Google uses a machine learning technology called RankBrain to help with search results, but much of how it works is shrouded in mystery. While much is still unknown, Google just opened up a bit more on how it works.
You can read more about this story on this post at Search Engine Journal.
Subscribe! You can subscribe to this podcast on Apple Podcasts, Anchor, Google Podcasts, Spotify, Breaker, Pocket Casts, RadioPublic, Stitcher, Overcast, and a variety of other places. Be sure to rate the podcast wherever you choose to listen!
Back in 2015, Google announced a new machine learning based search algorithm called RankBrain. If you’re not familiar with Rankbrain, the Wikipedia entry includes a nice summary: “If RankBrain sees a word or phrase it isn’t familiar with, the machine can make a guess as to what words or phrases might have a similar meaning and filter the result accordingly, making it more effective at handling never-before-seen search queries or keywords.”
Google has said that roughly 15% of the searches on their site — which amounts to 500 million searches per day — are phrases that have never been searched before. Having RankBrain to quickly help understand those searches is very important for Google.
A user recently asked Google’s Gary Illyes a bit about RankBrain, saying “Lots of people keep saying that part of the RankBrain system includes UX signals, including Dwell Time, Bounce Rate, Click Through Rate, etc. As I understood it, RankBrain was about trying to fathom what results to serve for unrecognized searches. Can you please confirm/deny whether RankBrain uses UX signals of any kind?”
Illyes responded with a variety answers that mostly answered the question. He said:
“RankBrain is a PR-sexy machine learning ranking component that uses historical search data to predict what would a user most likely click on for a previously unseen query.”
He also added: “It is a really cool piece of engineering that saved our butts countless times whenever traditional algorithms were like, e.g. “oh look a “not” in the query string! let’s ignore the hell out of it!”, but it’s generally just relying on (sometimes) months old data about what happened on the results page itself, not on the landing page.”
And lastly he said: “Dwell time, CTR, whatever Fishkin’s new theory is (Fishkin meaning Rand Fishkin, who founded Moz), those are generally made up crap. Search is much more simple than people think.”
Things like RankBrain are why it’s important for Google to really understand your content. They’re having to serve up answers for queries they’ve never seen before, and clarity will always be to your benefit when it comes to SEO.