Reviewing YOUR Search Engine Expectations

A vast majority of companies already have one or more engines, and many are dissatisfied and shopping around.  For example, we've seen one client running Vendor A, thinking of moving to Vendor B, only to have another client the following week planning to move from Vendor B to Vendor A.  Stop!  If your sole reason for selecting a new vendor is that you are unhappy with your current vendor, you are setting yourself up to be disappointed again.

No matter how frustrated clients are, we try to make sure they understand why it has failed, and how the other vendor can fix this.  If a new vendor has sold you on something nebulous like "better relevance", your money might better spent on a bag of magic beans.

The biggest general problem companies seem to have with search is mismatched expectations; this trouble manifests itself in some common complaints that we hear:

1. The engine is more complicated to setup and maintain than we thought. Do you have relatively simple "generic" search requirements?  If so, then get a simpler engine, there are plenty out there.

   Conversely, if search is core to your business, or you have somewhat unusual or demanding requirements, then this may be a staffing problem, not a "vendor" problem.

2. The engine is not flexible enough to do all the things we need. In many ways this is the opposite of issue 1.  In some cases businesses just outgrow their initial search engine.  Think long and hard about your requirements before you make a switch.  If you really need flexibility, ask lots of very specific questions.  Make sure you understand the requisite skill-sets required to support each vendor.  If search is critical to your business, plan to cultivate some real in-house search expertise and retain some real technical resources.

3. "Users still can't find anything!?" - Usually either
   3.a: The engine relies on "Single-Shot Relevancy" and doesn't offer search navigators.
   - or -
   3.b: The system hasn't been maintained since it was first setup.

   Don't plan to just "set it and forget it".  Search engine performance should be monitored at least quarterly, if not monthly or weekly.  If you're not sure how to do this, get some professional help.

4. This stuff is SO expensive! Be careful with that one.  TCO (Total Cost of Ownership) is a huge factor in search projects.  Ask all vendors for a complete estimate, including professional services.  Also, "free" software like Lucene is amazingly powerful, but you will need some real programmers to implement it; for larger projects this may not be "cheaper" than the "expensive stuff".

5. "Garbage In = Garbage Out" Despite some claims to the contrary, most search engines will not magically fix mangled content.  If your data is horribly broken on the way in, your search will suffer.  Fix your titles, meta tags, summaries, etc. See our series "Poor Data Quality gives Enterprise Search a Bad Rap" parts _1_ and _2_. With careful planning, some vendors can clean this up, but your current vendor can probably do that too.

6. Process, Process, Process! If search is a core component of your business, don't trust machines to take the place of human judgment.  We see clients "throw" search engines at random data sets all the time, without doing any up front analysis.  How much data do we have?  In what repositories?  In what locations?  Do we have highly secure data?  Do we have duplicate (or almost duplicate) data, or multiple versions of files?

   "Process" means doing an audit of your current data and future business requirements, a checklist of questions and discussion topics for humans to ponder.  It's actually more entertaining than it might initially sound, and if you're not sure how to do it, get help.  Don't skip it!

The Good News is that this is a buyers' market.  There are MANY good enterprise engines out there (maybe even the one you already own!) but this is definitely NOT a "one size fits all" market.