Google announced a new signal to their current algorithm called RankBrain in early 2016. RankBrain is a machine learning technology that helps provide meaning to natural language queries to deliver better results. RankBrain started to roll out in early 2015 and has been fully deployed globally. While Google has not provided any statistics on how many queries would be affected, they did state that a “very large fraction” of the queries are being run through RankBrain.
As one of the newest parts to the Google algorithm, Google considers RankBrain as the third most important factor in ranking web pages. They mentioned that the learning done by RankBrain actually occurs offline. RankBrain is being feed historical searches and uses “vectors” to create mathematical entities associated with the words in order to solve complex or ambiguous searches. It will understand patterns between unconnected complex search queries and learn how these search queries are actually similar to each other.
RankBrain will not run against all searches. It will be used to provide better search results for more natural language “long-tail” keyword search queries. Here is a short Bloomberg Business interview with Greg Corrado, a senior research scientist at Google.
Our take is that Google is continuing to position itself to handle more voice queries from mobile users. In a recent report Google noted that “more Google searches take place on mobile devices than on computers.” Plus, they need to keep Google competitive with Apple’s Siri.