Really good read if you are interested in on-page SEO and semantic search - https://www.link-assistant.com/news/optimizing-for-semantic-search.html Summary Masha describes how Google uses TF-IDF algorithm to determine keyword frequency - this is common knowledge I think but its interesting, basically how many times a keyword appears on a page vs how many times it appears elsewhere on the internet. There was also a discussion a few days ago on BHW about keyword frequency on a per niche basis, and I wonder if Googles algorithm will calculate the density of keyword on a page, vs density on the web as a whole vs density of the keyword on other sites in same niche (Like working out a mean average for the industry / all results for a keyword). Also Masha talks about how Google are focusing now much more on the intent of a user when searching for a phrase rather than the actual phrase itself. Also, importantly Google groups together lots of different phrases, keywords and so on for each search and then they have a big list of keywords and phrases, people, places that are relevant to a topic and then use that information when analysing a page to determine how relevant it really is to any particular phrase. A good example of this in the article was - if a page was about New York but it didnt mention the term Brooklyn then Google wouldnt consider it very authoritative on that topic, even if it did mention the term New York 5 times per 1000 words or whatever. Anyway, its definitely worth a read.