SEOmoz’s Rand Fishkin recently published a graph of his opinion of how Google’s ranking algorithm has changed over time. What did this graph say, and what does it mean?
The graph demonstrates that the importance of PageRank has dropped significantly since 2002 and that the trust and authority of the web site (as viewed by Google, of course), is significantly more important since 2002.
To me, this makes total sense. Incoming links used to be very important to SEO success, but site owners figured out that they could game the system by buying links. By placing more emphasis on a site’s trust and authority, Google attempts to reduce the ability to game the algorithm.
So what constitutes trust and authority? Perhaps…
- Links from news sites
- Links from government sites
- Links from educational sites
- Links from quality blogs
- Links from Wikipedia, Technorati, other “authority sites”
- Age of your web site
- Customer reviews on authority sites
- Amount and depth of helpful content
- Perceived usability of your web site
What do you think makes a trustworthy, authoritative site?
April 20th, 2009 at 5:42 am
Vijay R says:
Some things don’t gel with the original post. First, the data to back up these factors is just not there – Rand only says that its his opinion. Second, PageRank, the algorithm, *is* a means of determining authority and trust based on graph theory.
Some of the things which have not been considered, which may be very important (I speak as a computer scientist who has a little familiarity with information retrieval theory).
* possible enhancements, again based on graph theory, which Google may be using. For instance, use detection of cliques to find spammers.
* Google has a ton of user feedback, in terms of clicks on links. They are almost certainly using this feedback loop to (machine) learn relevance and authority. More recently, their wikifying of their SRP could be seen as an attempt at getting similar kind of feedback.
* Region/demographic specific enhancements. Demographic may become very important given their new initiatives in getting people to register with Google (gmail, etc.)
* Semantic analysis of data to “learn” from signals in the web page, user queries, etc.
Vijay