Bill Slawski is the Director of SEO Research at Go Fish Digital, and he’s made a name for himself by reading, interpreting, and blogging about Google patents. With a background as a lawyer, he has a history of dissecting complex technical documents and surfacing key insights. He has been writing about these types of insights for years on his personal blog, SEO by the Sea.
Go to his site and you’ll often see intense diagrams, like this one.
He’s certainly adept at translating this sometimes-esoteric information into content strategies that bring real results. He shares some of those stories on today’s podcast.
The highlights:
The insights:
Right now, Bill is studying a patent called Searchable Index.
“One machine learning model Google is using right now is one that one of the inventors worked on: Sibyl. Sibyl was used on sites like YouTube to give them a better grasp of it. YouTube gets indexed by looking at text on a video page: the title, the description, and so on.
So it made sense for them to start watching the videos and learning what they’re about. They’re using machine learning to do that.”
Bill gives the example of a quote from the movie Apocalypse Now: “I love the smell of napalm in the morning.”
He says that when someone searches that quote, they aren’t looking for information about who said the quote, or information about the quote.
“They want to display the video.”
This patent helps Google find and display the video.
He points out that Google used to have a very rudimentary algorithm and in part trained us, the searching public, how to search in very simple ways.
These days Google is far more complex and is able to draw relationships between different entities in ways that it didn’t in the past. Thus search can be a little less precise.
“Imagine the web as a huge scattered database of information, and some of that information is hard to get at. It’s stuck in a PDF file. It’s not easily transferable because we can only read the information on some PDFs.
We can’t necessarily extract the data from them. We can’t do extensive queries like: how many people in the US today are addicted to painkillers?”
The web’s not going to know the answer to that.
“But there may be some sources on the web like the Bureau of Justice statistics that may come close to knowing some information about it.”
He says that semantic SEO addresses this problem by using schema to providing key-value pairs on pages.
“We’re using it in local search to provide address information, to provide geographic coordinates for places, to tell us what’s nearby and what’s far away. Which companies are owned by other companies, who the CEOs might be, and so on.”
Schema, Bill says, helps SEOs describe what the data is, allowing us to play a role in making the web more searchable.
Bill uses an example of these value pairs and how targeted related entities can improve a page’s ability to rank.
For example, once Bill worked for the Baltimore Visitor and Convention Center. They wanted to optimize a page for Baltimore Black History.
“We wrote a page. We included the phrase Baltimore Black history on it three or four times. We tried to rank the page and after 3 or 4 months we were ranked about 114.”
Time for a new strategy: writing a walking tour of Baltimore, with all of the sites which relate to Baltimore’s black history, like the 9-foot statue of Billie Holiday, or Frederick Douglass’ six townhomes.
The page ranked for all kinds of long-tail keywords, but ranks for Baltimore black history, too! All because each of the entities described is relevant to the subject matter they were initially trying to rank for.
This worked way back in 2005, and Google has only become more sophisticated since then.
He later used a similar tactic to help an Arlington, VA apartment complex rank.
Bill says that local search was “sort of a proof of concept of how entity search worked.”
He describes some of the ways that local SEO evolved.
“The guy who worked on supervising engineers at Google for entity search was Andrew Hogue. He was in charge of what he referred to in his resume which he had online, as the annotation framework.
The annotation framework included people like Daniel Eggner, who worked for Google Local Search. His idea was to build a business directory. Instead of a Yellow Pages, you can turn to Google, look a place up, see how close it is, what else is available, look at reviews.”
He gets a little more granular with this.
“Google contains things like data wrappers. They were a template full of fields. An address, street address 1, street address 2, city, state, and a zip code in a certain format. Google was able to identify the format. You’re doing a key-value pair type thing where Google was saying: this is a business address.“
Bill is especially concerned about how information flows to people; that is, accessibility to important information such as assessing the number of damages from natural resources or being able to find informational sources.
“Helping people develop a type of technology and learn about it. That’s why I like to see more of a semantic web.”
Want to see what Bill’s working on or thinking about in the SEO world? He’s incredibly active on Twitter.
