The Learning Center
AKA... The Blog
AKA... The Blog
To better understand what it is, we’ll first pose the question: Why do we need it? Google has always relied on humans to refine search queries. Programs and software help, but down the line queries always land on human eyes. Google employees create synonym lists, known as stemming lists, so Google knows when you type in “walking trails” you may mean hiking trails. The lists that are in place are just fine, but when queries that Google has never seen before come up, humans usually need to check on these with the possibility of updating the stemming lists.
The problem is 15% of searches on Google are new. Google handles an average of 3.5 billion searches a day and 1.2 trillion a year. I’ll save you the math, 15% of 1.2 trillion is over 4 billion. That’s 4 billion new searches a year that need to be refined. Humans need help. The answer is RankBrain which Google first rolled out in early 2015. While still being somewhat new, it is different from when it first rolled out. RankBrain is now being used for every search query Google receives. Initially it only factored into that 15% of never-seen-before queries.
RankBrain initially handled new queries which Google had never seen before. It would take these new queries for which Google had less relevant data and would replace them with queries Google had seen before and therefore had better data for. Ultimately it allowed Google to serve results they felt more confident in.
Today, RankBrain does not do this for new searches. It does this for all searches. For example, if you were to search “best pizzerias in Washington D.C.” RankBrain might determine your query is similar to “best pizza in D.C.” If Google has more data on “best pizza in D.C.” RankBrain would substitute this for your query, and you would see results as if you had searched for “best pizza in D.C.”
It is important to note that RankBrain isn’t actually moving the search rankings around. It did influence the results, but it did not change the order of results based on a specific ranking factor such as links or content. It simply substituted one query for another.
So how is RankBrain the third most important ranking factor if it doesn’t actually influence results? This is where it gets a little confusing because Google has confirmed that it is a ranking factor. It’s likely that RankBrain helps classify pages based on the page’s content. However, Google hasn’t stated this. All we know is RankBrain is a factor in rankings. If you’re worried, the best thing you can do to ensure rankings is the same as always: focus on great content.
RankBrain uses machine learning to teach itself without the aid of humans. Don’t worry RankBrain isn’t Skynet. It is not constantly learning, in fact all learning is done offline. It learns to make predictions on how to refine queries based off of batches of historical query data. I guess you could say someone at Google “feeds” RankBrain this historical data, RankBrain learns from it, these predictions are then tested, and if proven good then go live. This process then repeats. So it’s not completely in control. Google is watching it to ensure predictions are reliant.
Again, you shouldn’t be worried about RankBrain taking over the planet, but if you are worried about your website’s rankings, the answer is simple. Focus on great content. Great content, the second most important ranking factor, can lead to traffic and more importantly links to your site, the most important ranking factor. As always, remember there are tons of other factors and having good standing in all them is the best thing you can do.