Is ranking the same thing as visitors?
No. It does however, tell you how sites compare against each other according to rank. Considering the size of the Internet and various other factors, a small difference in rank when measuring in the hundreds of thousands or millions probably isn't that big. The closer you get to the top of the ranking however, the more that difference means.
For instance, if your site ranks at 1 million and your competitor ranks at 2 million, you probably are not that much ahead in actual traffic. If however, your site ranks at 15,000 and your competitor ranks at 100,000, there is probably an enormous difference.
Will advertising on a better ranked site be more effective?
Not necessarily. You have to know your audience and the advertising has to be well placed. What are the demographics of the audience of the site and how well do they compare to those that buy your product?
To take it to the extreme, advertising a new sugared soft-drink on a site for teens will probably do far better than advertising it on a site that is designed for diabetic seniors. Advertising fishing gear on a sports site will do far better than it will on a site that is dedicated to interior decorating. Common sense is always required.
What factors may skew rankings?
Rankings are an inaccurate measurement at best. In fact, short of monitoring where every person in the world goes online, there are no guarantees of who is doing exactly what. Rankings can only be relied upon as indicators of performance.
Several factors may skew rankings with Alexa such as the browser of choice. Alexa cannot monitor AOL or Netscape usage, only Microsoft Internet Explorer. Therefore, since consumer sites are frequented by more AOL users that business sites do, the consumer sites are probably ranked a bit worse than they really are. For the greatest accuracy, you should probably compare business sites with business sites and consumer sites with consumer sites.
Alexa may have a few less bells and whistles than the high dollar ranking companies, but they claim millions of users. Most of the rest only work with a few thousand. Statistically speaking, the larger the number of total users (population sample), the more reliable the results will be and the less likely that any single user or event can skew the results.
