In my own experience, when I'm 'on a quest' for information, I most often use search functions in a site, or rely on 'overly obvious' navigational clues. When I'm designing a site, I like to cross-pollinate links throughout the site, using both in-text and graphical anchors. The staticred.net site is a good example of this: on each page, there is a link to the most current project. Timestamp is an even better example of this, as I've built in several ways to access the same functions in the web-app. (And, so is oven, a web-app I'm building to handle inventory and ordering on a website)
Of course, none of this is to say that clear and easy to use navigation isn't important; most often a user will not enter your site through the front door; they need a way to quickly identify where they are and how to get around. The thing that may not be noted about tests by Neilsen et al is that they are controlled tests; they're not necessarily letting the users go wherever they want, and may not be accounting for the fact that the user has likely already found the information then want by entering through a Google search, and are now looking to see what the rest of the site has to offer.
When I built the Matrikon site (below), I built it with this in mind:
Matrikon: Main page
Matrikon: Interior page
If users get to the main page, they can quickly find the following:
- Information for Oil & Gas and Mining - at the time, two of the most important target markets for the company
- Upcoming trade events - always a good idea to say where your company is going to have a public presence
- Quick links to 'portals' for specific industries - these alone drove a 400% traffic increase to the industry pages within 3 weeks.
The same tactics were employed on the interior pages with the addition of graphical cross-links to related products. And they worked.
That is, for the month they had the site up. But that's another story.