Skip navigation

How do I rate thee?

There are a variety of tools to assess your website's performance. Unfortunately, you're also likely to get a variety of results

Globe and Mail Update

If your company has any kind of consumer-facing Web presence—and most companies do—or you're an entrepreneur with your own website, chances are that you've run into one of the biggest grey areas on the Internet: How do you compare your website's performance to that of your competitors? In other words, how do you figure out whether your site is just a collection of words and pretty pictures, or whether it's actually reaching your customers?

Measuring how many people come to your website isn't that hard—most web servers can run tools such as AWStats, which tell you how many visitors and page views you get a month, where viewers came from and how many pages they checked out on each visit. Google offers such a suite of tools for free. The tools, called Google Analytics, require you to add a small piece of code to your pages.

So, now you've got all kinds of information about your website. But how do you know whether your stats measure up to other companies in your industry? The simple answer is that you don't. That's because your competitors probably don't publicize their own internal figures for traffic and "conversion" rates (how many visitors actually wind up clicking a link to buy something or find out more information).

In the absence of that data, you have to use the services of a traffic-measurement company, which tries to arrive at those numbers through external means. There are several of these companies, including comScore (which acquired Media Metrix last year and recently acquired M:Mobile, a mobile measurement company); Nielsen (the TV, radio and newspaper measurement company, which built a web business by acquiring NetRatings); and several web-only businesses such as Compete.com, Hitwise and Quantcast.

Each of these companies uses one or more of the following methods for measuring web traffic: software that is (voluntarily) installed by surfers and tracks their behaviour; software that sits on an Internet host's servers; a piece of code installed on websites; and web or phone surveys that ask users which sites they visit. The problem is that each firm comes up with different numbers, and there's no way of knowing which one is more reliable.

In some cases, figures from comScore have shown traffic decreasing at a popular website over a period of months, while numbers from Nielsen or Hitwise have shown just the opposite. In other cases, there's a massive gap between the figures from a rating firm and those from a site's own server logs. For example, in the case of MLB.com, the major-league baseball site, Nielsen reported that it had 6.2 million unique visitors in December, while the site showed that it had 19.4 million.

So what's the reason for such large discrepancies? There are a number of factors. A site's own server tools (such as AWStats) may not distinguish between visits from actual people versus hits from search-engine "spiders" such as Google's web-crawling robot. In other cases, it's the traffic-measurement firm whose data is distorted because of a variety of flaws—including the fact that people often lie when taking surveys. Downloadable software from companies like comScore, meanwhile, tends to be installed by non-web-savvy users, and therefore can undercount the impact or influence of a technology-related site.

Other traffic-measurement techniques tend to undercount visits from Macintosh users, or from people who surf while they're at work (where most of us can't install software on our PCs). It's gotten so bad that the Interactive Advertising Bureau asked Nielsen and comScore—the two largest measurement companies—to submit to an audit of their techniques to see whether their numbers were an accurate reflection of the traffic that major websites are getting.

Newer web-based ventures such as Hitwise don't use surveys or installed software, but rely instead on large Internet providers who agree to hand over their server logs. In the case of Hitwise, they also claim to measure something better than pure traffic—what the company calls "share of Internet visits," or how many hits you get compared with competitors. But the criteria that go into calculating those numbers remain proprietary, and some sites have criticized Hitwise for undercounting in much the same way that Nielsen and comScore do.

The bottom line? You should get as much data as you can, from as many different sources as you can, and look at it all together to see if there are any patterns—and then take the whole thing with a grain of salt.

Recommend this article? 1 votes

Back to top