Rand Fishkin
Founder and CEO
SEOmoz
3-18-10
Your annual survey of SEO’s reveals what the community thinks are the top ranking factors for achieving top rank in organic results. Does your own research validate the outcome of that survey?
The research we do doesn’t always match perfectly with the aggregate of expert opinion. Our methodology follows two main things:
1. Correlation Analysis — do top-ranking websites tend to have these factors (more so than lower ranked sites)?
2. Causation Analysis — we build ranking models to emulate the search engines’ ordering systems as best we can and attempt to derive causative answers. Obviously, we’re not at 100% accuracy, so there’s a substantive margin of error on this, but it’s accurate enough (in the 70% range) to draw some conclusions.
What we’ve determined is that the top ranking factor from the survey, [keyword-rich anchor text links from 3rd-party sites,] is mathematically wrong. [We found that] root-domain diversity is the highest-order ranking effect. [In other words,] getting links from a diverse set of domains [is the most important ranking factor].
This shows in simple correlation analysis as well. If you order Google results by root-domain diversity, [you’ll see it has the highest accuracy of any individual metric].
We’ve also observed that while experts think H1, H2, H3 tags are important for SEO, it appears that only the H1 tag actually matters at all and even then, only slightly. Experts also felt the alt attribute of images was not a very important place to use keywords, but the research suggests it’s on par or possibly slightly more important than the H1 tag.
What are the implications for marketers? What does this tell us about how Google thinks?
Root domain diversity of links is a clear signal that Google weighs broad popularity. It’s tough to [generate diverse links] in a scammy or spammy way. To get 1,000 unique root-domains to link to you, you have to have some pretty compelling incentives. It’s relatively easy to get 1,000 links from the same ring of websites.
If Google is trying to determine the authority of your link footprint, domain diversity is among the most challenging factors to game (particularly getting links from high quality, authoritative sites that have their own diverse link profile). It tells marketers that gaming the algorithm is incredibly challenging, and that attracting natural links is going to be an easier path to high rankings.
So, by valuing diverse domain links, Google essentially creates the imperative for marketers to create more good content. The more good content people product, the more people will need Google to find it.
Google, along with many marketers in the space will repeat precisely that advice, but I think it’s a bit naive, acutally. The lesson here is that Google rewards people who earn links. You don’t just earn them by producing “great content,” lots of “great content,” earns no links at all. You need to incent others to link to your content and successfully market that incentive.
For example, if I write the best article in the world about “Venetian Restaurants,” I won’t automatically be #1 for that query. But if I create a badge for each of the top 50 restaurants in Venice to put on their websites with a link back to mine, then I’ve incented them to link to me [and Google will reward that].
Another way to think about this is through creating embeddable content, like Slideshare, Vimeo or Scribd. Everyone that posts a presentation to Scribd gets a piece of code to post to their site that includes a link back to that document on Scribd’s site.
Great products and great content are not enough. You need a hook. You need an incentive to get people to link to it. If you can’t create a hook or give people a reason to share your content, you’ll lose out to someone who can.
How would you describe SEOmoz?
We’re a software (SAAS) company that provides tools to make SEOs better at their jobs.
How many customers do you currently have?
Over 5,000.
What does the future hold? For Google? For SEO? For marketers?
If you look at what Google’s doing with social search results (results from my social circle,) real-time results that pull in Twitter, Google Buzz, etc. [you’ll see that] they’re trying to think about sharing activities beyond just a link on a web page.
1-2 years from now, we’ll look at Google results and realize we can influence them through Facebook, Twitter, StumbleUpon, and other social media platforms, similar to the way we do through links today.
It becomes less critical to focus on methodology of how marketers can share content (ie, by generating back-links with keyword-rich anchor text) and focus on whether or not they [share content] at all. Don’t worry about telling people to link to your page by saying “Venice Restaurants” rather than tweeting the page, just get them to share it. [Google will take it all into account.]
In some ways, this will reduce the complexity of SEO, in some ways it will increase it. Early adopters will be rewarded when it comes to creating sharing activities.
