File structure of web sites?

["Geoff Hart" <geoff-h -at- mtl -dot- feric -dot- ca>]

Barbara Stuhlemmer is seeking <<...insight in to the best way to 
configure the file structure of a web site? We currently have a lot of 
small files in a few large directories.  The concern is that the file 
structure may inhibit the performance of the server and reduce the 
usability of the web site.>>

Unless your audience and information are both identical to the one 
on which some hypothetical standard is based, it's unlikely that 
you'll find any published solution that is much use. Audience 
analysis and usability testing of your assumptions based on that 
analysis are the only sure way to develop a good solution. You 
need to find out the content of the information that your audience is 
most likely to want to receive when each file downloads: for some 
topics, that may be a five-line summary, whereas for others, they 
want the whole 20-page white paper. Then you need to test your 
solutions to ensure that they really do meet audience needs, and 
that they're feasible given constraints on your server.

That being said, some general guidelines:

- Try to come up with a design that downloads at least the first text 
components quickly so that users can quickly determine whether 
they've found the right page (or set of pages). If they've found the 
right page, they can start reading immediately while the rest of it 
downloads; if not, they can stop the download fast instead of 
waiting for the whole thing to complete.

- Make the hierarchy relatively flat rather than relatively deep (more 
choices on each page, rather than fewer choices and more steps 
until you reach the final set of choices). The balance between flat 
and deep depends on the nature of the information and thus, on 
how users will try to access it. And avoid the "everything on one 
page" approach you see with so many portal sites: the more 
information you put on a page, the harder it can be to find specific 
information buried in all that clutter. It'll take some work to come up 
with a good compromise solution.

- If you have tons of files, stay away from search engines; the 
typical search engine remains more of an exercise in frustration 
than a truly efficient tool for locating information. That may change 
over the next year or so, but imho, search engines are still works in 
progress. Instead, consider producing an index (just like in the 
back of a book) that presents users with all your keywords so they 
can read until they find the keywords that relate to what they're 
seeking. (Search engines don't work largely because they conceal 
the keywords, and make you guess what keywords the author of a 
page used to describe the page. The better portal sites, such as 
Yahoo are popular precisely because some human has done the 
work of categorizing pages and placing them in relevant 
categories.) With this approach, you'll find that the index gets 
downloaded a lot once users learn it's there and learn that it's 
useful; if it is truly useful, then you'll also minimize the load on your 
server because users will go directly to the page they need rather 
than downloading dozens of files in a vain attempt to find what they 
need. You can build small indexes manually, but for larger jobs, 
look into a tool such as HTML Indexer (http://www.html-
indexer.com/), which automates the process.


--Geoff Hart @8^{)}  geoff-h -at- mtl -dot- feric -dot- ca  (Pointe-Claire, Quebec)
"If you can't explain it to an 8-year-old, you don't understand it"--Albert Einstein