Machine learning is officially a component of Google’s search algorithm. Machine learning you guys! I’m really excited. Google recently announced that RankBrain will be one of the main factors in deciding which links to display in SERPs for a specific query. Here’s what you need to know about RankBrain.
Any news surrounding Google updating their search algorithm is often accompanied by mixed feelings of panic, frustration and a whole lot of confusion. SEO, webmasters, and marketers scramble to maintain their rankings and protect their domains against any possible penalties.
This new development, however, isn’t a cause of worry for you, in fact as an SEO and a data science enthusiast, use of machine learning in search is a great news for search users like me and you.
What is RankBrain?
In case you’re unfamiliar with RankBrain I’ll give you a quick review. RankBrain is the new addition to Google’s search algorithm called Hummingbird. Just like previous updates Caffeine (2010), Panda (2011) and Penguin (2014) this new update also makes the algorithm smarter by enhancing how Google finds relevant sites and delivers search results.
So RankBrain is an update but it’s still a little different than its predecessors. Where Caffeine, Panda, and Penguin had a direct impact on your sites ranking in Google SERPs, RankBrain does not impact how links are ranked by Google. Instead, it focuses on what results are displayed.
What is Machine Learning?
Machine learning is developing algorithms that can detect patterns from data and as a result, learn to improve their response by making predictions based on the data available to them. Sadly, machine learning isn’t the same as Artificial Intelligence.
A machine can qualify as an AI if it can exhibit intelligent behavior and a human-like ability to learn. At the moment we’re far from it but if you need to kick your imagination into gear, I highly recommend watching Ex Machina.
So RankBrain is not AI which means that caring and feeding of Google’s core search algorithm still falls on the shoulders of its capable army of human engineers. RankBrain is a machine learning algorithm that, according to Google, since its introduction has been one of the dominant factors is picking which results to display based on a query.
What does RankBrain really do?
RankBrain’s sole responsibility is to help Google deal with the 15% queries it processes every day that are completely unique. These are the questions, terms, and combination of terms (long-tail) that no one has ever been entered in Google search before. Just 15% Google’s daily search volume that’s a really high number (over 450 million searches per day).
RankBrain learns what people might mean when a unique query is entered into Google’s search bar by building connections or vectors between related terms. As more and more of these connections are built, the algorithm gets better and better at understanding the intent behind the query.
Let me give you an example,
Earlier in this article when I couldn’t remember the name of the movie, Ex Machina, I searched with the following query – ‘the movie with the AI girl’ and voila there it was!
What does it mean for search?
In a nutshell, it means the search is growing up and by merging machine learning with currently existing algorithms, Google is strengthening their case for expanding the role search and other Google technologies will play in our connected lives across multiple mediums and devices.
In an article, I wrote last week I put a lot of stress on starting your SEO or content marketing projects by first understanding the role of search in information creation and delivery. Indeed the increased importance of micro-formats and an ever greater presence of Knowledge-graph related search results are all pointing towards Google’s long-term commitment towards enhancing the search experience.
As SEOs and content marketers, it’s time to let go of keywords and start exploring how our own audiences create and share information.
In Google’s own words,
This [machine learning] has a very broad range of potential applications: knowledge representation and extraction; machine translation; question answering; conversational systems; and many others.
We’re open sourcing the code for computing these text representations efficiently (on even a single machine) so the research community can take these models further.
At the end of the day, as search users, for us, this development means more relevant results for our sometimes obnoxious queries whether we’re on mobile, desktop and screaming OK Google while driving.