Sometimes when you set off on an internet search you know exactly what you’re looking for. Googling “Mark Zuckerberg net worth” or “2015 home run record” will likely provide you with nice round figures that you use to settle a quick bet with your friends. Google has gotten pretty good with these types of searches that use an economical number of search terms to zero in on a very specific result. Google’s Knowledge Cards are a reflection of this anticiapatory nature of Google’s algorithm. Even if you misspell your search term as “Gorge Cloonie,” you’re likely to get a Knowledge Card filled with facts about the actor George Clooney like his birthdate, filmography, and more.
Google has been getting increasingly better at fixing your mistakes like this and providing a pretty good guess for what you’re searching for, especially when the query consists of commonly searched-for terms. But over the past few years, Google has algorithm updates have gotten better at finding the relationships between words used in search terms. These updates help the search algorithm do better when it’s faced with combinations of search terms that are seen less frequently. For help with these unique queries, Google is looking to artificial intelligence to help it find the best matches.
Nicknamed RankBrain, the new AI component to Google’s search algorithm has been handling a “very large fraction” of searches for the past few months, according to senior Google research scientist Greg Corrado. RankBrain turns written language into mathematical vectors that the AI computer can understand. The artificial intelligence kicks in when the algorithm encounters a word or phrase that it hasn’t seen before. When this happens, RankBrain makes a guess at what the phrase means and associates it with another, recognizable phrase to point the searcher in the right direction. The Google algorithm uses hundreds of “signals” to determine what to display on the search results page, and RankBrain is now one of those signals. In fact, it’s now the third-most important signal in the algorithm, according to Corrado.
Google wants you to know that so far, RankBrain is performing well. Google asked their expert search engineers (who write the code that the rest of the algorithm relies upon) to take a look at some web pages and provide their best guess about which pages that would likely end up on top. The engineers were right 70 percent of the time; RankBrain was correct 80 percent of the time. Corrado says that RankBrain is now so indispensable to accurate searches that removing the feature “would be as damaging to users as forgetting to serve half the pages on Wikipedia.”
So what does this mean for SEO? First of all, RankBrain is not taking over for Hummingbird, the most recent update to the Google search algorithm. RankBrain isn’t replacing Hummingbird, instead it is just becoming a part of it. Other parts of the algorithm include names that will be familiar to those in the SEO field, like Panda or Penguin. Search Engine Land has a good breakdown of some of the other important parts of the algorithm: “Payday [is] designed to fight spam, Pigeon [is] designed to improve local results, Top Heavy [is] designed to demote ad-heavy pages, Mobile Friendly [is] designed to reward mobile-friendly pages and Pirate [is] designed to fight copyright infringement.” In other words, RankBrain shouldn’t change your entire SEO approach, but it is one more thing to take into consideration.
Keep in mind that RankBrain was designed to be used with the algorithm that Google’s coders (who are very much human) have been writing and tinkering with over the past couple decades or so. If RankBrain works as Google wants it to work, that should mean that its artificial intelligence is built to learn how to simply make that algorithm work better. If RankBrain helps searchers find pages that Google engineers would have you find anyway, the same SEO tactics that have worked in the past should continue to work with RankBrain-assisted searches. As Forbes’ Mike Templeman puts it “The increase in machine learning technologies will make the algorithm more agile and will begin to reward the marketers who are doing SEO the right way more and more. So, it’s a great day for people who practice SEO the right way and often get frustrated that old spammy tactics still produce results for their competitors.” So, content remains king, as do inbound links and proper website structure, and proper tagging, and so on.
So even if the implementation of RankBrain doesn’t mean that everything you know about SEO is changing, it is still worth noting that Google freely admitted that RankBrain is now the third-most important factor that goes into ranking pages. But this begs the question, “So what are number one and number two?” Search Engine Land tried to pry that information from Google, but they came up empty: “Google won’t tell us what the first and second-most important signals are. We asked. Twice.” Search Engine Land’s Danny Sullivan believes that revealing RankBrain to be the number three factor is Google’s way of letting us know about its innovative AI breakthrough while still keeping most of the algorithm’s nuts and bolts shrouded in mystery. But that doesn’t stop Sullivan from taking a shot at it: “My personal guess is that links remain the most important signal, the way that Google counts up those links in the form of votes.” And what about the second most important factor in determining search results? Sullivan says, “I’d guess that would be ‘words,’ where words would encompass everything from the words on the page to how Google’s interpreting the words people enter into the search box outside of RankBrain analysis.”
If Sullivan is close, then your SEO strategy shouldn’t need a RankBrain overhaul. After all, if links and content are still among the most important elements for ranking, and if the third-most important element learned everything it knows from the other elements, the quality SEO practices that worked yesterday should continue to work with the implementation of RankBrain.