Google’s March 31 2005 patent application reveals some details of Google’s focus on quality.
The Description section reveals some current and future Google likes and dislikes. The patent concentrates on ways to deliver quality search results. Below are some inferences I take away from reading the patent.
Google rewards (or reserves the right to reward):
- updated content (with the monkey wrench thrown in that sometimes older documents are better), especially for FAQs
- new unique content
- content that Google has successfully matched to other users’ search terms
- anchor text (the text the reader sees in a link)
- links from “independent peers”
- a high and probable rate of back link growth
- updated outbound link anchor tag text
- an accelerating rate of document changes
- time-based relevancy (search for “world series champion” in 2005 yields different results than searching for “world series champion” in 2006)
- an accelerating rate of incoming links
- the trustworthiness of the site hosting the inbound link; trust is given to known entities (the government, Yahoo) and authoritative sites (outlined in yet another patent)
- an accelerating rate of document traffic
- entering a time period (e.g., “summer” or “weekends”) when the document has proved more popular
- documents containing advertisements to trusted sites (e.g., Amazon)
- documents containing advertisements that receive high click-throughs
- documents with high stickiness
- documents hosted on domains registered for ten years rather than 1
- documents hosted on name servers that post quality DNS whois info
- documents hosted on name servers that host a variety of domains
- documents hosted on established name servers
- documents that slowly rise in quality ranking scores
- documents with accelerating bookmarking activity
- documents with accelerating cookie activity
- documents with accelerating cache activity
- varying anchor text in inbound links
- inbound links from high scoring sites
Google ignores (or reserves the right to ignore):
Google penalizes (or reserves the right to penalize):
- stale documents
- spam
- links from dependent sites
- new documents with a spike in inbound (aka “back”) links
- a document that “hits” on widely varying queries
- aging links that point to your content
- artificially-dated content
- a high but improbable rate of back link growth
- a decelerating rate of document changes
- a decelerating rate of incoming links
- content that becomes inconsistent with the anchor text of its inbound links
- a decelerating rate of document traffic
- entering a time period (e.g., “summer” or “weekends”) when the document has proved less popular
- documents containing advertisements to lest trusted sites (e.g., porn sites)
- documents containing advertisements that receive low click-throughs
- documents with low stickiness
- domains that serve doorway (spam) inbound links
- domains registered for one year rather than 10
- low quality DNS whois info, including physically incorrect address information, constantly changing contact information, use of several hosting companies
- name servers that host mainly porn, doorway, commercial-word-domain-names, bulk domains from a single registrar
- new name servers (in combination with items from the bullet above)
- documents that quickly rise in quality ranking scores
- documents with decelerating bookmarking activity
- documents with decelerating cookie activity
- documents with decelerating cache activity
- no variation in the anchor text in inbound links
- inbound links from low scoring sites
- an increase in the number of topics that make up the document
{ 0 comments… add one now }
You must log in to post a comment.