Google officially announced the use of its RankBrain algorithm in late 2015. At the time, it was used for only 15 percent of searches processed by Google. As of mid-2016, it was confirmed that RankBrain began to process 100 percent of all Google searches. It has grown to become the third-most important ranking metric, yet many people have no idea what it is.
RankBrain is labeled as a “machine-learning artificial intelligence system,” but that doesn’t help understand what it does. If anything, it causes it to sound confusing and mysterious. Let’s consider what the algorithm actually does in simple terms.
RankBrain’s primary function is to interpret new or ambiguous searches. At the time of the algorithm’s release, Google stated that around 15 percent of their daily searches were terms or phrases that had never been searched before. It was difficult to provide accurate results for those searches.
Google handles around 3.5 billion daily searches on average. This means that more than 500 million daily searches are new or ambiguous phrases. RankBrain is designed to figure out what those phrases mean so that Google can provide the best possible results to the user.
The actual “machine learning process” takes place offline. Google engineers feed the algorithm batches of ambiguous search phrases. It creates links and search results based on those phrases. The engineers then verify those results and make refinements. Each time the algorithm is updated, it will have processed more batches, created more links, and have been refined by human testing. Therefore, it is always learning to provide the most accurate results possible.
Business owners and marketing professionals are still struggling to learn how this algorithm will impact their SERP rankings. One common theory is that RankBrain places an emphasis on user signals. In the past, links and content were enough to stay at the top. Now, most top-ranked sites have high levels of engagement. This may be the machine learning algorithm at work.
As time progresses, Google will certainly reveal more information regarding how this algorithm operates under the hood. All that we know for now is that it helps interpret new search queries and that it is having a significant impact on SERP rankings.