A scientific approach to search engine optimization
Tuesday, May 6th, 2008By now we all know that search engine optimization is a must for virtually any website. In fact, natural search traffic should generate the bulk of a mature website’s visits or that site may quickly find itself out of business. On singletracks.com, our mountain biking website, unpaid search engine referrals generate 87% of our visits - more than 100,000 in just the past month.
Clearly search engine placement is important for traffic generation yet what I’ve seen from search engine optimization (SEO) consultants has left me unimpressed. There are countless SEO forums, blogs, and corporate sales sites that espouse certain SEO “truths” but rarely seem to back these rules up with little more than statements like “in our experience” or “as far as we can tell.” SEO is just too important to leave up to gut feel and subjective experience.
Search engine crawlers and ranking engines are computer programs which means they don’t run on feelings and subjective rules. To understand how search engines view and rank webpages, it’s important to use rigorous and controlled scientific testing so we can base SEO “rules” on the results of these tests. To that end, we’ve been running our own highly controlled tests over the past several months to see how some typical SEO myths stack up against reality. In several cases, we were pretty surprised at the results.
We won’t bore you with any more of the details of our strategy or our testing methods - suffice it to say we’ve taken out all the noise in our tests to answer individual SEO questions like this one: When placing keywords in a URL, where are they most effective - as part of a ‘directory structure’ or as the name of the document? For example, which is better: domain.com/keyword/ or domain.com/keyword.html?
In this test, our page (domain.com/keyword) was indexed by Google first by almost a week, though this is not always an indication of a better ranking. Over more than two months of testing, however, domain.com/keyword remained the top result for our test keyword - a result that surprised even one professional SEO consultant we spoke with. Remember, we used a highly controlled test keeping keyword density, title tags, etc. constant while changing just one variable - the page name - to achieve this result. So if you have a choice - go with domain.com/keyword over domain.com/keyword.html.
We have more than a dozen such test results we’re dying to share on this blog over the next several weeks - come back soon to see what else we found!
