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

[Jared M. Spool]

rich at richardwiggins.com wrote:

>I can conceive of a site that doesn't need search.  I'd define that as a
>site that doesn't have any links on it.  So if you launch
>gettysburgaddress.com, and all it has on it is one page with the text of the
>Gettysburg Address, then you don't need a search box.

I also believe that small sites don't need Search.

>The more complex your site becomes, the more obvious the need to have a
>search box.  If your home page and each of the tiers beneath has an average
>of, say, 20 links, 20**3 is 8000 links to look through.  You can group and
>label and categorize to Da Vinci levels, it's still a lot of choices.

Yes, it *is* a lot of choices. That's why it's important you really 
understand how categories work. (I wrote a little about this here: 
http://www.uie.com/Articles/strategies_categories.htm and will be 
presenting some more information at my IA Summit presentation, Scent of a 
Web Page.)

>A site that's really simple -- a brochureware site with a home page and 5
>linked pages and no tiers beyond -- probably doesn't need search.
>
>A university, a Ford Motor Company, a General Electric, an HP, an Amazon, a
>Linksys, a NY Times... they all need a search box on the home page, and they
>all need Best Bets or equivalent.

I'm not saying these sites wouldn't benefit from a well-constructed search 
and a decent best bets implementation (and probably a nice faceted, 
guided-navigation approach, ala Endeca). However, I'm still not convinced 
that every site that with more than one page *needs* these things. Nor am I 
convinced that every site that currently *has* these things automatically 
satisfies users better than sites that don't.

Based on our discussion, my sense is that you want all designers to always 
put search on their site. A noble goal and one that I think lots of people 
probably agree with.

I'm just saying that I don't have any evidence to suggest doing that is a 
good idea. That doesn't mean I think it's a bad idea -- I'm just waiting 
for the evidence before I endorse it. (It does seem to me that if it was a 
good idea, the evidence supporting it would be easy to come by. But, so far 
it has eluded me.)

>What were the clothing sites you tested that worked well sans search, by the
>way?

They were small online retailers: The Gap (http://www.gap.com) and Old Navy 
(http://www.oldnavy.com). The third site in the top three was Lands' End 
(http://www.landsend.com), who is slightly larger and where people did use 
Search, but not frequently.

The two worst performing sites were Macy's (http://www.macys.com) and 
Newport News (http://www.newport-news.com). Users used Search frequently on 
both, but to no avail.

As for size, Macy's online is larger than the Gap, but smaller than Lands' 
End. Newport News is about the same size as Old Navy. Size doesn't seem to 
matter.

I'd like to believe the notion that the more content you have, the more you 
need Search. I don't believe it with the evidence I've seen so far.

Jared


Jared M. Spool
User Interface Engineering
http://www.uie.com    jspool at uie.com

Don't miss User Interface 7 West, March 23-27, Burlingame, CA. 
http://www.uiconf.com