As an SEO, I've grown accustomed to frequent news about updates and other changes made to Google's algorithm. But today's news from the world's leading search engine took me for a loop.
Here at Wikimotive, the idea of machine learning and AI as a part of future search results has been something we thought of as semi-futuristic. We knew Google was dabbling in machine learning AI, but assumed the technology was years away from being tested. But it turns out it's already here, and it's called RankBrain.
To quickly get you up to speed on this new search technology, we've put together five quick facts every SEO needs to know:
We all knew it was coming. The robot takeover is here, and it's all Google's fault.
With the RankBrain system, Google has officially put search results in the hands of a machine-learning artificial intelligence system. Unlike the search algorithm it's a part of, which requires updates from Google's search engineers, RankBrain is able to learn and better supply results related to completely unique searches.
This means it's able to give users the best results possible for longtail and conversational search queries.
In 2013, the Hummingbird algorithm brought better conversational search results to Google. This meant you could ask Google questions and get accurate results instead of formatting your query to match what Google understood.
Hummingbird was able to better understand words and how they were connected in order to supply quality and relevant results for many longer queries.
RankBrain builds on to what Google started with Hummingbird, but is able to detect patterns and connect dots that were unable to be detected by the non-AI algorithm. Because of this, Google can now learn and begin to connect unique searches with others in order to provide better and better results for those that use long and conversational queries.
As a company, Google is made up of some of the world's best and brightest software engineers. You'd think that all of that time spent creating and tweaking algorithms and updates would give them the ability to outsmart a new AI like RankBrain.
Unfortunately, that's not the case.
In one experiment, the company put its search engineers head-to-head with RankBrain. They were asked to look at certain webpages and make a guess as to which page would rank at the top of Google search results. The engineers were correct 70 percent of the time, but RankBrain was correct 80 percent of the time.
I'm sure they all had questions about job security afterward.
While the exact number hasn't been made public, Google did announce that RankBrain is an active part of the process that serves a "very large fraction" of its search queries.
At this point, there's no word of any reported traffic effects from RankBrain, which has reportedly been going through a rollout process since early 2015. As we learn more about the technology, though, it's expected that SEOs will put together the pieces of the puzzle to determine where the changes are being made, as well as the fallout from said changes.
Because RankBrain is a part of Hummingbird, it acts as just another one of the hundreds of signals used by Google to rank pages and serve results. However, Greg Corrado, a senior research scientist at Google, told Bloomberg that RankBrain has already become the third-most important signal for serving search results.
So despite being new technology, it's clear Google is betting big on RankBrain.
There are still plenty of questions on the table about RankBrain, especially amongst SEOs. Wikimotive will keep a close eye on all developments involving RankBrain and report new findings as they come in.
Originally Published to Wikimotive.com on October 26, 2015.