industry pundit Steve Arnold, who seized upon Manber’s comments to once again don the Cassandrian mantle of doom regarding the search industry, noting for what seems like the umpteenth time that “enterprise search is leaving many users dissatisfied.” Once that was out of the way, Arnold had an interesting observation, “An employee needs information to perform work. Problem is that “work” is somewhat fluid. Some employees need to answer a customer’s question about an invoice. That’s one type of search. Other employees need to get a document signed and want a police style “where is she” service. That’s another type of search. Other employees don’t search at all. These folks are sitting in a meeting and need information pushed to them germane to the task at hand.” […]
Rebecca from Vivisimo
Yep, you are absolutely correct. I emphasize that most users of search systems are dissatisfied. Why? Few vendors believe this. Few IT professionals really care about Buffy and Thad in marketing. Few senior managers know much about internal information. Will this change? Yep, search enabled applications. The financial noose is tightening around the necks of many vendors, and those who are not delivering services that generate a happy face inside the “o” of ROI will be following in the footsteps of Wikia and Yahoo. So, make users happy, and Martin White and I will stop focusing on this fundamental problem with search and retrieval in an enterprise.
Stephen Arnold, October 22, 2008
Stephen, if I understand your argument correctly, you’re saying that enterprises don’t care enough about internal information to invest either money or effort in developing the means to access it, and as a result they build systems that yield poor returns in terms of user productivity and satisfaction. If that is the gist of your argument, I have to concede that it’s true more often than I’d like.
But there are exceptions. I think that the most successful enterprise customers are those who don’t think generically in terms of “enterprise search” but rather in terms of particular problems driven by enterprise information needs. Those needs correspond to tangible (and sometime enormous) returns, thus justifying the comparably small investment needed to obtain them.
Daniel Tunkelang,
In general the shift is to search enabled applications that contribute a fungible benefit to the organization. The day of the fuzzy wuzzy “one size fits all” search system may be nearing its end. The reason eDiscovery is hot is not that it is search. It is hot because search is embedded in something that can deliver a financial payoff and reduce certain costs associated with a legal matter. A fuzzy wuzzy search solution is not too good at this type of implementation. Companies want systems that work and deliver.
Stephen Arnold, October 23, 2008