Google published a new podcast episode where John Mueller, Gary Illyes, Martin Splitt and a guest from the Google search quality team named Dewey spoke about how the search company tackles search result spam and how Google ranks search results.
The new episode can be found here or listened to on YouTube.
How Google ranks search results
It is always interesting to listen to a Google representative talk about how Google search works. And in this latest episode Gary Illyes from Google went deeper on how Google ranks its search results.
The short version is that Google first comes up with a short list, let’s say around 1,000 results, for a given query. That list is generated based on if the query and the content on a page are relevant and topical. Once the list is generated, Google then applies a lot of its ranking signals and factors to that shorter list. Gary Illyes said that is where “the magic” happens.
Gary Illyes explained that these documents are assigned scores or numbers and Google “assigns a number and we calculate that number using the signals that we collected during indexing plus the other signals. And then essentially, what you see in the results is a reverse order based on those numbers that we assigned,” he said. Some examples of algorithms used are RankBrain and even the HTTPS boost – although the HTTPS boost is a tie breaker and would not ever really rearrange the search results he said.
This part of the discussion in the podcast started at about 28 minutes into the conversation. I highly recommend you listen to it when you can.
Google spam prevention and machine learning
Before Google spoke about how it ranks search results, Dewey from the search quality team spoke about spam preventative measures. One thing he said that really stood out to me is that Google uses machine learning models to deal with the most obvious spam. I guess this shouldn’t surprise anyone, but it was nice to hear Google confirm that.
Dewey from Google said Google uses a “very effective and comprehensive machine-learning model that basically took care of most of the obvious spam.” This machine learning model lets the Google search quality team focus on “more important work,” he said. More important work might be around hacked spam, online scams and other issues that the machine learning models do not pick up.
Google’s machine learning models have years and years of data it uses to improve its spam prevention methods and search and it seems Google is very confident of its abilities.
Why we care.
Like I said above, it is always interesting to listen to Google representatives talk about search. The way they talk about search may clue us into what really matters with rankings. Like how Dewey from Google was saying that often it is sad to see SEOs focus on a single metric, often an external metric that Google does not even use, instead of focusing on making better functionality, quality content, and an overall better user experience for your users. Google has hundreds of ranking signals, so focusing on one or two probably doesn’t give you the best chances to rank well in Google Search.
About The Author
Barry Schwartz a Contributing Editor to Search Engine Land and a member of the programming team for SMX events. He owns RustyBrick, a NY based web consulting firm. He also runs Search Engine Roundtable, a popular search blog on very advanced SEM topics. Barry’s personal blog is named Cartoon Barry and he can be followed on Twitter here.