URL Design - Page 29

Tim Berners-Lee has said that if he had known that the Web would be as popular as it is, then he might have thought harder about finding an alternative to the slash-slash part of the URL, which is particularly annoying when speaking URLs over the telephone. In principle, URLs are machine-readable code and should not have anything to do with user interface design. In practice, it is an unfortunate truth that URLs are exposed to users in many aspects of web usage, so we do have to consider them as a design issue.

Considering the popularity of the Web, there is no need to speak out the "http://" part of a URL when giving it over the telephone or when including it in a television commercial. Most companies simply refer to their website as www.company.com rather than http://www.company.com/ (the syntactically correct form). Although HTML purists deplore this abbreviated form of stating the name of a website, it seems perfectly acceptable to me, especially because almost all browsers add the missing protocol specification to the front and the missing directory specification to the end. The Web is now so ubiquitous that it is understood that anything starting with www and ending with .com (or .uk, .de, .jp, etc., outside the U.S.) is a website.

I recommend making both company.com and www.company.com aliased machine names for your web server. Currently, most users do include the "www." when typing in URLs, but sometimes they forget. Also, when speaking URLs over the telephone, it is nice to avoid the very awkward-sounding "www."

The most important component of a URL is the domain name (the machine name immediately after the http://). If users can remember your domain name, they can at least get to your home page, from which navigation and search are hopefully sufficient to allow them to find the page they need even if they don't have the rest of the URL. Most companies try to get their company name as their domain name, and I would definitely advise anybody who starts a new company these days to pick a name that is available not simply as a trademark but also as an Internet domain. Having an obscure domain name is going to cost big time in lost customers. Good domain names that are easy to remember and easy to spell are the Internet's equivalent of a Fifth Avenue real estate location in the physical world.

Our usability studies have shown that users rely on reading URLs when they try to decipher the structure of a site or the possible results of following a hyperlink. It would be preferable if browsers had better ways of making site structures explicit and of previewing the destinations of hyperlinks, but right now they don't, so users read URLs the way the ancients read cracked turtle shells: to divine a hostile environment with no known laws of nature.

Because we know that users try to understand URLs, we have an obligation to make them understandable. In particular, all directory names should be human-readable and should be either words or compound words that explain the meaning of the site structure. Also, your site structure should support URL- butchering where users hack off trailing parts of a URL in the hope of getting to an overview page at a higher place in the site hierarchy. Of course, it is better if users can navigate your site structure using your navigation buttons, but we know that a lot of users use URL-butchering as a shortcut: Such users should get reasonable results (typically a table-of-contents-like page listing the information available at the desired level of the hierarchy).

One day browsers, servers, and proxies will all include spelling checkers, but at this time users are doomed if they don't get every single character exactly right when typing a URL. Web designers can reduce the frequency with which users meet the dreaded 404 by making URLs easier to spell. Rules for easy-to- spell URLs are:

• Make the URL as short as possible (the longer the URL, the great the possibilities for making errors).
• Use common natural language words as much as possible because users normally know how to spell these words.
• Use all lowercase characters. If you use MiXeD cAsE, some users are guaranteed to forget some of the caps and get errors (depending on the server). In general, you should never rely on the difference between uppercase and lowercase letters in a user interface because such a distinction is a sure prescription for frequent user errors. Confusing upper- and lowercase characters is a so-called description error. Because the two objects are almost the same and because the most salient part of the description of the two objects (the name of the character) is exactly the same, users are very likely to confuse the two.
• Avoid special characters (anything but letters and digits) as far as possible. If punctuation is necessary, stick to a single character throughout all your URLs. Use all underscores or all hyphens, for example, but not a mix of the two.

Compound Domain Names

How might one make up a domain name to refer to a website that has multiple words in its name? For example, a site for Jakob Nielsen might be called jakobnielsen.com, jakob-nielsen.com, jakob.nielsen.com, jnielsen.com, and many other combinations of the two words. (The underscore character is illegal in domain names, but hyphens are allowed.)

Creating compounds by using dots (e.g., jakob.nielsen.com) only works for a company that owns the primary domain (in this case nielsen.com, which is taken by the Nielsen ratings). And if you have the primary domain, then why make a longer and more complex subdomain for your website? I recommend using the standard "www." as the prefix for websites because people know what it means and because having an address start with "www." is a nice indication that you are talking about a website and not something else (it used to be the case that this goal required the use of a full URL, complete with "http://," but these days, only very meticulous people bother doing so).

Thus, the three reasonable candidates are:

• Run the words together: jakobnielsen.com
• Use an abbreviation: jnielsen.com
• Use a hyphen: jakob-nielsen.com

Current mainstream practice on the Web prefers the first choice; simply run the words together to form a new "Internet word" for the domain name. In usability, the fact that most other people do something is reason enough to follow along because the most common practice is what users expect and find easiest to use.

Abbreviations work as an alternative for three or more words or when the result of running two words together would be very long and/or difficult to spell. My main recommendation is to run the words together if you are dealing with two reasonably short and easy-to-spell words.

Hyphens should be avoided because people often forget them, they can be mistaken for underscores, and they are rare (and thus a usability problem).

Search Destination Design: Further Examples - Page 28
Designing Web Usability
Fully Specify URLs in HTML Code - Page 30