> Hoping I'm not violating unwritten list rules by mentioning this -
If you are, then the rules are dumb. Search analytics should be part of any
product (even homebrewed) that is search-related. It was a
vote-off-the-island function when my FPOW reviewed search engines last year.
The extent to which search analytics-heck, even transaction logging-heck,
even simple reporting functionality-is missing from LIS products makes me
bubble and stew.
> One of the challenges of course is how to interpret search statistics.
> 90% percent of users don't go beyond page one: is it because our ranking
> works so well, or because they immediately change their query instead of
> looking further, or because they can't find the 'next page' link?
This practice is so common (known to happen with Google, OPACs, etc.) that
while interpretations of it may be important, it is even more important to
design around it (e.g., your first page should always be your best results).
Of course, that does not mean you can't do task analysis asking people to
figure out how to go to the next page or how to modify a query on the fly. I
don't want to hurt your feelings, but this practice is SO widespread that I
wouldn't tot it up to your product being transcendent. ;-) I mean, it could
be, but there are larger forces at work...
As a sidebar, I know a usability specialist interested in the issue of
Karen G. Schneider
[log in to unmask]