Google’s Gary Illyes on Emerging Search Trends – Virtual Keynote 1

Introduction

On February 11th, 2016 I had the opportunity to do a live video event with Google’s Gary Illyes and we took on a wide variety of topics in Search and SEO. We then had Mark Traphagen lead a 30 minute Q&A session in which we took questions from the audience.
Our second Virtual Keynote with Gary Illyes, focusing on SEO tags, can be viewed here.
As you might expect, there were a number of interesting gems that emerged including a new example of a RankBrain query, an update on what impact the new Penguin will have, why Google does not certify SEOs, how AMP works, and much, much more.
Below is the complete transcript, along with some specific notes from me on key points. Or, you can watch the full video here:

Virtual Keynote Transcript

The Digital Essentials, Part 3

Developing a robust digital strategy is both a challenge and an opportunity. Part 3 of the Digital Essentials guide series explores five of the essential technology-driven experiences customers expect, which you may be missing or not fully utilizing.

Get the Guide

Eric: Hello everybody, I’m Eric Enge of Perficient Digital, and this is Gary, whose last name shall not be pronounced, from Google. He actually complains any time anybody actually tries to pronounce it, to be fair.
We also have Mark Traphagen, who’s going to be running the Q&A side of things. Mark will be active on YouTube during the show and interacting with you all over there, so thanks for joining us. Thanks for coming, Gary, appreciate it!
Gary: Thank you for having me.
Eric: Absolutely. So, here’s the official book language on Gary. He’s a webmaster trends analyst at Google, working out of the Zurich office. He is dedicated to creating a better search experience for users by helping webmasters create websites that are both user and search-engine-friendly.
When he’s not looking for ways to improve web search, Gary can be found helping webmasters on social media, such as Google Plus and Twitter, and at conferences, doing all kinds of stuff. So, you also told me you’re pretty instrumental now on the webmaster PR side of things, so that’s pretty cool to know as well.
Gary: If you think about it, usually when Barry has a question, or Jennifer from the SEM Post has a question, then it certainly feels like the first thing is to ping John or me on Twitter, and if we don’t answer that, they go to PR, and then sometimes we have to help PR give an answer. I’m involved with many things, not just PR. I was working for quite some time on search tracking as well and different infrastructure pieces within search, so yeah, my hand’s in many, many jars.
Eric: Yeah, well, you’re instrumental in the HTTPS thing too, right?
Gary: Yes, these guys made that.
Eric: Oh, those guys right there, excellent! So that would suggest you’re fairly technical.
Gary: Yeah.
Eric: Okay, so how do you explain this then?

Gary: I cannot confirm or deny anything about any operation without the express… or, yeah, I just s***.
Mark: What was that, Eric? What were you looking at there?
Eric: That was the SEO credit score that Gary got on the Perficient Digital Quiz… sorry, the SEO quiz that I published on Search Engine Land. Gary reported that he got 15 out of 20 on the quiz. To be fair, there are at least two questions that I acknowledged needed to be changed after chatting with Gary, and I suspect a couple of the other issues really related to just probably taking it a little too quickly and not paying enough attention.
Gary: No, I would never do that.
Eric: So, fair enough. But anyway, one of the big things that happened lately is Amit Singhal left Google after 15 years. He’s had an amazing presence within the company. Can you talk about that a little bit, what he’s meant and your thoughts on his departure?
Gary: Well, not too much. Basically, what he said, that’s what I can tell you. But obviously his departure is really bittersweet for my team. On the one hand, it’s always a bit sad to see someone like Amit leave. He had an enormous positive impact on many, many teams inside the wall. He had a huge positive impact on search, for example, and also on our users’ lives.
He probably approved thousands of launches, and all of those launches affected users’ searches. On the other hand, Amit is really passionate about many, many things, and I certainly understand that, after 15 years, he would like to try something else. And to be frank, I’m quite happy for him, that he could try something new that he’s passionate about.
Eric: Yeah, that’s great, in his post, he talked about wanting to spend some time doing charitable work, giving back, spend some time with his family—that’s a great thing to do as well. Obviously he had a huge impact on the industry and search overall, so… I don’t know him personally, but it sure sounds like it was a good time for him to make that move, and there have been a lot of changes in the search team over the past couple of years.
Matt Cutts stepped back, is still on leave. That kind of led to you becoming a bit more of a public face as well. So there are those who refer to you as the new Matt Cutts, but I think you’re rather a different animal than Matt, personally.
Gary: Yeah, I’m definitely different than Matt, especially because we don’t believe that there’s a single person behind Matt who’s replacing Matt. I think that’s a misconception. There’s actually a whole team replacing Matt or that’s replaced Matt.
The only thing is that I gained more visibility because I’m going to bigger conferences, and because I have a big mouth on Twitter, and I like trolling people and give smart-ass responses. But I think, other than that, actually, the rest of the team is doing much, much better, and they are focusing on SMBs for example, which is great. And I strongly believe that their impact is much bigger than mine.
Eric: Yeah, but you did end up with a very visible face, as you said, from being at a lot of the larger conferences, and that’s what makes it interesting to have a chance to talk to you, because you get to talk to a lot of people at those conferences as well. But the new person that’s taking over for Amit, I’m going to try to pronounce it, John Giannandrea, or something like that?
Gary: Yeah, we call him JG.
Eric: JG? Yeah, it’s easier. So we’ll settle for that too. So, JG is taking over. And I know you can’t really say too much about him at this point, but one of the things that’s interesting is that he, evidently, led the role of RankBrain and has been integral to machine learning at Google over the past few years. Does that mean that we’re going to see an increasing shift over to machine learning? Can you share some thoughts about that, or machine learning and Google in general? I know that’s probably a huge topic.
Gary: It is. It’s a freaking huge topic. I just want you to keep in mind that machine learning and deep networks are not a magic syrup. It’s not like you will add it and suddenly everything will become better. When you resort to machine learning, there are tons of things that you have to consider.
You have to consider, for example, how will you debug the output, and how will you oversee what the machine learning algorithms spit out. And you also have to have a team that will actually understand the output clearly. It’s not enough to just train something and then add it to search, for example. But for search, I actually don’t have anything more to add.
We do think about how we can use machine learning for different things. Obviously we are always exploring ways to improve search and make our products better, but it is a big thing, and we try to tread carefully and not jump into it and then perhaps accidentally ruin something.
Eric: Well, all right, and I’ve done my own little exploration of machine learning, and gotten to the point where I’m able to develop some simple programs and train some simple neural networks.
Gary: Yeah, I think I saw your Twitter Predictor, or Twitter Shares Predictor?
Eric: Yeah, it was a pretty simple tool… (you can see the retweet predictor here)
Gary: I thought that was pretty awesome.
Eric: Well, I had fun with it. Basically, we had 1.9 million tweets’ worth of data (you can see our study on Twitter engagement here), and we knew things like social authority, or the images, or number or characters, and things like that. And in the grand scheme of machine learning things, it’s a very simple program, but it was useful to me to get some idea.
I think one of the things I’ve learned about it is it constructs an algorithm, which evidently gives me the equation that allows me to predict the chances of getting a retweet. But you’re right, there’s a whole process that you have to really put a lot of effort to figure out whether the resulting algorithm makes sense.
Fine, it constructed an algorithm and it fit the data somehow, but it seems to me there’s an enormous amount of testing that has to follow, because you don’t have this human intuition that went into the creation of it, which by itself would be flawed. But instead, this machine, it’s a mechanical thing, and at the end, it’s coming out somewhere when you have this program, and it’s either useful or not.
Gary: With anything that we launch, we do rigorous testing. We run sometimes an experiment for months to make sure that what we launch will actually work or actually will improve search. And with machine learning, there was nothing different. We were testing it for a very, very long time and we just wanted to make sure that RankBrain will actually do something great and will not decrease accuracy and relevance of the results.
Eric: Right. In the case of RankBrain, as I understood, and maybe you can’t comment on this, but a lot of what was going on is understanding relationships in language so you could better appreciate what things were closely related to one another, and that could help you understand more complex queries, and perhaps also more complex… how to parse webpages.
Gary: I mean, if you think about, for example, a query like, “Can you get a 100 percent score on Super Mario without a walk-through?” This could be an actual query that we receive. And there is a negative term there that is very hard to catch with the regular systems that we had, and in fact our old query parsers actually ignored the “without” part.

(Eric’s Note: In the absence of the word “without” this query will focus on returning results that depend on a walkthrough, which isn’t what the user wants, and RankBrain fixed this query and others like it.)
[Tweet “Google’s Gary Illyes: RankBrain helps Google understand what humans mean by words like ‘without.’ More at”]
And RankBrain did an amazing job catching that and actually instructing our retrieval systems to get the right results. It’s definitely something that works really well with natural language queries, and its accuracy is way better than, for example, ranking engineers that worked for 10 years on search.
Eric: Yeah, that’s amazing. So that had to be something that was in the works for an awfully long time, an awful lot of testing to get to the point where you were comfortable rolling it out.
Gary: Yeah.

Related: See our in-depth study of RankBrain’s effect on Google Search.

Eric: Absolutely. And maybe you can’t answer this either, but is machine learning used at all in things like Penguin, or Panda, or other algorithms?
Gary: I will not comment on that.
(Eric’s Note: There are many known ways that Google uses machine learning today. For example in Google News, it’s used to identify similar articles as shown here:

The highlighted portions are the related content identified by a machine learning algorithm.
In a 2001 interview I did with Peter Norvig, I also learned that Google used it to build Google Translate. They did it by finding sites on the web that offer multiple language version of content and seeing the outcome of their translations.)
Eric: OK. See, I knew you might not be able to answer that one.
Gary: But I answered the previous one.
Eric: Well, there you go, so that was good. One for two. Very, very cool. So let’s talk about mobile a little bit.
Gary: That’s my favorite topic.
Eric: I’ve heard that in the past, actually. I think it was actually when you were saying you were going to refuse to talk about it at a conference. So, what are the most common problems people have on their mobile sites?
Gary: I would say that roboted resources are still the biggest problem. When we try to figure out if a page is mobile-friendly or not, then we will use the rendered version of the page. And for rendering, we will try to bring in resources like JavaScript and CSS and whatever, and if we cannot bring in something because the resources robot, for example, isn’t up for crawling, then we might not get the content of the page right, or what devices it is meant for.
And I think that’s a huge problem, and we are actually working with different CMSs, or at least worked with different CMSs on the robot, most of the resources that they could. Because, to me, it feels kind of stupid because it’s like, we won’t fetch those resources too often. We wouldn’t put servers to their knees by fetching those resources too often. So I definitely think that people should allow those. We are not going to index those resources anyway, so there’s no good reason to not allow us crawling.
Eric: And as I understand it, at this point you can do a pretty good job of leveraging CSS and JavaScript, or parsing CSS and JavaScript, I guess I should say, to understand the page layout. And that’s extremely helpful to
you, to be able to do that.
One of the things that you made clear recently is when somebody has something that requires a click, like opening up a JavaScript window or something like that, and you don’t really pay attention to that content inside of that at this point.
Gary: Yeah, that’s correct. Well, actually it depends on how the content is broken. Because if the content is hidden in the page, like the CSS rule, then, or JavaScript, well, that would be also CSS, just indirectly, then we would be able to see that content. The content’s weight in search wouldn’t be that big, because the content is hidden. The user has to do something to actually see the content.
But if you bring in the content with an Ajax code, for example, like an XML HTTP request, then that’s a different story. And generally we will not make that call. So for example, if you have to click a button and then a JavaScript will query a server end point and will grab some HTML and then inject the HTML in the page, people don’t do that, in most of the cases at least. I know that there are teams at Google that try to bring in that content as well. I’m not sure which teams. But in web search, we don’t really do that.

Eric: One of the reasons for treating that content differently is if it’s not immediately visible on the page, then it’s less accessible to the user. It’s available to get to it, but it’s less accessible. So it makes sense to give it less weight, right?
Gary: Yeah.
Eric: Yeah, absolutely. So, back on the mobile side of things, what’s the current situation with interstitials and how you’re viewing those and handling those?
Gary: I’m assuming that you’re talking about app install interstitials.
Eric: Right.
Gary: There’s nothing new around there. We are doing exactly what we announced. We are going after app interstitials, but there’s nothing new, really.
(Eric’s Note: What Google does here is take away the mobile friendly designation from you if you use these. This went live on November 2, 2015)
Eric: Okay, fair enough. What about AMP?
Gary: AMP is awesome.
Eric: AMP is awesome, yes! So, I’ve been reading up a little bit about it, but can you just give everybody a little bit of an overview as to what it does and how it works?
Gary: Yeah. So basically for each content you have on a single page, you would create another page with a new kind of markup, which is still HTML, it’s just a bit different. And anyone would be able to load that content much faster because of how it’s distributed across different caches that are set up on the internet.
The pages would be fast for several reasons, because they share a common HTML with well-defined text, for example, which doesn’t mean that developers can’t define their own text, but that’s beside the point. Then you can use limited JavaScript, one of the bottlenecks nowadays. You can also use limited CSS, for example, and all of these will speed up the pages considerably.
Finally, as I said, anyone who wants to set up a caching network can, and then the AMP pages would be loaded much faster from a location closer to you. Imagine that you are, for example, in India and you are trying to access a server in the U.S. Your connection would have to travel across the globe.
With cache or with Mcache, potentially you will not have to travel across the world with your connection, but rather access a local Mcache, which already has your content loaded, which can save sometimes several seconds, and that’s great, I think. Did I answer your question?
Eric: Yeah, so I’m going to play it back a little bit. Some of the elements of the page structure are coming from a common library, if you will, and because that’s cached all over the globe, those portions of the page structure load very fast. And the rest of your content, which is custom to you, still has to come from your web server infrastructure, but the pieces that are in common are already loaded, and so that makes everything go faster. Did I get that roughly right?
Gary: Well, you can also cache a local copy if you want. So it’s not necessarily just a library that you can cache, but you can also cache the content, and then you will report… I’m actually fuzzy about this part, but you will also have to refresh that or to abide to some refresh rates at which you have to fetch the content again or drop the content from your cache. But basically you have tons of freedom, unlike with other quick pages.
(Eric’s Note: What makes AMP work is basically two things: (1) You can only use limited code elements on your pages, such as no forms, and very limited Javascript, and all content must be pre-rendered in the page, and: (2) the caching that Gary refers to above)
Eric: Right, so as I understand it, at the announcement, Pinterest and Twitter were already participating in leveraging AMP as well, did I read that right?
Gary: As far as I know, yes.
Eric: Yeah, so as you said, it’s not something that’s just restricted to something you have to do for Google. It’s an opportunity for anybody who…
Gary: Yeah, I mean, it’s totally open. The library is on GitHub, as far as I know, so you can even contribute to it if you’re a developer. Yeah, we believe in open summaries in general, so we wanted to be open with this as well.
Eric: Right. Great, so let’s talk a little bit about app indexing. What are the current things that Google does that support app indexing? And I think a lot of the viewers probably don’t know just how much you’re doing in that area, even though I don’t think it’s changed for a bit.
Gary: To be frank, I don’t think that we are doing that much [new] nowadays around that. We have about 100 billion links within apps indexed nowadays. It’s still an important feature and I definitely recommend that when you are creating an app, that you make sure that app indexing would work on that, but otherwise there is pretty much nothing that I can recall we do for indexing nowadays.
Eric: Yeah, so don’t you still have situations where you can get the app page you’ve shown? But also in the search results that can happen if you have it installed.
Gary: Yeah, sure.
Eric: So that’s something that I think is very useful. And then at least historically there were situations where you would actually recommend people to install the app under certain situations?
Gary: Yeah. I think we still have that option but not from within app indexing. It moved to a different area, and I think nowadays it’s sitting in apps universal instead of app indexing. Essentially, not much changed, as far as I know, but if I remember correctly, the engagement on those installs, or install buttons, weren’t high enough or there weren’t enough engagements to spend more resources on it, and we just moved to another feature instead of spending resources on maintaining it.
Eric: Right, that’s common, where you’ll try something and then if it isn’t working that well, then you’ll back off from it.
Gary: Yeah. And I think that’s creating rule that we run tons of experiments with many, many things. With ranking, we run experiments all the freaking time and also on UI, we like to run experiments and see how people rea
ct to it.
Sometimes we just want to leave something there, we want to learn something, and then let people decide if they like this or not. So for example, with these installs, we launched it, I don’t know, a year ago or something, maybe more, I don’t know, and then we saw that, after 10 months, people didn’t engage with it too much. So we just removed it because we also don’t want to have too much clutter on the search result pages.
Eric: All right. When you’re dealing with devices that size (referring to Smartphones here), it’s a little messy to have clutter.
Gary: Yeah.
Eric: Sure. What’s up with Penguin?
Gary: It’s about 70 degrees outside. Why would penguins be here?
Eric: That’s a great question!
Gary: I don’t know. Someone asked me on the computer a few days ago, and I said that it’s… I don’t remember which version I said, but we were aiming to launch Penguin this quarter. I’m not sure if that will happen. I haven’t checked with the team for a while. We do definitely check in with them and ask them, “What’s up with Penguin?” but I think, as any human, they have a threshold for nagging.
Eric: Being badgered.
Gary: Yeah. And we try to not reach that threshold, because if we do, then they would just not answer us anymore. I know that they are running the experiments, and no, you can’t see that externally, whatever you might think.
Not you, Eric, but some other people. They are running the experiments, but we will also not launch something that we are not happy with. With Penguin, it can have a very strong effect on a page, and we want to make sure that if Penguin affects a page, then we are absolutely sure that that page should be affected. We don’t want to negatively affect a page without a good reason.
[Tweet “Google’s Gary Illyes: Penguin not launched yet because of extensive testing to make sure it will work right. More at”]
Eric: Right, well, that relates to what you were talking about before, is since Penguin is an algorithm that can potentially have a very large impact, the amount of testing you have to do before every release is probably significant.
If you go through that testing and you find out that, “Oh my gosh, we’re having negative impacts, more negative impacts than we want,” you have to start, “OK, we’ll do another pass at it.” And because we’ve talked about Penguin coming out even some time last year, I’m imagining you’re going through a cycle of not being happy with the results and just trying to get it right.
Gary: Yeah. I mean, first there’s lots of brute tuning going on, and after a while, you reach a phase where you have to actually do really, really tiny fine-tuning on Penguin and algorithms in general. And sometimes that fine-tuning can actually take way more time than the brute tuning. We are working hard to launch it as soon as possible. I can’t say more than that.
Eric: Fair enough. And we’re going to go to Q&A in just a second, but one last question for you from me: Give me some SEO misconception that drives you nuts, and let’s dispel it right here. How’s that for a challenge? Because now you’re thinking, “Oh gosh, there’s so many!’
Gary: No, I actually can’t think of anything because I reached that phase where, when I hear some misconception, then I’m just like, “Oh, OK.” There are very few things that annoy me. I don’t know actually. This morning, we had that question about whether increased crawl rate means that an update is going on, or will go on in a few days. That’s definitely a myth, for example.
Crawling and ranking are two separate things. We don’t necessarily recrawl the web when an update is going on, or we don’t recrawl the web just because we had an update, because sooner or later, we will recrawl it anyway.
[Tweet “Google’s Gary Illyes: Increased crawling activity does not indicate an algo update is imminent. More at”]
Eric: My wild guess here is that crawling the web is something that you do frequently.
Gary: Yeah, you could say that, yes. Sometimes for breakfast, sometimes for lunch, but we definitely recrawl the web. Anyway, it has no connection, really.
Eric: Fair enough. So, Mark, I know you’ve been collecting questions and things like that. Well, what do you have that we should be taking from the audience?
Mark: We’ve got a lot of great questions, Eric, from our pre-registered viewers who are with us today, and we want to get through as many of those as we can. I tried to focus in on the ones that I think will have the most general interest for people. We had some that had very specific questions, people having problems with their sites and things like that, trying to get some free Google consultation from Gary, but we’re going to try to benefit everybody here.
So, I’m going to punch right in, Gary, with one that I thought was particularly interesting. I think it’s a very creative question from Kristin Drysdale, and she says, “If you were making your last tweet that you knew all webmasters and future webmasters would hang on, the last word from Gary Illyes, what would it be?” What would you leave in a tweet with all of us if it was your last tweet?
I hope she’s not meaning in your life, but as a Google spokesperson, what would you want to leave us with?
Gary: So, the problem is that I’m quite a troll, and I would probably just say something utterly stupid. It probably wouldn’t be anything constructive. It would be funny for me, and that’s it.
Mark: Is there any word of wisdom? If you had a last chance to leave a word of wisdom with all of us in the SEO world, what would it be?
Gary: Jeez, I don’t think about these things. Come on, guys!
Mark: Why don’t you think about that if you… go ahead.
Gary: OK, this is another tweet that I did last weekend, I think. I was quite amused because I was reading these articles, not yours, Eric, about machine learning, and sometimes they were probably like 432 pages long. And I was just scanning them, and scanning, and scanning, and there were these huge speculations about how this works and why this works and why are we doing this and how are we doing this.
And then at the end of the articles, people just wrote that, “Well, you know what, actually you just have to focus on the user.” And I’m sitting there, it’s like, “So I read 423 pages and that’s what you tell me, that you have to focus on the user?” And actually that’s what we were telling you for probably like 15 years or even more. Seriously?
Eric: To be fair, I actually wrote an article like that (you can see my article on machine learning here).
Gary: Yes, I saw.
Eric: But having said that, obviously I’ve been involved in SEO for a long time and I’ve seen so many different people just get focused on trying to work the engines and the algorithm and trying to find the latest trick to make something happen. And as you’ve said earlier, machine learning is no magic bullet, but it is a new tool, a new class of tools can come out of machine learning that puts Google in a better position to accomplish its objectives, which is to surface the best content, right?
So, I know when I say something like that, I’m thinking that, “Gosh, I’ve got this community of people, or a lot of people in the SEO world, who still look for the latest ways to manipulate the algorithm, and the presence of machine
learning is just another tool that Google can use to make those pursuits less fruitful.”
Gary: Sure. Machine learning can be used for many, many things, but I’m actually just amused that the general conclusion is that you just have to design for users. That’s like, “Duh.”
Mark: So there you have it folks, some day on Gary’s tombstone when you go to visit…
Gary: “Focus on the user.”
Mark: “Focus on the user,” that’s what it’ll say on your tombstone.
Eric: Well no, then it will say, “Duh,” at the end.
[Tweet “What would Gary Illyes’s last words to the SEO community be? Find out at”]
Mark: “Duh.” All right, let’s move on. This question comes from Kenichi Suzuki, who says, “Do you have any data that indicates the growth of voice search?” So he’d like to know how many people are using that feature now. Is that something that you track, looking at how many people are using voice-activated search?
Gary: Konichiwa, Suzuki-san. I don’t actually have the latest number on that. I know that we obviously do track this thing. Why wouldn’t we? But I don’t actually remember it.
Mark: Do you remember if it’s growing? Do you see it growing?
Gary: Oh yeah, definitely. In general, people speak more like a person to their phone, so for example queries like, “When is my car reservation?” or “Remind me to buy flowers for my wife.” We are getting more and more queries like this. In fact, we get, I think, 30 times as many action queries by voice as by typing, so that’s really a voice thing.
Mark: Is that affecting the way that you are designing the algorithm or the way that you’re responding to the customer?
Gary: Yes, if you get to SMX West, Beshad will talk about this more. I don’t want to undermine his talk.
But for example, I just remembered a fact, last year, the number of mobile voice searches have more than doubled than the year before, which is, again, I don’t remember the number, but obviously if you see that the number of voice searches are doubling, then that means that we must get quite a few.
[Tweet “Google’s Gary Illyes: The number of voice searches on mobile have more than doubled in past year. More at”]
Eric: Right. And just for people who are not aware, when Gary refers to Beshad, that’s the opening keynote, I believe, at SMX West. He’s the VP of conversational search, so there’d be a lot more data available if you’re able to get to SMX West and see that presentation. (his full name is Beshad Behzadi, and you can find details on his keynote here, by clicking on “Learn More” for the opening session of the conference)
Mark: Eric and I will be there as well and speaking, and I’m sure Danny Sullivan appreciates this little free ad for SMX West. A great conference, you should come.
Eric: There you go.
Mark: All right, let’s move on. This is a question from Stephanie Salmon from U.S. News, and she asks about archived news content. There’s been a lot of articles recently about that and some worries people have that that might be seen by Google as not valuable content, kind of old news that’s still on your site. Should you worry about that?
Gary: No. I definitely think that people should not worry about that. They should keep it up. What would happen if we burned up all our libraries, for example?
Eric: That would not be a good thing, yes. You would lose a lot of historical human knowledge.
Gary: Yeah, and I think the same goes for new sites as well. At least in U.S. libraries, I know that you can access newspapers, newspaper archives as well, and news sites, when they aren’t hiding their content, function similarly. It’s like a library containing a wealth of information about humanity, and I think that should be definitely preserved and that people should be able to access it. Ultimately our goal is to make all information universally accessible and usable. I think we don’t want to give up on that.
Mark: It makes sense to me that Google would understand that there are times that we want to go back, we want to research something, we want to know something that happened many years ago, and it’s good for it to be there, right?
Gary: Yeah.
Mark: Okay, thank you. This question comes from our friend, Bill Slawski, and it’s going to be kind of a Bill Slawski type of question. So it’s the…
Gary: It’s about patents.
Mark: It’s going to be about the Knowledge Graph APIs. He asked, “How are the result scores for different entities in Google’s Knowledge Graph search API derived?”
Gary: I have no idea.
Mark: I knew you were going to say that!
Gary: I’m sorry, I have absolutely no idea. In fact, I wouldn’t even know who to ask, but I can dig that up if someone pings me on Twitter about that.
Mark: OK, well, there you go, Bill. I know he’s listening, so you got your invitation. Expect your tweet as soon as you get off here, Gary. All right. Let’s move on then to, this is from Robert Nowaczyk, I believe the name is, who just has a very simple question, “Is RankBrain operative for the whole world now? Is it global?”
Gary: Yes.
Eric: So, in all languages.
Gary: To the best of my knowledge.
[Tweet “Google’s Gary Illyes confirms that RankBrain is now global, for all languages. More at”]
Eric: Even Hungarian? (note: teasing Gary here, as he’s Hungarian)
Gary: I’m not sure about that.
Eric: Does RankBrain know how to pronounce Illyes, or however that’s pronounced?
Gary: Actually, Mark was pronouncing it pretty good (Note: before we went live with the show).
Mark: Well, as I told you, before the show started, I lived for two summers in Hungary, so hopefully I picked up something there.
Gary: First you said three summers, now you said two summers.
Mark: No, two, two summers. Darn, that wasn’t recorded. I was going to say it’s on the record. All right, you don’t want go get caught lying by Google, folks. You do not want to get caught lying by Google. Let’s move on to another question, save my neck. This is from Monica Schmidt, who asks, “Will virtual reality, or augmented reality, change the terms of influence in search, process, and result?” And just simplifying that, is virtual reality, which we all know is coming, is that going to have an effect on search? Are you guys thinking about that?
Gary: No idea. VR is pretty new. I mean, it’s been increasing the past few years, its usage, and we’ve seen different results or different implementations which had different qualities. I have no idea where this is going, but it’s definitely exciting. Obviously, we can try something with VR and search. We will try. Hell yeah, why not? It’s interesting, it’s exciting, it’s fun, and that’s pretty much what Google is about.
Mark: Right. I just wonder if there’ll be new ways, such as you mentioned earlier, that people ask questions, they query quite differently by voice than they do by typing. So will there be some way of querying through virtual reality, the virtual reality world interacting with a search engine that way?
Gary: I don’t know. I mean, we could speculate about this, but we are not there yet. We will definitely try different ways. Obviously we, as in developers, in general, all over the world, will have to figure out how to actually make controlling the display, the UI, easy when wearing a VR set. Nowadays I’ve tried only a couple VR sets, and controlling the UI wasn’t all that easy. So we also have to figure that out. There are tons of things that we have to figure out, and then we will run experiments, we will come up with fun ideas, probably.
Mark: No, I can think that will happen. I might be in a virtual world, and see something. Maybe it’s a simulation of something in the real world, and I look at it, and I want to know what it is, “What is that building?” or, “What is that event?” to be able to find out without leaving that environment. Very interesting.
Gary: I mean, I can also imagine something like, nowadays people spend more and more time on exploration. They go to a website and then they will just step on Wikipedia. Great example. You go to Wikipedia to read about Google Glasses and you realize that you ended up on how to make light bulbs seven hours later.
Eric: Oh seven hours later. That’s somebody with too much time on their hands there.
Gary: Yeah, I’m not sleeping much. Anyway, and I can definitely imagine something like having something in front of me and, with my hand, just scrolling or something like that. But I can imagine the same thing in search as well. Even now, we have immersive experiences for different things and those could be, in theory, controlled with a VR set, or presented a VR set.
Mark: So I want to shift gears back to Penguin. You knew that was going to happen.
Gary: Obviously.
Mark: And Jesse McDonald asked a question. His question was very specific. I’m going to generalize it just a little bit more. His question had to do with real-time Penguin. I’ve been told that the next Penguin update will be real-time. What will be different for us in the SEO world, if anything, do you think? How will we experience that differently once that happens?
Gary: My problem is now that we didn’t think yet about how to communicate this, so I’m making things up as we go. So the first thing will be that if your pages are affected by Penguin, then generally you will be able to get rid of that effect much faster.
So you will definitely be able to see that there is something going on, like your rankings are dropping. You can easily think back what you did, what you changed, if anything, on the site or external to the site, like off-site SEO, or link building, or whatever, and then revert those changes and see if that fixes it.
Mark: There’s more real-time feedback in a sense, where you can see more quickly if you’re doing the right thing to get back.
Gary: Yeah. Of course, we will have to recrawl the pages, and sometimes that can take a lot of time.
Mark: So it’s “real-time,” air quotes.
Eric: Well, it’s real-time with the data you have available, but you don’t have the data until you recrawl some things and you discover the data, right?
Gary: There you go.
Eric: So there’s some natural latency in there. And as I understood it, historically, there was some hesitance (at Google) to make things real-time like that because potentially people were doing the wrong things to do a lot of experimentation to figure out what wrong things they were able to get away with and which ones they couldn’t.
But it seems like you have a lot higher degree of confidence now …
Gary: Yeah
[Tweet “Google’s Gary Illyes says real-time Penguin will make both problems and resolutions visible faster. More at”]
Eric: …. that would cause you to think that, “Now we can make it a bit more real-time,” which, of course, is good for people who have made innocent mistakes. Or maybe they made not-so-innocent mistakes, but they’re willing to reform.
Gary: So, what I’ve seen is that it’s not people making innocent mistakes, but rather their SEOs. I know this will be shocking, but there are bad SEOs out there.
Eric: Damn!
Gary: I know, I know.
Mark: Some of them swear. I’ll quickly say, some of the good ones swear.
Gary: Actually most of the legitimate sites that I looked at that were protected by Penguin were small businesses. Or well, actually, not that many small businesses. Well, in general, businesses that hired an SEO who did something nasty. And this just goes back to, there are things that, as a site owner or a content owner, you can’t do without an SEO. I don’t say drop all SEOs, because SEOs can do awesome magic.
I know some of the sites that you’re working with, for example, and I know that they are doing amazing. Or there’s Bruce Clay, for example, who has a great team. There are a few companies that are really, really good, but you have to be careful how you choose your SEO team, and look at what they did in the past. Talk to the other clients, perhaps, but you have to figure out a way to make sure that the people that you are letting touch your website will do a good job, that they know what they are doing.
Eric: So, that analogy would be, as someone who is a father and has a daughter, there’s a certain amount of scrutiny that goes into guys that are allowed to date your daughter, and you might not actually have control over that, but the same mental process you go through there, maybe should apply to your SEO agency, because you’re letting them get a hold of your company jewels, so to speak. And you’ve got to be careful.
Gary: You know, that sounded weird.
Mark: All right, that’s all we have time for…
Eric: All right, I’ll stop. Next question?
Mark: So Bharati Ahuja asks, “Do you think that Google should have an SEO certification program just like it has an AdWords certification?”
Gary: We were thinking a lot about this. A lot. And then a bit more. And we always end up with the conclusion that, no we shouldn’t. There are many reasons why that would be a good idea. There are way more reasons that it would be a bad idea. If we endorse SEOs, then that will also mean that we actually have to, perhaps, train them, which would be nonsense because, first of all, it wouldn’t be scalable to train I don’t know how many people.
Second, if we do go down that path, then we will have to hire more people, which costs money. We have to get that money somehow, so we would ask for money. Search doesn’t ask for money for anything, so there’s already a conflict there. There were tons of things that we brought up against this. My short answer is no, we will not have that any time soon. That’s it!
[Tweet “Google’s Gary Illyes talks about why they have not created a certification program for SEOs.”]
Mark: This is one of those ones that Gary’s going to hate.
Gary: If you already know, then why?
Mark: Yeah, just because it’s not going to be apparent on the top of your mind is what I mean. Don’t think about this one.
This is from Gary Yorke, who runs a small business, and he says, “What would be your top recommendations for business owners who are not SEO-savvy and on a limited budget?”
Which is, limited budget, I’m thinking he’s thinking like, “I’m not going to be able to hire Bruce Clay. What can I do as a small business owner? What are the main things that I could do to improve my SEO in a good way that’s not going to kill my budget?” What are the simple things that everybody should be doing?
Gary: I think you want to reach a state where you are comfortable with a lot of content that you have on the site. It’s a myth that you always have to publish new content. If you have stable, static content, that can also work pretty well, and you don’t have to recrawl that that often because as soon as we realize that you don’t refresh the content, we would just crawl less, but that doesn’t mean that you cannot rank with that.
[Tweet “Google’s Gary Illyes says idea that you constantly need new content just to maintain rankings is a myth. More at”]
Once you are comfortable with a lot of quality content you have, you want to make sure that people know about your service, people know about your website. So you probably want to talk to people about your website. You want to convince them that you have quality content or quality services.
For that, you probably want to implement something like reviews or ratings on your site, which again, can be done with gadgets or widgets or whatever they are called nowadays. So it’s not complicated.
And there are social networks. I see that on Facebook sometimes, people post that, “Hey, I found this amazing little grocery shop, and they have Italian stuff, and they are actually importing Italian stuff, and they are amazing!” And actually this happened when I was in Dallas.
There was this little Italian grocery shop, I don’t remember the neighborhood, but they had the most amazing sandwiches ever. And they sometimes post on Facebook or whatever social network and they tell people that, “We got this new product. See what others said about it!” and that’s it.
And the more you talk about your business and the more people start talking about your business, the more visitors you get and potentially more customers. They will be also very targeted, because that’s what you actually want, targeted users, because they know why they are going to that site.
Someone told them that it’s great for something, like Italian pizza or whatever. So they are going there for that purpose. And yeah, I would do that, talk about how awesome my site is and tell everyone that they should go there and buy stuff. Just don’t stand, turn to the black side.
Eric: Well, there you go. So from time immemorial, it’s not enough to produce a great product. You’ve got to market it effectively as well.
Gary: Yeah.
Mark: You didn’t hate that question so much after all. That’s good.
Eric: Way to go. Well, anyway, yeah, it looks like we’re at the top of the hour, so it’s time to call an end to it. But Gary, thank you so much for taking time to talk with us today. It’s been awesome. I’m sure a lot of the viewers got some great value from that and Mark’s fantastic job with the Q&A, as expected.
Gary: You look a little green today, Mark.
Mark: I feel a little green. I don’t know why. It’s strange, isn’t it?

Eric: Yeah, so for those of you who don’t know, Mark is in our video studio upstairs, so he’s got a green-screen wall behind him. So maybe next time we’ll have him wear a green shirt and make Mark disappear. It’ll be a floating head. We really ought to try it out.
Mark: There are people out there who would love to make me disappear, so that might be good.
Gary: Oh, come on!
Mark: I just want to take a quick serious moment to thank all of the people who registered for this event and especially those who submitted questions. We have many, many more questions than we could get to Gary in this half hour, but thank you, everyone, for participating, thank you for being here with us.
Gary: Thank you for having me, guys.
Eric: All right, thank you and have a great evening, Gary. We’ll see you out in California at SMX West and we look forward to that. Thank you, everybody.

 *   *   *   *   *   *   *   *   *   *   *

We’re grateful for Gary’s comment in this interview that sites with SEO managed by Perficient Digital are “doing awesome.” Find out more about our awesome SEO and other services!