[Sigia-l] ROI/Value of Search Engine Design - Resources?

[Christopher Fahey [askrom]]

Boniface Lau wrote:
> When I read that someone did a search and received 210 results, I
> expected something like the following:
>
http://ourworld.compuserve.com/homepages/boniface_lau/images/all210.jpg
> 
> But when I went to Amazon, I got the following:
>
http://ourworld.compuserve.com/homepages/boniface_lau/images/laptops.jpg
> 
> Your suggestion that Jared and I were describing the same phenomenon
> flies in the face of information architecture. Mind you, you are
> posting to sigIA.

Thanks for the very kind clarification. 

Your two screenshots are superficially different, but they are
functionally the same thing (from the perspective of the search engine
program). The actual functionality of the search engine is what Jared
was talking about, not the way the user interface displays the
functionality. 

The three laptops listed in your initial search results are no more
relevant to your search query than the 210th result is. The first hit of
your initial three hits is no more valid than the 210th hit of the query
(22 clicks later, a nice Apple laptop):
   http://LessLink.com/210th_laptop/

There is a big difference between "returning three results" (which
implies that the three results are the three very best matches to your
search, which is not the case) and "hiding all but three of 210
equally-valid results" (which is actually what happens). 

Amazon is betting on the likelihood that a user's search query will be
sufficiently well-formulated that they can get away with not troubling
the user with the 207 other results. Sometimes this trick works because
sometimes the item you seek will be in the first three displayed hits
and you won't need to click "view all 210 results". Unfortunately, in
the case of the "laptops" query, the first result is no better than the
very last. Amazon is not, as you say, "doing pretty good". In a
discussion about the effectiveness of search engines, it is needlessly
splitting hairs to distinguish between showing 210 equally-valid results
on 22 screens instead of on 21 screens.

Amazon is great, but it's very disappointing that a query for "laptops"
doesn't return a page with a very very prominent link to this page:
  http://www.amazon.com/exec/obidos/tg/browse/-/565108/

It's easy for me to see the logic behind Jared's assertion that "on-site
search engines are worse than nothing", but it's also easy to see how
Amazon might improve the search engine as I describe above. This same
technique (subverting the 'purity' of a search algorithm -- i.e.
cheating) is used all the time to cause paid advertisers to appear at
the top of search engine pages. Just look on the right edge of this page
to see folks who, however cynically, understand the limitations of
search engines:
  http://www.google.com/search?q=information+architecture


-Cf

[christopher eli fahey]
art: http://www.graphpaper.com
sci: http://www.askrom.com
biz: http://www.behaviordesign.com