I just attended the Inbound Marketing University Webinar “Advanced SEO Tactics: On Beyond Keyword Research” led by Rand Fishkin of SEOmoz, and I must admit I’m feeling a little frazzled. Why? So many of my cherished SEO beliefs have been turned upside down.
Search Engine Ranking: It’s Not All About the Content
The content on a Web site is not nearly as important to SEO as the context that site inhabits on the Web: Which other domains link to it? How many other domains link to it? What is the quality of those links?
According to a survey of 70 SEO experts, your site’s search ranking depends much more on external factors such as inbound links than on onsite content.
Expert opinion: Relative Importance of Algorithm Components in Google
See that purple slice? The content on your page is about 15% of what Google considers when ranking your page. Anchor link text (the green slice) also affects SEO, but inbound links and host domain authority (does Google have reasons to “trust” that your Web site is legitimate?) are by far more influential.
Inbound Links Trump Onsite Keyword Usage
The very large, very bright orange wedge in the following image represents the overwhelming influence of link popularity—the number of other domains that link to your site—on your search rank.
Relative Importance of Algorithm Components in Google - link popularity predominates
What About Keywords?
Aw, shoot. You just spent 6 months researching the best keywords for your site. Now you need to go do some link building. But since you’ve got those keywords anyway, where will they do the most good?
Keywords in your Domain Name Really Matter.
Keywords in H1, H2, H3 tags? Not so much.
So you’d think that if site content is not the most influencing factor in your search rank, it should be second, right? Nope. That honor goes to your domain URL. Keywords in your domain name (www.yourdomainname.com) exert nearly twice as much influence on your search rank as keywords in site content.
Correlations between Web page data and search ranking
When it comes to site content, your title is one of the most important elements on the page, so be sure to use keywords as close to the beginning of the page title as possible. (Good: Used Cars for Sale | Thomson Motors; Not As Good: Thomson Motors | Used Cars for Sale).
Heading elements exerted little to no influence on search rank, according to Fishkin’s data. That doesn’t mean that you should stop paying attention to headings. Fishkin recommends sticking with best practices. You are creating pages for people, not search engines, and people appreciate a clear, descriptive header.
Alt Text Rules
I was very surprised to hear that high search rank is closely correlated to the use of alt text on images. The alt attribute contains a brief description of an image. When you view a screen with images disabled, you’ll see the alt text in place of the image. Search engines are blind and deaf, so the only way they can recognize an image is by its alt text.
Alt text is revealed on a Web page when images are disabled
One webinar participant asked if this correlation was due to the fact that anyone who includes quality alt attributes is already on the ball when it comes to SEO, and Fishkin conceded that this might indeed be the case.
In any event, it is always a good idea to include alt text for images. Purely decorative images (such as spacer gifs) should contain a null (empty) alt attribute.
Key Takeaway 1: Focus on Link Building, but Don’t Abandon Best Practices
I know that I, for one, will be redoubling my efforts at link building for SEO. It’s important, however, to continue dotting those i’s and crossing those t’s in our Web development practices. Code correct HTML markup, write good heading text, and concentrate on delivering high-quality content to your audience.
Key Takeaway 2: Links Trump Content, but Content Earns Links
(or, Great Content Is Useless Unless It Compels People to Link to It)
This clever tweet came from Rena Bernstein during the IMU Webinar
As rattled as I was to hear that my precious H1 elements were not sacrosanct, I took comfort in the knowledge that unique, substantive content is still the name of the game. Without useful content, there’s no reason for people to use the Web at all. The key is to keep focus on both content and context. We want to create great content that results in other people discussing it and sharing it. The links generated from that online discourse improves our search rank.
Key Takeaway 3: No Site Is an Island
A large part of how easy you are to find on the Web is reliant upon your relationship to others. Are they interested in you? Do they talk about you? Do they link to you? When someone runs a Google search using a keyword, Google looks for the best, most relevant results to deliver. One way it decides if your Web page is a good-quality search result is how popular it already is. If others found your page useful, then the searcher might as well.
See the Presentation
If you care about SEO, you need to see this presentation. The webinar has been archived and is available for on-demand viewing (You may have to register in order to see it). You can view the slide presentation on Slideshare.
[...] nicht genannt. Ich denke, das sollte von Rand auch nur ein kleiner Vorgeschmack sein. Hier noch ein Artikel über seine Präsentation, von der Rand selber twittert “Great write-up on my IMU [...]
[...] der Rankingfaktoren könnt ihr bei dem Kollegen von SEO-Book lesen, der das ganze aus dem englischen übersetzt [...]
I don’t think this data makes any sense when taken individually like this. I’d want to see Fishkin’s data before jumping to any conclusions, but I have an incredibly hard time believing that user-hidden fields (meta, alt text) have a higher weight than structural markup that the user interacts with (H1, et al). Google is trying to determine what the page is about, which is derived from what the visitor sees. Meta tags and alt attributes are prime locations for keyword stuffing … and I would think are inherently less trustworthy.
Taking the aggregate of a Google crawling session gives these results a little more validity, but they lack the sensationalism. Inbound links are certainly a major part of Google algorithm. We’ve known that for years as a result of the “Google Bombs” for “miserable failure” and the like. If inbound links are the keystone to the algorithm (and I’d guess they are), then I can understand how high correlation between inbound link words, meta keywords, alt tags and the domain provide Google with the extra detail to rank a page higher. (I’d argue they also show a greater burden of maintenance, possibly indicating a higher quality of site, but I’m a purist
That doesn’t make alt tags *more* important than H1 (after all, without the H1 and P tags, the outside world would never have linked to this page in the first place).
Here’s the experiment: create two pages, one with keyword loaded, semantically coded (H1, H2, P) content and no meta / title / alt attributes, and the other with keyword-rich alt attributes and meta tags but no content. Link to both pages the same number of times from the same quality of sites. If the no-content site wins, Google has a problem on their hands, right?
I think your wrap up pretty much says it all. Work on the content first, and get inbound links in any (ethical) way you can. Code your pages using the semantic markup HTML provides (that goes from H1 to title attributes in links). Now you’ve got accessible content in addition to content Google’s going to love.
Interesting post … definitely has me thinking …
nz
Great insights—thanks! I am guessing that the high correlation between the elements I found surprising (alt text, for instance) and search rank has more to do with the fact that a developer who pays attention to the details is doing a lot of things right (including adding high quality alt text). My concern is, and always has been, losing sight of the needs of the human user in pursuit of Google Glory.
The archived Webcast is now available: http://event.on24.com/eventRegistration/EventLobbyServlet?target=lobby.jsp&eventid=146656&sessionid=6&key=DEE3CCD13D0B78A48E511E9A62636FD5&eventuserid=25935508