BILL SLAWSKI

Does researching Google patents help you to better understand SEO?

 

There are times people refer to me as a patent analyst. Mostly, these are editors or those who write magazine articles. I have always reminded them that my purpose for looking at patents isn't to learn about patents. I'm channelling what Google and Microsoft are doing because I am looking at search-related patents. Google is going through lots of evolutions and changes.

 

Within the last year, they've focused more on artificial intelligence and machine learning than they ever had before. I don't have a degree in computer science. I've got an English degree and a law degree, which means I'm supposedly trained in analysis, logical analysis, and reading.

 

People with computer science degrees once told me I was wrong about what a patent says. They thought I could have read the patent better if I read six lines from the patent and what it says. Most of the time, the information is there; it's just that people are too lazy to go and look for it. It may be difficult to understand certain actions of Google. And everyone's shocked, because it studied the patents to get a good idea of what Google's thinking about, maybe they don't do everything.

 

From an SEO point of view, you can start to predict the future a bit better. Google came out with the patent seven years ago and I read it. I knew it would take some time to go through it, but it was really important. Google explored the idea of how to substitute query terms. They wouldn't necessarily look for words that were adjacent to each other, but they'll look at a whole string of words. A full complete sentence is more spoken queries and written queries. I spent five weeks writing or going through the patent, trying to figure out what it says and writing something about it. And the day I decided, Danny Sullivan gets online, and he starts talking about Hummingbird. It can't be about anything else.

 

Does researching Google and Bing patents help you to predict the future of SEO?

 

Earlier this year, Google came up with what is called website representation vectors. Google is using neural networks to classify websites on a level where they are identifying the niche and the level of expertise in writing the content of the websites.

 

They gave an example, and it rang some bells in my mind. If we have a medical website, and it is written by doctors, those are expert writers. If it's written by doctors in training or student doctors, those are like intermediate level experts. If it's written by laypeople, (some medical websites will be written by laypeople) it means the content on the website is by people who do not have expertise on the topic. The questions people ask require different levels of expertise. There are people who ask questions about regular experiences. For instance, ‘what’s a good breakfast?’ on lay person’s website is not asking for the same level of expertise as ‘what are the symptoms of diabetes?’

 

What Google then did was to classify query terms and query logs by using a machine learning approach to identify the knowledge domain of those query locks.

​

So, in addition to identifying the knowledge domain of websites, or identified knowledge domains of queries when people perform a query, Google will try to find an answer to their query from the appropriate classification of the website – be it expertise, intermediate or laypeople’s knowledge domains. This makes it easier and more efficient to search. For instance, if we ask, ‘What are the symptoms of diabetes?’ The information on ‘diabetes’ requires a medical expert knowledge domain query. Thus, they only have to look for answers amongst mental expert websites.

 

One of the things they have been able to achieve is to ensure that before a website is used as a place for answers, they try to see if it passes certain thresholds for quality. They also make sure it meets quality scores. This all reminded me of the medic update and our expertise – trustworthiness. This is because they are looking for signs of trust and signs of expertise. They are also trying to identify sites as being authoritative sites.

 

What made you get into the SEO industry and when was that?

 

It was 1996. I had a friend who told me one day about how much he hated his job. He was working as a service manager and car dealership.

 

He said, ‘I don't want to go back. I don't want to do this anymore’. I replied that I was reading a book about people that was helping people incorporate businesses in Delaware. I said to do this, you have to maintain a mail address, receive notices from the courts that somebody might be trying to see somebody. These are the technical abilities needed. So as an agent, you sell and charge him a certain amount of money per year. If there's any type of lawsuit against them, your job is to receive the notice and you inform them. My friend agreed to do that. We had another friend who was practicing law and was renting. There was an empty office in the house and they talked to each other and the one friend who wanted a new job would work in the empty room in that office. The only problem they had was that they saw the need for a website. However, they could not fix the website on their own.

 

I then bought a book and learned how to build a website in two weeks and learned HTML. After two weeks I got the website and we launched it. Once we got that out there, I started learning how to promote it.

 

One of my friend’s sisters worked with digital Corp. She sent us an email one day informing us add our website to AltaVista. Once we got into that, Google came at us, we said, we need to be in that now. That's when I started doing SEO and search engine marketing, showing up and we saw needs to be in them.

 

How did your love of SEO turn into a business?

 

I was working to promote a website. I joined a forum where small businesses wanted website promotion. We joined and we answered people's questions about promoting websites and forming legal business structures and other related issues. That turned into answering questions about SEO. I ended up having other people ask me for help with websites.

 

So, this small business forum turned into the creation of a site forum and became more popular. Until in 2005, Lauren Baker asked me if it was okay if he posted a job posting on the forum. I asked him what it was meant for. He said, ‘we're hiring an SEO in Maryland’. I said that's about half an hour away. And so, what is it? Tell me about the position because I'm interested. He told me and it sounded good. I applied; it was half an hour away. It was him who gave me a chance to leave the courthouse, leave what I was doing there and do SEO. And I said, Okay, I can do this. This is a good experience.

 

What has been the biggest change in SEO to date?

 

The biggest change didn’t take place overnight. It was a long time coming. We often think about SEO; we think about a link graph. We think about websites connected by links, and search engines trying to follow those links. We find the web pages indexing, and give people access to an index to answer their queries. Google started crawling the web to find facts. There were connections between those facts, relationships between entities on webpages, or between entities and attributes of those entities or classifications for those entities.

 

It was a type of change that led to Google coming out with the knowledge graph. Google engaged in question answering, showing featured snippets or people who also ask questions. So, it's a knowledge illusion when Google changed the name of their search division to a knowledge division in the early 2000s.

 

That was a reflection of a movement away from search as a name. Knowing what those relationships are between those things is knowledge. Google has become more of a knowledge engine than a search engine. The connection is how to locate the proper search by people

 

How does Google find the best website and the best page?

 

It's an ongoing process. So, Google is attempting to analyse the intent behind your query.

 

When somebody asks, ‘What is the hotel in the Middle East? It's shaped like a sail from a sailboat.’ Somebody is asking for a specific hotel. They have one in mind that they want to find. They're searching for an entity. So, Google knows this is a question about knowledge. When somebody says, ‘What are the best restaurants in New York City?’ Google knows that the answer people want is most likely a list of links to the URLs to restaurants.

 

Is Google getting it right with the results they show?

 

I think they make mistakes from time to time. But most of the time they're getting it right. They are constantly evolving these methods. So right now, I'm reading a patent about using machine learning to answer questions. And it's a much more complex method than what they've used in the past - working to answer what we refer to as featured snippets when you change from an algorithm.

 

Well, let me introduce a definition of NAT (Network Address Translation) rule. I've done lots of reading of books about computer science, machine learning and artificial intelligence. So, one author said - “An algorithm is problem-solver”.

 

So often, many of the patents I read start by noting this is a problem that we have now in our industry. This is the current shape of how we solve that problem. It has some problems; it doesn't always work. Well, we could do it differently. Here's how we plan to do it differently. This is our process, and they describe a process that solves that problem. So, the algorithm is solving problems.

 

How does Google understand what the searcher wants?

 

There is what is known as Natural Language Processing (NLP) which is connected to how people talk. The hard part isn't how to understand the way people talk. The hard part is getting the computer to understand the way people talk. That's where the challenge is coming in. So, people could ask a question from a computer. They ask in normal natural language; they appear to be able to understand what somebody is saying and what they're asking for. There are different accents and different parts of each country say things differently, which means the same thing. So, how does Google understand that?

 

It's sort of a two-stage process here. When Google gets a voice query, it often does a speech to text, translation, and then it responds to the text. However, an accent could be considered noise. In some cases, it can be considered information. In other cases, if Google can recognise the accent, Google can personalise the response. So, if Google recognises that you have a British accent, and you're searching from New Jersey. When you're asking about football with a British accent, you might want something different than American football. The same things can be said in 1000 different ways.

 

I mentioned football as an example. I chose that on purpose because there is a difference between what I think of when I say football and what you think of when you say football. And if you're from Australia, and you talk of football, it's a completely different game. So, Google needs to do some type of preferred country biasing when they're answering queries. This implies that the country from which a person is sending queries may go a long way to determine the responses. They may have checked off in the browser that they have a query from a first language user. Consequently, Google will try to provide answers on what may be looking for using the context where the person is from and what their accent is. This means that context matters.

 

What are the top things SEOs need to stop thinking about?

 

The focus of any SEO has to be on critical thinking. If you're going to write about SEO, make sure you have supporting evidence and ensure you can make an argument for yourself. That means a realistic argument has a thesis; it has analysis and has a conclusion. SEOs need to cross all the t's and also dot all the i's. They give a complete answer. Make sure that if you're going to show a list of correlations, and you're not going to leave it at correlations without proving that there's any causation. Don't just leave it there.

 

I've done a study with 9 million websites that show that the top 10 websites have at least 2,000 words long each. This implies that the longer an article might be, the more chance a search engine has to index important content within that article. This means articles that are longer might rank words for more terms. That may not be true, but it's at least an attempt to causation. So, if we can do the correlation study, it is important to understand why that correlation exists if there's a connection. There's a well-known study that shows that convenience stores purposefully place diapers near beers because they found out that the more diapers they sell, the more beer they sell.

 

They did a study and understand that it seems a lot of young fathers or a lot of new fathers would often reward themselves for having picked up the diapers by picking up some beers. There's definitely causation. There's causation and this correlation does happen. If they can do a large correlation study, figure out some of the causations and see how they're connected, the correlations are an opportunity to come up with interesting information that you might find reasons for.

 

What do SEOs need to do to get away from the robot mindset into a marketing mindset?

 

They need to do marketing. Unfortunately, SEO is marketing on the web. If you're marketing a website, you have to think of the audience the website is meant for.

 

Who are the pages on your site for? Are there specific pages for specific members as audiences? If so, what tasks do you provide for them? What information do you give to them? What problems do you solve? What answers do you have? You have to include those things and pay attention to the audience.

 

If somebody writes a large study about SEO and they completely ignore the audience, they have missed out. SEO is about understanding who the audience is. They want to target the interface. It's like keyword research. You can't complete keyword research if you don't know who your audience is. You have to know which words they use to describe the goods and products that you offer. Notably, they will likely use the search for this. If anyone is unwilling to take these steps, the person may not be able to complete good research.

 

I mentioned that writing articles, having evidence, making arguments, and supporting them is crucial. I think there's a two-way street here. People writing articles need to include those things. People reading articles also need to ask the right questions when they see articles.

​

Is this something that works? Is it something that will provide effective SEO that will help me know that you have some study that says, top-ranking pages are usually 2,000 words, doesn't necessarily help you? Because it doesn't tell you why. And that should be a red flag saying - Do you want to just plan and write two thousand-word long articles?

 

How important is technical SEO?

 

There are some things that do provide help with rankings of websites and help some websites rank better than others. There are mistakes people make that often are outside of guidelines from search engines, and if you do those, you want somebody saying no, this is not a good idea. Don't do this. So, you want to have technical SEO to be able to look through your website and find all those things. You shouldn't tell people, 'here is our link resource page, if you want to be on it just linked us first and send us a copy that we can put on here.'

 

Now that's a good way to get out of the manual family. And no, so there are things that don't work well. And you shouldn't do it. Technical SEOs should be able to go through your site, identify those things. One year that said, you think the Google guidelines in our SEOs Bible, they have a lot of information in them that's useful. But there are some things that they don't necessarily cover.

 

I had one client who had service pages where the client had described the services very clearly. I said, Okay, this is a problem. And this is why maybe you're not getting too many of these customers. You need to flesh out this content, you need to make it more clear about what you're providing, in what people can expect. they rewrote the content and their return on investment increased incredibly, almost overnight, because people could understand what they were providing. And that's not something that's necessarily pointed out in the webmaster guidelines, but it's an essential part of SEO.

 

What is the most important thing in an SEO – client relationship?

 

It is essential to communicate well with your clients and always have a line of communication open and available. Make sure they understand that you are interested in helping and answering questions they have. Because once you achieve that level of communication, you can get a lot more done. As long as they feel comfortable asking new questions, and they should.

 

A lot of time it's just down to communication. Right? If ideally, you should have one person who's a point of contact and your client's team, and that's the person you work towards because this way that person can become the evangelist for your company. If it's the right person, and you'll be able to tell that soon. But they're the person who, as long as you, you limit the communication through one person.

 

On the other side, you're making sure you don't have two or three different things going on at the same time. You just have one thing going on with one person, they're the person who gets communications from their team sent to you, through them. And that makes things easier. It allows for more effective communications.