Google’s Latest Change Isn’t Human
Last week, Google announced their Q3 results: and they were outstanding. Wall Street took notice of the top-line results, and Google stock saw a large bump in value. However, keen eyes noticed that Google management was touting something that search-industry people probably weren’t expecting to be in an earnings call. What they were touting was “RankBrain.” This is the first instance of Google acknowledging this combination of artificial intelligence and machine learning as it involves their core search business. Several people have been talking about Machine Learning and Google, notably Moz’s Rand Fishkin.
In the past, discussion of this type was in the abstract. Today, the news is that RankBrain is powering Google’s ranking of a number of search queries. It turns out that the results have been in testing for over a year, and it was rolled out earlier this year. Google is constantly feeding it new information to keep the learning current and updated.
How does Google use RankBrain?
An eye-popping 15% of searches have never been seen by Google before, and RankBrain is used to answer those queries. The indication is that instead of reaching maturity, search queries are continuing to expand. Previously, Google stated that roughly every quarter that 20% of search queries that happen they had never seen before. Whether it is 15 or 20%, it is a high number, and RankBrain improves the way unexpected new queries are handled.
Google goes on to say that it is still one of many signals (hundreds, actually) used to determine search results.
What is most pertinent is that Greg Corrado, a senior research scientist with the company, says that “in the few months it has been deployed, RankBrain has become the third-most important signal contributing to the result of a search query.” Google’s testing revealed that users gave the thumbs up to the RankBrain results. The tests also show that RankBrain did better than the engineers at Google in predicting the results of search queries (70% to 80% correct estimates).
Expect more news in the future as the search industry tries to determine what this means. If you allow me to venture a guess, I would say accuracy and authority are even more important than ever. There are no specific mentions in the news about RankBrain regarding Local SEO. If we assume the goal is results that benefit the searcher, then the importance of location is still a huge factor in mobile searches — probably even more so than machine learning.