4 - Web Measurement Standards
One problem with the Web: there are no benchmarks. There aren't even any standards. Even the most basic terms - a hit, page view, impression, visitor, - have no consistent definition.
A common woe among metrics managers are constantly having to define (and defend) their core metrics. There are a handful of quotes and anecdotes from metrics managers that underscore the problem.
There are a handful of organizations and sites that are attempting to come up with industry-standard definitions. None have become industry standards.
When it comes to benchmarks, it's difficult to determine whom to benchmark against: clearly, you don't measure a retail site against a search engine or a news site, but even when you whittle down to your industry segment, goals are dissimilar. A retail site that is designed to enable the customer to order one item as quickly as possible is different than one designed to encourage the customer to browse and buy many items.
And to complicate matters further, Web metrics are kept secret. Data about customer behavior is sensitive, and a wise company keeps the most meaningful metrics well under wraps.
There are third-party sources that attempt to remedy this, such as the "Net Ratings" service (started by Nielsen), which uses a body of sample users who allow their usage to be monitored as a method for measuring sites. However, their metrics emphasize page-views (their model draws from broadcast advertising, where more viewers is more better) - and unless your site is intended to generate advertising revenue, then the metrics are probably not in line with your goals.
And in the end, comparing your site to other Web sites is not useful or meaningful. Your Web site can be "winning" the numbers game, when compared to others, but not doing anything at all for your company.