If you’ve never heard the term “phrase-based indexing” then let me give you a quick primer. The concept is based on clustering. Google’s search bot will go out and analyze pages that use certain phrases together. For instance, “baseball” and “home run”. Add to that the key phrases “base hit” and “strike out” and you’ll start to get a picture.
If a large percentage of pages on the Web that discuss baseball also use the other key phrases in the cluster then you have a high value set of phrases. Google can use this information for several purposes.
One purpose is to identify phrases that are popular among spammers. If the search engine can identify those then the spam can be filtered from the search results. But another purpose is to use the phrase clusters themselves for ranking purposes. A web page that successfully incorporates the cluster of phrases into its content could rank higher for the initial key phrase – “baseball” for instance – than a web page that simply re-uses the initial key phrases over and over again attempting to achieve the right density.
This is particularly telling because what it is really saying is that natural language writing is preferable to stilted keyword-based writing. The search engines have been treating content this way for several years. Understanding this will improve your SEO.