Almost 2 decades of experience: Jason started promoting his first website in the year Google was incorporated and built it up to become one of the top 10,000 most visited sites in the world.
Today, he is a freelance search marketing consultant and working hard on Kalicube.pro (Measuring Brand Authority).
Jason Barnard (The Brand SERP Guy) On Discussing Answer Engine Optimisation
[00:00:00] Kelvin Newman: Our next speaker is Jason Barnard. And at the end of his session, we’re going to get all three speakers back up to do a Q&A, so start thinking about your questions for that. But Jason is going to be talking about answer engine optimisation, which is what his company, Kalicube, specialise in. Jason has been in the SEO world for 20 years plus and has looked off to some pretty serious sites, driving in the page views in the hundreds of millions a month. So without further ado, Jason.
[00:00:26] The Brand SERP Guy (Jason Barnard): Thank you. Well, I’m absolutely thrilled to be here. Thank you for having me. Today we’re going to cover the answer engine landscape, machine learning, position 0, and the strategy you should implement to become the choice of Google as the answer for its users.
The Very Quick Development of Voice Assistants
[00:00:57] The Brand SERP Guy (Jason Barnard): So, the landscape. The dinosaur isn’t dead yet. The dinosaur is still typing on his keyboard. This drives an awful lot of search. Position 0 is the answer when you’re typing. We all know that these guys are coming on fast. Voice assistants are developing very quickly. These assistants are also moving onto watches, into cars. Munich Airport’s got a virtual system in the form of a robot. And the future according to Google is going to be this. Apparently, they’ve got a bit of a Star Wars obsession. So, I’m going to use Star Wars a little bit today.
Two Sides to the Answer Engine Equation: Understanding the User Intent and Providing the Answer
[00:01:40] The Brand SERP Guy (Jason Barnard): There are two sides to the answer engine equation. First is understanding the user intent. Obviously, we’ve got the typing, the dinosaur, we have Google, we have the assistants, we have Cortana, we have Siri, and we have Alexa. The back end, giving the answer, providing the answer, Google and Bing. It looks a little bit like this. On the left-hand side, understanding the query, understanding what it is the user wants, what is their need, what question are they asking, what is their problem, what problem are they trying to solve.
[00:02:21] The Brand SERP Guy (Jason Barnard): And on the right-hand side, providing the answer, providing the solution to the problem that they have. This right-hand side is actually more complicated than the slide shows. When you’re looking at voice assistants, the voice assistants need other sources of information for the answers than just Google and Bing. I won’t go into that today. That’s obviously the future, and it’s a big, big, big, big subject.
Retrievers Versus Machine Learners
[00:02:56] The Brand SERP Guy (Jason Barnard): So, machine learning, in all of this, vitally important. It’s what’s allowing Google today to become the answer engine it’s always dreamed of being. They say they’re going to become or they’re becoming an AI company, machine learning company. I read an article recently, interesting, about the retrievers and the machine learners. I didn’t really know that distinction before. Retrievers get the information in and then write an algorithm that will provide a set of results from which you may choose.
[00:03:25] The Brand SERP Guy (Jason Barnard): Machine learners have been trying to do that for years and years and years and years. And up until about two years ago, they couldn’t do a better job than the retrievers. Couple of years ago, they managed it. So now, the machine learners are kings of the hill. They’re the guys that are or the people who are really starting to run the search algorithm at Google. The changes that something Google are very demonstrative of that.
Three Types of Machine Learning: Supervised, Unsupervised, and Semi-Supervised
[00:03:54] The Brand SERP Guy (Jason Barnard): Machine learning. Three different types. Supervised, basically labelled data. You pull it in. You tell the machine where you want it to go. You give it the result that you’re looking for. And you ask it to find a route. In answer engines, basically, that’s going to be Wikipedia, Wikidata pulling in data that’s been human curated, verified by humans.
[00:04:16] The Brand SERP Guy (Jason Barnard): Unsupervised is where you give it a set of tools, a set of rules. You give it the mathematical formulas to get to the result that you want, and you just let it loose. First one, a very good example is AlphaGo, if you know about that. 10 million games, human games fed into a machine that then manages to beat the world champion at Go. Unsupervised AlphaGo Zero. You teach it the rules to Go. You leave it to play against itself for two days and it comes out and it beats the first machine a hundred times out of a hundred. So with unsupervised learning, what Google managed to do was beat 4,000 years of human endeavor in two days, which is pretty astonishing stuff. And that gives you an idea of where this can go, how far it can go.
Machine Learning Is More Likely to Be Semi-Supervised
[00:05:08] The Brand SERP Guy (Jason Barnard): I’ve put semi-supervised in red for a very good reason. I do not believe that Google would ever let unsupervised machine learning loose on the machines that make it money. So, we’re much more likely to be looking at semi-supervised machine learning because supervised machine learning cannot be done at the scale Google needs it. There isn’t enough human label data in the world to be able to feed the machine forever. It’s going to have to learn to learn by itself. And I’ll repeat, learn to learn. And that’s the really interesting thing about the machine learning part of all this.
[00:05:43] The Brand SERP Guy (Jason Barnard): And as SEOs, my personal opinion is you would do very well to have a general overview of what semi-supervised machine learning is, teaching machine to learn. And we have machine learning on both sides of the equation. I showed you earlier on, the front end, understanding user intent and the back end, providing the answer.
RankBrain Is Associated With the Front End but Also Uses the Back End Which Understands the Available Options
[00:06:07] The Brand SERP Guy (Jason Barnard): RankBrain. That’s the famous one. That’s the one that everyone was talking about. Generally associated with the front end using word vectors, query patterns, universal sentence, and code. I had to write that because it’s such a long phrase, and I can’t remember it. With RankBrain though, it’s not only there to understand that user intent what the question is, what the problem is. It’s also using the back end, which is the Knowledge Graph, which is aiming to understand the available options. This is the part that I’m really interested in.
Position 0 as the Answer at the Top Spot That Google Is Providing and as Real Estate
[00:06:43] The Brand SERP Guy (Jason Barnard): So, position 0, the answer. Google, I’ll concentrate on Google. It’s the machine we have today providing position 0s that we’re all interested in today. Top spot rules, position 0 is the one we want. It’s the answer that Google is providing today. This is an extreme example. Google had this as a test for a couple of days. By the week, they’ve withdrawn it. This is the extreme example. It’s where we’re going to end up, where Google always end up with its Star Trek machine.
[00:07:14] The Brand SERP Guy (Jason Barnard): Is position 0 real estate? I think it’s a bit bigger than most of us imagined. That’s the one we all know. Rob talked about that earlier on. We’re aiming to get that. That’s one of my clients. You can also steal a little bit of that position 0 real estate by having the image. There was an image on Techopedia, but Google came and found out because ours is better.
[00:07:36] The Brand SERP Guy (Jason Barnard): Related answers. Rob also talked about that, really nice opportunity. Semrush did a study in France. I’m not sure if it’s a specific thing in France, but 60% of these answers do not rank for the actual question they answer in the result, the original result. And that’s probably because the answer is being given within the context of the first question. So, it’s a subcontext. Sure you’ve got quite a lot to say about that.
The Knowledge Graph: You Can Have a Knowledge Panel Without a Wikipedia Page
[00:08:07] The Brand SERP Guy (Jason Barnard): Next, the Knowledge Graph. The Knowledge Graph is nodes with attributes and relationships to other nodes. Here we have a node which is trying to tell, and it’s in Wikipedia. A lot of people think that you have to have a Wikipedia page to have a Knowledge Panel. That is not true. Here’s a company who doesn’t have a Wikipedia page, and yet it has an Knowledge Panel. For me, this supervised machine learning, this semi-supervised machine learning, the machine is starting to analyse and make guesses at information and knowing that this is a company and that it can rely on the information it has.
An Example of a Knowledge Graph and the Future of Nodes, Related Nodes, and Attributes
[00:08:49] The Brand SERP Guy (Jason Barnard): Semi-supervised, back to Star Trek. Here we have William Shatner, lots of related answers. This is music and film and TV. There’s a lot of these, a lot of information. Why? Because if IMDb, MusicBrainz, Wikipedia, all these human curators sources that people since the very beginning of the internet have been pushing this information forward. So, Google can rely on it. It’s human labeled data, and it can use this within it’s supervised machine learning. What you then get with our friend, William Shatner, is he did what he thought was singing, but some of us disagree, albums, and we can dig down and find related answers. Google knows that he did these albums. It knows that he did these films. You can see at the top, there’s a digging down kind of idea from William Shatner to movies and TV to Star Trek, the original series.
[00:09:42] The Brand SERP Guy (Jason Barnard): Why am I showing you this? Because this is what’s going to happen to companies. Today, Google doesn’t have the confidence to show all this information. It doesn’t have this Knowledge Graph with the nodes, the attributes, and related nodes. These related nodes are not yet present for Google in its Knowledge Graph. So, who knows when? We will have a company. We’ll search for a company, and they will give you the sub-organisations, divisions, product, offers, chairman of the board, so on, so forth. We’re going to have this idea of nodes. What I see here is a node with attributes linked to other nodes.
Google Is Weakest When It Comes to Shopping Actions
[00:10:20] The Brand SERP Guy (Jason Barnard): Lastly, shopping actions. In answer engines, this is where Google is weakest. It’s up against Amazon, who’s obviously very strong here. They’re partnering with Walmart, that first logo. They’ve got ad words, and they’ve got targets. So, I don’t think they’re going to be losing the game for very long.
The Future of Position 0: More Searches, More Flavours, More Devices
[00:10:34] The Brand SERP Guy (Jason Barnard): The future of position 0. It’s going to be present in more searches, in more flavours, on more devices, and it’s a voice that’s going to be driving it. So today, we want the position 0 in the type search. Tomorrow, we’re going to be the answer in voice search.
The Strategy: Google Will Give More Freedom to Deduce Relationships Between Nodes as Well as Attributes
[00:10:56] The Brand SERP Guy (Jason Barnard): The strategy. And I’ve got 10 full minutes to tell you about the strategy. In this part, think supervised to semi-supervised machine learning. We’re moving from a world where Google’s Knowledge Graph is filled with information that has been curated by humans, verified by humans, and Google is allowing it to learn from that. Over time, it’s going to, excuse me. Great. Over time, it’s going to allow the machine more freedom. I reiterate. It’s not going to give it complete freedom, but more freedom to learn for itself to deduce relationships between nodes and deduce attributes of nodes that are related to each other.
The First Pillar of Strategy is to Communicate, Then Wrap It up in Some HTML5
[00:11:51] The Brand SERP Guy (Jason Barnard): The first pillar. We have two pillars for this strategy. The first is to communicate. You want to communicate to Google who you are and what you do. You want it to understand you. You want it to understand your brand, your products, your offers. To do that you need to come straight, first of all, on your content. You need good content that’s clear, unambiguous, and you stick on topic. Then, you wrap it up in some HTML5. HTML5 is really useful for identifying the important parts of the content and what role each piece of content within a page plays. Obviously, you want to concentrate on your article, your main sections, but don’t forget about the aside. There’s supplementary information. You can actually insert attributes within the aside, teach Google a little bit about yourself, use it to throw information into Google’s Knowledge Graph.
Schema Markup Is Not Only for the Confirmation of What You Have on Your Page but Is Also for Attributes
[00:12:51] The Brand SERP Guy (Jason Barnard): Next, Schema Markup. There’s a great talk about that earlier on in the auditorium 1 that I saw. Here we’ve got typical organisation markup. What I see here is not only the validation, the confirmation of what it is you have in your page, which is what Google tells you Schema Markup is for, but I also see attributes, characteristics. So, we have our node, which is the organisation, and here we have lots of attributes, and here we have lots of relationships, member of, owns, founder. It’s a relationship, the relationship to the founder. The founder is Jacques Bonifay. That’s a nice relationship. That’s a nice way to teach Google, to communicate to Google who you are, what it is you do, and in this case who you’re related to.
Corroboration: Giving Consistent Information for Google to Gain More Confidence
[00:13:48] The Brand SERP Guy (Jason Barnard): Next, corroboration. I really don’t like the word corroboration. I’d rather say confirmation, but I don’t think it means quite the same thing. First place to corroborate what it is your saying is on your own site and on your own social networks, your LinkedIn account, your Facebook account, your Twitter account, all need to give the same information. It’s a little bit like NAPs in local search, if anybody’s interested in local search. That concept becomes very important for your brand. Consistency in the information you’re giving to Google will enable Google to gain more confidence in what it is understood about you, and that’s vitally important.
[00:14:31] The Brand SERP Guy (Jason Barnard): EV certificates, that’s a third party confirming who you are. So if you want to go to SSL, why not go to EV? You’ve got a third party confirming who you are, then you go further and you go into third parties. Newspapers. This is supposed to be a thought leader here. Not a very good icon. I do apologise. Local government very, very reliable source of information that can confirm who you are. Business directories.
The Context of Links and Mentions
[00:14:58] The Brand SERP Guy (Jason Barnard): Next, the context of links and mentions. I like Majestic’s tool because it categorises links. Google’s a big fan of categorisation. Taxonomies, more we think in that way, the more comfortable Google will be with us. In this example, we’ve got a very, very, very well contextualised back link profile. And I also put mentions. What I like about the new world we’re coming into now of answer engines is we’re moving back towards a more traditional press approach rather than link-building. We’re looking at press relations. I find that very, very encouraging. The thing about mentions is if Google has understood who you are, it can link the mention to you and give you credit for that mention. And it helps it to categorise you. Once again, you don’t need the link. If it’s understood who you are, the mention is quite enough.
Convincing Google That You Are the Most Credible Solution Available
[00:15:56] The Brand SERP Guy (Jason Barnard): So, Google is training it’s machines to learn to learn. And it’s up to you to communicate, to teach, to inform. The second part is to convince. You need to convince it that you are the most credible solution available. If it has two answers, you and your competitor, and it needs to choose between the two, it’s going to choose the one that’s the most credible. And it’s up to you to convince it of that.
The Importance of User Behavior
[00:16:27] The Brand SERP Guy (Jason Barnard): User experience, user behavior, vitally important. You need to appear for the right keywords, so that when the person comes to your site, you can solve the problem that they have or answer the question that you’ve been asking. You need to look at the signals, such as the click-through rate, whether they’re strolling, whether they’re bouncing, whether they’re Pogo sticking, whether they’re dwelling on your site or not. Google looks at all this. And the more it moves forward, the idea of, for example, click-through rates being a ranking factor has been pushed to one side by some people. Google is starting to say that it does have an effect, that it’s part of their overall algorithm. That must be machine learning because the amount of data is very difficult to deal with. So, user behavior is becoming increasingly important.
Getting as Many Reviews and Positive Feedback and Buzz as You Can
[00:17:11] The Brand SERP Guy (Jason Barnard): Next, reviews, lots of reviews. I tell my clients to search for their brand plus the word review. You look at the first page of results, and that’s what Google thinks the world thinks about you. And if it’s rubbish, then Google thinks that you’re rubbish. At least it thinks that the people think you’re rubbish and it, by association, it’s going to think your rubbish. You need to make that page really good, pick full platforms, and work them to death. Get as many reviews as you can. Get that positive feedback going.
[00:17:45] The Brand SERP Guy (Jason Barnard): I often recommend Trustpilot because it’s, for me, the strongest platform for SEO. It sends out lots of very good credibility signals to Google. I tend to find as well with them when you search for a brand or a brand plus you, you then own two places in the results. You own the one on your own site, but you also own theirs. And they almost always ranked for most brands, in this case, with Mailjet. I also looked at G2 Crowd and Get Up, whatever your industry is. And that’s the easy way to find it. Run name plus reviews, pick the top four, and work them to death.
[00:18:21] The Brand SERP Guy (Jason Barnard): Links, mentions, and buzz. As you can see, the two pillars actually play off each other. What you’re doing on this side will help the understanding. What you’re doing on the understanding will help your credibility. You need to get a really nice buzz going as much as possible around your brand. Make sure that it’s all positive buzz, of course. Get the thought leaders on board, press relations, or obviously, the artist. Sorry, I almost said a word in French.
Inherited Credibility: Bigging Yourself up
[00:18:51] The Brand SERP Guy (Jason Barnard): And this is my favourite because I made it up yesterday, inherited credibility. And here we are really looking. What I see here is a node, lots of nodes, and relationships to those nodes. Now, why is that interesting? Because if I have a partnership with a really big company or I’ve won this really groovy award or I’ve been mentioned in the press, if I can communicate to Google that relationship, some of their credibility rubs off on me.
[00:19:21] The Brand SERP Guy (Jason Barnard): So basically here, I’m bigging myself up. And I’m bigging myself up by saying, oh, look at all these entities I have a relationship with. Some of their credibility could come to me. And you don’t just have to sit there and expect Google to do this on its own. You can use Schema Markup, as we saw earlier, to demonstrate, explicitly place in Google’s algorithm the relationship that you have with these entities. Wikipedia, Crunchbase, and Wikidata place the information explicitly so that Google sees the relationship, and you can big yourself.
[00:19:58] The Brand SERP Guy (Jason Barnard): So, Google is teaching, is bringing its machine learning to learn how to judge. If that was all a little bit quick, there is a handout on the way out from Trustpilot that I wrote, 12 pages on this exact strategy, and it takes up all these points in the same manner.
[00:20:19] The Brand SERP Guy (Jason Barnard): So, the takeaway, very quickly. If you haven’t taken those, just take a handout on the way out. The takeaway is you need to communicate, to convince Google so that it understands that you are the most credible solution or answer to its users for any given problem or question. Thank you.
[00:20:42] Kelvin Newman: This was originally recorded at a BrightonSEO Conference. If you want to listen to more episodes or find out about the conference itself, you can do at brightonseo.com.