13 Career Jeopardizing Enterprise Search Issues

July 6, 2014

The ArnoldIT team has combed through our archive of enterprise search data. We have identified the top 13 surprises that enterprise search delivers to licensees. Get hit with several of these surprises and you might find yourself seeking a future in a different line of work.

13 Users don’t like the new system or the old system, for that matter. Dissatisfaction with enterprise search, regardless of vendor, runs at 55 to 70 percent. See Successful Enterprise Search Management at http://www.galatea.co.uk/index.php?option=com_content&task=view&id=35&Itemid=53

12 No one pays attention to search costs until the CFO conducts an audit. Cost overruns plague nine out of 10 enterprise search deployments. The reason is that a comprehensive compilation of costs is not part of an enterprise search deployment. When a system crashes, the costs for emergency and rush work comes from a different line item. Customization is usually taken from a different budget allocation. Consultants and contractors are paid from another budget allocation. When the costs are added up, everyone seems surprised at the money spent for a system few find satisfactory. The hunt for a scapegoat is on.

11 Open source search and proprietary search solutions differ little in costs. Aside from an initial licensing fee, the costs for customization, optimization, contractors, programming, and enhancement are essentially the same.

10 Many major enterprise search systems are now based on open source software. The reason is that the cost for the basic functions are rising and are difficult to control. Therefore, vendors use open source and concentrate on extra cost add ons.

9 Every enterprise search system struggles to process content without human intervention, additional “connectors,” extract transform load activities, and original scripting. When content cannot be acquired, the few who notice will squawk, often loudly.

8 Latency creates a problem because new or changed content imposes significant costs on the licensee to cope with the need to process content in near real time and then refresh the indexes whether these are state of the art in memory systems or old style spinning discs and cached methods. When a user asks, “Why is the system too slow?, it may be difficult to make improvements when budgets are constrained.

7 Modern systems include adds on that permit faceting, query expansion, and linguistic functions. Unless these are “tuned” by subject matter experts or analysts, the outputs can generate irrelevant or off-query outputs. The notion of “smart, automatic” search and retrieval are often chimera.

6 Users typically do not conduct a thorough, research librarian type of investigation of query results. Enterprise search systems that generate laundry lists of results that are stale or irrelevant will be used by the person running the query. The assumption that online systems are “correct” is held by 95 percent of an enterprise system’s users. Only when a user cannot find a document that is supposed to be in the search index will the user realize that the system is not working as assumed. If the issues arises during a crunch, the prudent search manager will polish that résumé.

5 Enterprise search scaling is expensive and complex. The idea that scaling is seamless and economical is false. Improving the “performance” of an enterprise search system requires correct identification of the particular factor or factors creating latency. More frequent updates may not be possible without re-engineering an enterprise search system’s infrastructure. How much has Google’s core method changed in 14 years? What about Amazon? What about Autonomy, Endeca, Lucene, etc.? The answer is, “Not too much.” Search is very, very complicated.

4 Interface does not improve precision and recall of a search system. Interface and cosmetic design changes are easy to talk about and “more fun” to work on that figuring out how to process content more quickly and update the searchable indexes in with significantly less latency. If users grouse, an interface change won’t silence the critics or slow the proliferation of bootleg systems in units that are dissatisfied with the search status quo.

3 Search with text mining functions often rely on standard methods and algorithm  configurations that licensees cannot modify without specialist training. As a result, many systems output results that may be based on assumptions not germane to the licensee’s content. Hence, outputs purporting to provide insight into business intelligence or predictions may be incorrect. Search is not text mining. Search is not a Silver Bullet for Big Data. Search is pretty much type a query and get a laundry lists of stuff that must be reviewed by a human. Automatic reports are often off point.

2 Search appliances are not money savers. The Google Search Appliance costs as much as Autonomy or Endeca to deploy. The cloud is not the big money savers marketers want me to believe it is. Cloud search solutions reduce the need for capital expense, but the on going costs are comparable to on premises solutions. A search appliance may be like handcuffs. The cloud may be overly complicated. No highways leads to the Magic Kingdom for search. If you think search is a slam dunk, you are misinformed.

1 Enterprise search systems are more alike than different. The reason is that computational methods have not changed much since the first commercial systems became available  in the late 1960s. The differences are created by marketing, not by significantly different numerical recipes. Most users cannot differentiate between or among systems. The concepts of precision and recall are unknown. Users believe that search systems are right almost all the time. Yikes.

Stephen E Arnold, July 6, 2014

Comments

4 Responses to “13 Career Jeopardizing Enterprise Search Issues”

  1. New Enterprise (intranet) Search – Stephen's Lighthouse on August 7th, 2014 9:33 am
  2. best kids bikes on August 15th, 2014 2:48 am

    Hello my family member! I wish to say that this post is
    amazing, nice written and include almost all significant infos.
    I’d like to look more posts like this .

  3. Vernon on September 5th, 2014 11:54 pm

    Great blog here! Also your web site quite a bit up fast!
    What host are you the use of? Can I am getting
    your associate link on your host? I want my web site loaded up as fast as yours lol

  4. Major on September 21st, 2014 6:20 am

    Hello, every time i used to check webpage posts here early
    in the break of day, because i like to gain knowledge of more
    and more.

  • Archives

  • Recent Posts

  • Meta