A colleague and I have been having a sporadic disagreement concerning
whether a keyword search of our Web site indicates a failure in the
site's structure.  I would be interested in your insights.

The context is that our Web logs track all user invocations of the
"search our site" feature, including the keywork(s) entered in the Web
form.  I believe that we can and should interpret multiple users
searching for the same or similar concepts as a failure in site
design, and use this evidence as one input to a redesign effort.

Most intra-site search engines -- ours, sadly, included -- are so
hideous from a usability perspective (unclarity of matching algorithm,
too high or too low volume of responses, inadequacy of responses) that
only a truly desperate person would turn to them.  If the site
hierarchy made an efficient navigation path possible, use of keyword
searches would be unnecessary, and, correspondingly, would dwindle.

From looking at the logs, I see two main patterns.  Some users
immediately invoke the search engine, without exploring the link
hierarchy.  I attribute this to learned behavior from other poorly
designed sites, and do not feel that this necessarily reflects poorly
on us.

Other users click around for a while before starting a search.  This
is where I feel we can draw some conclusions.

My colleague disagrees.  He believes that the search mechanism is an
independently valid and appropriate navigation strategy, and its use
should be seen as such.

Any comments?

        -- Michael Levi