 All right. Welcome, everyone, to today's Google Webmaster Central Office hours. My name is Jamiro. I have a bunch of special guests today who are joining live from the Google office in Oklahoma. So do you want to introduce yourselves quickly? Hi there. I'm Jamiro Alburriker. I'm a technical S.C.R.O. and C.R.O. Technical...whoa, what? I don't blame them. My name is Ashley Berman-Hale. I do technical S.C.R.O. as well. Hi, I'm Elliot Mellichamp and I do S.C.R.O. at Macy's. All right. Awesome. So we have a bit of an odd hour today, so I guess that's why we don't have that many people at hand, but that's perfectly fine. More questions for us. Yeah, more time for you all. Exactly. So I have a bunch of questions that were submitted and I didn't actually read through most of them. So we will kind of see how far we go and I'll try to leave a little bit of room towards the end for live questions that come up. And if there are questions that you all want to ask, then you'll be able to jump on in too. All right. So here's a question from Ben to start with. We have some branch pages for our different proposals. If we wanted to rank for these specific services at these branches, how important is it that we have an internal link to those relevant categories on our website or that our content on those branch pages mentions those services we want to rank for? Is Google clever enough that already given the category in branch pages? So I think in general you should definitely have content on your pages that we want to rank for. So that's kind of the first thing that needs to be done in order for these pages to show up. But otherwise there are lots of different branches. What have you all been seeing in practice? What do people do? We just make a thousand pages with the exact location and keyword match. Oh my, sorry, I'm not allowed to be sarcastic on me. Yes, yes, yes. Do that. Somebody just said yes on the internet right now. So I think, especially if there are different locations, make a different location page. And what I've seen with some bigger sites that have different branches is to at least talk about top services or top products. So not relisting everything, but maybe if there are big products or services that are unique or in demand for that location, add a piece so they know that you offer them there. I'm Jason on LD, another great option using the structured data to mark up individual departments that report content information. Oh yeah, of course. Also phone numbers, open hours, all of that. May I have one related with this? Somehow at least it's about localization and different branches, country, state, cities. We'll see, we can answer. Okay, so here's a thing, it's a site about specific dog breed. And we are kind of making lists and promoting breeders from different states and different cities that are not so many yet. It's a new designer breed. And think is that for certain states we have only, I don't know, maybe five breeders on a city level. On most cities we don't have any and on those we have any. It's usually one or maybe two, not more. Now we wanted to make specific pages for each state and each city, not each city, generally each city where we have breeders. Now the options would be two, three options we have here and I don't know which one is best and would like a suggestion. The first one would be to have a generic text and just the list of breeders, but that would be too much boilerplate and may look like duplicate content. The other one would be to just list for breeders, but that would be thin content and might be penalized for we don't have enough content on those pages. And the third one would be to try to write something about the state or the city, general things, original things, but that would deviate the content that we are talking about specific breeders and specific breed and not the town generally. So what would you suggest? From my point of view, I try to make sure that these pages actually have good content. So not like random content kind of rewriting from Wikipedia, not just thin content, not just boilerplate content, but really make sure there's something unique on these pages so that we can stand on their own. That's kind of the most difficult option, of course, but that's kind of what we're looking for. Yeah, but it's not really anything to write about. You can't just feel either you divergate from the main topic too much, talking about state city or the state on topic and you don't have much content. But people are actually searching, we get traffic, we have rankings for those pages but I'm just worried that they are not high quality. At this moment it works, but I'm worried that it's not really good practice. I think for those cases, what I would do is maybe combine a bunch of those pages and put them on one page. So you have a bunch of locations in a general area and you don't have more content for those locations than just make one area page and this is what patients are rather than creating individual pages that have very little content. For example, instead of states, you have Western US, Eastern US, Western Europe, Eastern Europe, something like that. I think any time you're like saying, oh Google might see this as being low quality content, then probably you do see it. That's already kind of like saying, well, you know you should be doing something different. Okay, thank you. Alright, so next question here is, we use a Google approved third party reviews companies to collect our reviews from us. These are hosted on their website and it goes on like how can we best showcase these to help our site so that we get maximum benefit from Google Trust. So I guess, first of all, we don't use these for rankings so we don't need to use these for SEO purposes. And if these reviews are about your business, then maybe highlight them on your homepage but not on all of the other pages on your site. That's a really common mistake that we sometimes see that people just copy and paste those same reviews across the whole site. So that's kind of what I would aim for there. It doesn't really matter if they're hosted by third party or not. And again, it's not something that changes your ranking. If they're hosted by a third party though, let's say a widget or a frame that they're using, what would they want to do for any structure data? So I feel like the follow up question would be how do I get my gold stars? The gold stars. Gold stars. So we do render the pages. If the widget includes structures out of markup in the rendered version by JavaScript or whatever, then that works. Make sure they don't mismatch some of these providers we're using at microdata and your site manager, jsnlb, jsnlb is preferred. So you get one language. Yeah. That definitely makes things easier. I like to mix them up. I like to go to a restaurant that has Korean, Chinese, and Japanese food. Well, these algorithms don't care that much about how you mix things up, but it makes diagnosing issues a lot harder. Especially when you have the same markup in jsnlb as you have in microdata and it's like, which one does Google use? It's kind of up in there. Your website sells a service which involves a financial transaction. Are you treated negatively in any way if your site is not on HTTP? Well, I do not feel good about financial transactions on HTTP. So when I'm the boss, yes, you would be treated negatively. Okay. I'm not the boss. You're not the boss. Not the boss. Yeah. So from search's point of view, we don't look at the type of site when it comes to HTTPS. There is a small bonus for being on HTTPS, but increasingly users care about HTTPS. And if you want to use modern web technologies, you almost need to use HTTPS, too. And you're going to get more aggressive browser alerts, too, if they're using Chrome, depending on the type of pages of information. So from a user trust perspective, I think they're going to get flagged a little more easily and it's going to re-trust with your visitors. Yeah. So especially if there is something like a passport field or a credit card field, then I believe the current version of Chrome already shows like not secure. And the page. So that's something to kind of keep in mind there. Let's see what else we have here. A quick question. Yeah, John. One quick question, please. All right. Maybe I'll take a quick question from the audience. Yeah. And regarding the Google businesses, the local businesses, we were facing some trouble in UAE, basically. So we have post boxes here and the Google postal code is not received for like, I think I had around 17 clients till now and I didn't receive any one of them. Right. Any manual thing to overwrite and register the business? I have no idea. No idea for that? I don't know how Google My Business handles this. Okay. And the second thing about what do you call? May 12th. Okay. There were some huge things going on in the SERPs. I don't know. We were, regarding, it was some algorithm change or it was like just, I don't know. We make changes like almost every day. There's probably some others there. Something big or... Oh, maybe it's like mess up with your stats. Yeah. All right. I just wanted to let you know regarding the maps, you can always call, if you have bad words, call Google support, press three and you'll be directed to Google Maps and just ask them what's happening with your clients. How's that? Oh, very cool. If you'll tap off and then just hang out there. Press three. Press three. Okay. Okay. I have no idea about Google Maps. I'm kind of happy that I don't have to answer all of those questions because there's enough web search questions to go around. So, okay. So this, it's like a quick question, which is a big wall of text, but I think I know what you're kind of thinking out there. In a recent WordPress update, it adds well equals no open or no refer to all links. So is that a problem? Basically, that's no problem. From the search side, that's something that's essentially a markup on a page every evening. So as long as it's not a realm of follow, then we pass all of our signals normally. If it has other rel attributes attached to it, that's totally up to you. That's, I've seen some people that try to take this out, some people that don't care. I don't know why WordPress made this change, but I'm sure they have really good reasons for that. They don't make changes just right. But at least from a search side, you don't need to do it. How does Google treat on-click these days? On-click. So if you add like an on-click handler to something. We try to recognize that, but we can't guarantee that. So especially if you have something like a data and you add like an on-click to it somehow, and your on-click function does some fancy JavaScript and then navigates somewhere else, then we can only go so far. Yeah, I mean, we don't know what to click on essentially. And we don't know which of your JavaScript you can execute to see if it does any kind of redirects. A really weird scenario that we sometimes see that I know that the rendering team was kind of battling with recently is all of the infinite scroll pages where if you scroll down, it changes the URL on top. Yeah. Where we kind of talk Googlebot to scroll down so that it actually does the lazy loading of images. And of course, if Googlebot starts scrolling down, then suddenly the URL changes and then Googlebot thinks, oh, it's a redirect. Which is not really the way it's supposed to be. So that's something where we kind of have to hold ourselves back a little bit and not click on everything, not do all actions to see what happens. So if you have a JavaScript-based site, I just make sure that you, Dom, has the normal A-links, with A-elements in there, with normal href links. Could you use a fetching render to go ahead and see what they look like? You can see what it looks like, but you can't see the code. So they're working on something to make it possible to see the render code, but you can't see the browser as well. What could cause good ranking in mobile, but not so good ranking in desktop? So what might be happening is that a lot of your competitors have bad mobile sites, and we've made sure them so highly. So your site kind of makes it a bit higher. But I'm not aware of anything specific where you would be kind of devoted on desktop or whatever. What if they're doing something with their configuration, depending on other mobile set? If they're detecting user agents wrong or serving content wrong, so maybe the bot doesn't see the desktop just much anymore? I don't know. It's hard to say without looking at the data. So I suspect these kind of technical issues are things that you would notice very quickly if you look at the site. Loading it on desktop, loading it on mobile to see kind of what's happening. Otherwise, I'm not really sure what that would be. Quality site plus links versus relevant site plus the link, which one is more powerful? I don't think there's like one answer to that. Quality versus relevance. We try to show both in the search results. So we think both sides are kind of valuable. I don't think it's worthwhile to kind of differentiate between that quality and relevance. Okay. Another domain. How does Google identify that it's the same company blog and not some random other website? For the most part, we don't understand that they're actually the same company, but we pick up a lot of similar signals when we look at two sites that are kind of about the same thing. So that could be a company blog. It could be different parts of the same company. Usually there's lots of information there about which company is rich. Would the markup, so the entity markup on the company help with their own both sites? Would that be to get into consideration? Same as. I can't get you to say, I don't know. Maybe let them work that out. Okay. Okay. Okay. Like whether we see it as one company or not. We'll just try to rank it. It's not that, oh, this is from this official company source. They're probably going to rank it higher. It's more like, oh, there's good content here. And it's linked from these other sites. And we have all of these other signals and it's something we should show in the search results. In fact, it's still, can I go ahead and put this on your primary domain? Subdomain or subdomain? Um, you can. So from the side, you could do it. I think putting it on the same domain makes it a little bit easier because we can kind of like put it all together. Subdomain would also work. Sometimes there are technical reasons why you can't do that. Subdomain on the same domain. But if you can put it on the same domain, it's usually the cleanest option. Oh man, what's happening on Twitter? It's like, oh no, oh mother world that I'm missing out on. Bobo, John has it. Actually, it's the most popular one of us. Okay. I have a question regarding interstitial penalty. Since we've removed our pop-up on mobile, we've seen a higher bounce rate. And that makes us question the mood. Our email signup pop-up is great incentive that people take advantage of. So even though it's intrusive, it does end up being received. It's obviously not going to be a bounce rate because you had to click out of the interstitial, which makes a lot of bounce. There was a click. Yeah. I also kind of suspect that it's kind of something like that. So it's not that they're engaging with your site. It's that they're engaging with the X button and like the back button probably, which probably isn't to your best. But one thing you could do is turn it into something like a banner instead of a pop-up so that it's on the page. And if it's really something that people love and they want to fill it out and like, they really give you an email address and you can still do that. What about just delaying it for a few seconds? So it's less about the interstitial. They're worried about the penalty. I know it's less friendly to users, but... Even if there's like a five or something different. The banners are nicer. Yeah, banners are different. May I have one more related interstitial banners? All right. Okay, we have one, I don't know what to call it, interstitial pop-up banner. It just shows on the top of the page. 19 pixels tall, wall-page-wide. It stays there even if you scroll. You can click on X to make it disappear. It's only on desktop side. And it has principally a VAT. Sign up for news, promotions, things like that. Would VAT count as an interstitial and get a penalty or not? As I said, it is site-wide long and it is 19 pixels tall. So it usually takes a small part of a screen, but it is where you click on X. I mean, it's more like a banner than VAT. So VAT's fine. It's no risk for penalty or something. Okay. Okay, thank you. As long as it's less interesting. Thank you. Sounds like you're on the right track. All right. Maybe I'll just switch over to questions from you all here. What has been on? What's happening over there? We're having a party. We've been drinking since June. I'm just kidding. It was 8 a.m. So Jamie and I are here for Google I.O. and the Women Techmakers dinner. All right. It's Wednesday. I just traveled from Canada to California. It's long for me, you know. Yeah, yeah. But I'm spreading the word. If there are any ladies out there for tech, join Women Techmakers because they do great events at these bigger conferences. And once a year, free dinner. Free dinner. Wow. Elliot, what are you doing here? Elliot lives here. You had a bunch of questions. I had one. Go for it. So I guess just mobile players, right? It's kind of hard that the mobile, like someone has a separate mobile site. Like what should that chemical tag look like? Should it be self-referencing? Or should it still kind of point to the desktop? That's a good question. So our goal with the mobile first index is as much as possible to make it so that webmasters don't have to make changes. So if we can figure out a way to kind of do the switch, which we think is the right thing to do, given that most people are on mobile nowadays, without webmasters having to make changes, having to switch to economicals, all of that, then that's kind of our ideal situation. There are some things where we can't do that. For example, if there's no confidence. So I saw a handful of sites today from engineers where they were saying, well, the desktop page has content in five different languages, but the mobile page is only in English. And then it's like, well, that sucks for you. But we have to figure out how to tell the webmaster about this. Because that's something we don't realize, because we're ranking them based on the desktop page. And with regards to canonical, our goal there is to just say, well, keep the canonical set up for the mobile now, which means on separate mobile URLs we point at the desktop URLs, we'll still pick the mobile to actually index, but we kind of have this connection between the desktop. I have a question. Agegames. Agegames. So we have lots of weird sites. Yes. The question about agegames. All of the weird impacts. Well, we don't want the only young ends to go ahead and to these sites. They're really in terms of age of 21. And it's very common when you click, no, I'm under this age to go ahead and be sent to another page. How does Google view if you were to redirect a user from that underage to an anti-drinking advocacy, anti-underage drinking? PBS. Yeah. If you took them to PBS, and take them to a warming Easter Rogers episode instead, would that be viewed as a... I think like where you send them doesn't really matter. The tricky part for us is rendering the pages because you're essentially providing us with this interstitial instead of the actual content for the renderer. And I don't have any perfect answers to that. So we've looked at that with the rendering team and they've looked into different options that we could recommend like maybe marking up the yes button or whatever is needed to actually make the interstitial go-way so that we can kind of navigate through that. Because I think Googlebot is like over. So... It seems so many terrible things for that. Yeah, that's true. That's true. Well, Googlebot. But I don't really have any comprehensive. So we're trying to work to just like make the existing set-ups work. Some sites use normal interstitials with JavaScript and we can just leave out the content behind that. That's no problem. Some sites have a robotic JavaScript. That works for us too. We don't want to recommend using a robotic JavaScript because then it means we don't see what actually the user is. Actually, it looks like poking potentially. We don't know. I think we don't know. That's kind of the tricky part there. What we should avoid is redirecting to some shared URL. So that's something that we sometimes see. Because then we see the redirect to this shared URL and maybe that shared URL is blocked by robots or just the same boilerplate for every page of the website. And then we don't know what's up with this thing. Whereas if we see the interstitial on the page on top of the normal content, we can kind of oh, this is an interstitial. We can ignore it. It's there for legal reasons. We can kind of see that. John. Yes. Just regarding like speed, you know, because you're talking about the whole mobile stuff. So if the site, according to Pingdom, like I'm just watching the specific site at about 7.3 seconds low time and then it goes back to 5.7.5.7. So for the past two weeks, it's been 5.7.5.7 and lost about 25% of its traffic. And according, I don't know, there's just a couple of blogs out there that I saw that also claimed that it can go as 20%. But I mean, on my end, I'm showing like a 25% loss in traffic, which is huge. I was, what I'm thinking, maybe the load balance or I don't know, there's so many things to it, but I've been watching it for two weeks and it's kind of crazy. So is that a silent penalty? No. I mean, shouldn't be. So with the speed with regards to figuring out which sites are really, really slow and which sites are kind of normal. It's kind of in the normal range. So the fluctuations that you're seeing when it's like a couple seconds longer, a couple seconds faster, that wouldn't be something better. It would be more a matter of like, oh, this page takes, I don't know, five or 10 minutes to actually load and I don't know if we should be showing this to you. So five to seven seconds doesn't, it's okay? Yeah, we don't have a straight cutoff time so we'd say this is a problem. If the user is on here, he's pretty annoyed after six, seven seconds, he's going to leave and obviously it's not good for you guys, right? Well, these are slow on mobile unfortunately. A couple of years ago at Google I.O. we told everyone to make their mobile sites slow within one second. That was ambitious. I think a lot of people listen. So the guy who announced this is now my manager so that means we would care about speed. So you'll probably hear more from us about speed anyway. But it was a really ambitious goal and I think it was important for us to kind of bring that out there because it did encourage people to think about speed a bit now. So that's something where they might not make it one second but if you can get it down for like two or three seconds it's already way ahead of the people. So next year we'll be talking about milliseconds and seconds and so on. Yeah, I mean it's hard. It's not trivial to make the site slow in less than one second on mobile. Because they had about 200 and 208 requests and from what I'm seeing now they kind of got it down to about 80 requests so I guess it's going down. So the bigger effect that you would probably see there is just with the dots to convert. People on the site, staying on the site, clicking into more parts of the site, all of that. That's something that probably you care about more because getting people from the search to visit the site is one thing but getting them to actually find something is probably worth more than just like you and your statistics. Okay, thanks. We'll be going through an experiment where you measured so if you are using Kingdom and you like it or web-page-test.org and run a bunch of sites and see how that correlates to the page through insights to get a gauge for more. I wouldn't, it's not, it's correlative, right? Not positive. If I get a gauge for the seconds that are measured on those tests, the thing is your actual time to store numbers, page through insights to sort of see where the benchmark side of that. And the thing is page insights, I mean, they're 99 out of 199. What are you asking for? No, it's just not, no, it's just not... You guessed it, improved. I'm thinking more, I'm thinking more, it's a non-stop of a DDoS attack as well for a long time. I think it's also important to kind of like not look at just the numbers when it comes to page through insights but actually it kind of comes about what would people actually see. Of course, of course. Where you should hire. So you have limited developer resources should they be working on a redesign with like a new technology that's faster across the board should they be tweaking HTML and CSS to like 10 milliseconds more. And somewhere in between there is finding the right balance. I did hear that John will hand deliver, I can't remember what it was, so if you do get 100 out of 100 for both desktop and also C, you should keep pushing for it. And deliver it. Okay. By drone. By drone. By drone. Showcasing the 100 of 100 sites. So actually, I don't know how far we are on this but one of our plans is to get rid of those numbers in page through insights. No, no, don't do that. The numbers are good but I saw there was a UI change during the weekend that actually paid some learning for the numbers so I saw four or I was like it could be optimized but it looks like your graduate is not like. I don't know what the size of that is. Please don't get rid of that. Please don't get rid of that. I mean, it's something where we see a lot of people focusing too much on just those notes rather than. No, no, I just think it's more of a, John, I just think it gives the people more of a, he just really gets, you know, a high from that. And it's more of a developer thing. No, I mean, it's kind of gamification. People love to get out of that, but it's also important from our side that we don't kind of mislead you with like this laundry list of things to do when actually you could be making inside a little bit better and getting a lot more conversions because you actually have something useful on your side. Self-confidence comes from within. Yeah. By the way, yours are frozen on my end. I hope we made a good face. John, I have a question for you. Okay. John, I have a question for you. Can you hear me? Okay. Okay, awesome. So we have a bunch of partners that syndicate our contents and some of them are complaining about actually adding the no follow to the links that live within our articles. My questions for you are, will this, excuse me, the no follow tags in the links, will the canonical tag actually work the same as a no follow if they can't place the no follow in those links? What do you sense? So I guess, can we substitute the canonical tag for the no follow? How do you mean canonical for the no follow? I'm sorry, you broke up a little bit. How do you mean? Can we place the canonical tag in the original article and then in place of the no follow tags that live within each one of the links that point back to our side? Yeah, so with the canonical, if we follow the canonical, we would focus on the main version of that article. So kind of the no follow or not links within that article would not be ignored. But I still think it's a best practice that if we have set up no follow links within the article, if it's like a paid placement, a tag or anything like that, then with those no follows in there, is the right thing to do. And is it okay to follow the link back to the domain or the original article if they can't put the canonical within the headcode? That sometimes helps us too, yeah. And so the relic tag is definitely the ideal situation where we can fold everything together with the canonical. That's a link back to the original article also kind of helps us. Okay, thank you. Do you want one small question? Sure, go for it. Yeah, I wanna blog. So in a blog, if I wanna just target distinct categories, for example, more which we focus on is marketing, all right? So if you wanna focus on something like travel or something on the same domain, can we do that? I don't know. So in a blog, if I wanna just target distinct categories, for example, more, more, all right? So if you want. Oops. Can you hear me? So can you repeat that briefly? Sorry. For my blog, whatever spanning, that can we just target distinct categories which are not related? For example, marketing, then travel or something like that? Sure. Will there be any issue in the dilution of the content or something like that? Because one of my author here loves to write about the travel or something. No, I know this guy, Matt Cutts. He writes about all kinds of stuff. Okay, cheers. You can do that, you can do that. Okay. He's right behind you, by the way. He's right there behind you. Yeah, I'll just let him know. It's fine, I'll let him know, I got it covered. I'm a good blogger. Question, can you clarify if AMP is gonna be supported in the mobile first index? I don't know. AMP supported in the mobile first index, yeah. Or, yeah. Like, what do you mean with supported? Maybe you should clarify that. Gary made a comment in Twitter a few weeks ago. And I hate to be the one that calls out that kind of stuff, but I'm gonna do it. The fiction. Yeah, that's what they say. We know where to find him though, right? So I know that AMP figures create some differences in different versioning, and he had mentioned that they weren't quite compatible with the mobile first index yet. And that was extrapolated, I think that maybe that's part of why the mobile first index is a moment in college. Will everything be kosher with AMP in the mobile first index? Yeah, we'll proceed with the planning. Okay, follow up question. So actually, I guess, let me just, like, one small clarification. So I think, like if you have separate AMP pages and you have mobile pages, that set up is fine. You can also have your AMP pages as the mobile page. That's something that's kind of tricky to set up if you have a separate desktop page or an app like the mobile page. You can also do that. Kind of like the mobile website. And you can just have an AMP version as a canonical page, which we will use for both desktop and mobile for the database. So ampproject.order, this is what we want. We want the one that should just work. So the trickier part, I think, is on AMP itself, there's some limitations with regards to what you can actually do there. So if you have this big file of structured data that you want to put on the page, then you need to watch out to make sure that it's like from the size point of view, not too much, but the AMP pages stay. So that's kind of the trickier part to watch out for, but if you have a normal setup with like the desktop page or mobile page or AMP page. So repeat that by slowing down your structured data for AMP. If you want to only have an AMP page instead of desktop or mobile, then you need to make sure that the AMP page remains there. But can you, you can piecemeal AMP components and borrow it for your own sake, because it's just really kind of hard, right? So even if you're not AMP valid, you can still benefit from AMP. Oh, sure. Everything should be fine. Well, like adding it like AMP is, I mean, you have to, you guys for all, sorry, like if I was gonna add it to a hundred thousand pages, it's not that easy to do that on an HTML site. Like a site, for instance, as an example, like Ikea, it's HTML. You can't just go and like go nuts, like do, it's not easy, right? Well, just because it's not easy doesn't mean you can't do it. Well, because you have like half a million pages, that's... I'm not good to do things that aren't easy, but it's true, like for some kind of site, it's harder to move everything to AMP. And that's something that the AMP team is definitely working on. I don't know what they're planning for Google I.O., but they do have a big session about AMP, so I'm sure they'll have some cool stuff to present to us. And they have something on Progressive Web Apps with AMP, and Paul has promised that on-state pool is a cool AMP. So... Yeah, I'm sure we're gonna do that. I'm waiting for that one as well, I'm gonna hold them to it. But with Progressive Web Apps and AMP, what are the best ways that SEOs that are exploring the technology and we're working with them easily? How can we test for the standard SEO stuff? Is the technology moving a little bit faster than our industry and we wanna make sure stuff is called an impact properly? What do you recommend? Um, having good SEOs. So that's kind of the... Higher. Yeah. No, I think that's actually really important. So especially with the newer technologies, everything around Progressive Web Apps, single-page apps, they need kind of the modern JavaScript frameworks that are out there. That's something where if you only have developers who are good at these technologies, that doesn't necessarily mean that these pages will work for the jobs. So sometimes we see really fancy sites that are really nice to use or fast, they're kind of interactive. But when we look at them, we see, oh, they all use the same URL or they don't link between the different pages or they link it away that you can't actually pick up. And these are all things that SEOs have to talk about. They know about title tags, they know about descriptions and those are the things that are really important for search. So it's like, you shouldn't just focus on technology, you shouldn't just focus on SEO, but you really need to kind of make sure that those sites are working properly. And with the PWAs and with all of these modern frameworks, single-page apps, you need to understand that Googlebot is running these pages and you're trying to figure out how you can diagnose that on your side. Is Angular still terrifying when you use cache with push and sl and let's just say the placeholder and you want it to set up? So, caches. That's a different question. So the Google cache is just a live knowledge. So if you look at the cache of a single-page app, you'll just see this. Don't panic, it's supposed to be that way. Yeah, so you should just search for the content correctly and you should be able to find it. I know we can get the DOM and get a lot of good details, but I do still live here for the original days of just viewing the source and stealing code for people and see how they made that because it's fun. So it's like good community love and it was making it harder and harder. Just kidding, I had to throw that in there. I love the new technology too. At least you don't have to program it with all the stuff. Can I ask you a related question? Bring my geosuits back. What's up, David? A related question on that. Is there any chance that we'll ever get reporting in Search Console about how frequently Google is trying to render the full DOM across pages on our site? Probably not because we're trying to actually render every page. So it used to be like maybe a year or so ago we would render like most pages, but nowadays we try to render pretty much every page that we actually could make. So it wouldn't make sense to highlight that in Search Console because we should be able to render every page. I assume we'll get some more tools in Search Console with regards to rendering content and how we keep that up, but I don't know what specific plans are. Also I set up a kind of a working group for JavaScript sites and modern sites where if your developer SEO were interested in that, we're going there. It's going to be kind of a place where we have those kind of discussions around how those modern sites work well in Search or how they don't work well. So that's kind of useful I think for both sides. For us to kind of see where people are having problems, but also for you all. So tomorrow the whole I.O. They'll maybe releasing, they're going to say when Google first index is going to be live tomorrow or no? I can't pronounce that. Okay, I understand. I don't think there's anything. So sorry, just closing out the whole dom thing. That means for every crawl we can assume every URL that you're requesting you are also rendering at some point. Yes. So in that case, is it safe to assume whatever our crawl, whatever our crawl budget is, whatever the crawl pattern is in our particular site is in part informed by your experience rendering the pages? Yes. So when we do that we don't fetch all resources all the time when we render a page because I think that would kill your server probably. So we try to cache a lot of the CSS and JavaScript on those pages, a lot of the images on those pages so that we don't have to kind of constantly request shows when we crawl through that site. So if you look at your server logs you probably won't see us crawling your JavaScript page as often as you probably would. So if I have an intermittent issue with like a tracking tag making it so the load event doesn't happen for like two minutes or something crazy like that how often will you need to see that before you decide I'm going to pull back on trying to request this URL? I don't know. I don't have any specific numbers on that. So one thing kind of to mention there is if you're making significant changes in the JavaScript then I would consider adding some kind of version into the URL so that we do understand it's in the JavaScript file and we can drop the old cache and pick up a new one. That's something that I believe most of the newer frameworks do automatically so that kind of helps us. Rakan, thanks. John, one small question. All right. Yeah, for the AMP basically what I understand is it's coming from the cache, all right? So like if it's like an AMP page is going to a 404 okay how long will it take for a webmaster to fix it without losing a ranking because recently we were having some changes with the SSL, okay? So I think most of the AMP pages were not found so the rankings were there for some hours and then it vanished, okay? So approximately how much time do we get to fix these old things? So if it's a matter of the AMP page being 404 then we would probably just drop the connection between AMP and the page. So we would just show the normal page results. If the normal page is also 404 then that depends a bit on different factors. So it's not something where we have like a strict timeline and say, oh, we made a 404, therefore one day or two days, it kind of depends on the URL, it depends on the calling, all of that. Okay, thank you so much. All right. So what am I missing on Twitter? Yeah. I would have one more if possible, John. Can I ask a question after that? Sure. It's a site, another site. And they have pretty good rankings for all their pages but after looking on them, I personally think they are rather thin, rather short, something like 200, 300, what's articles. So I suggested them matching some of those pages like two pages in one, something like you said earlier in the other case because while they rank at the moment and they have no penalty, I don't think the quality is high enough and it may happen something bad in the future. Now the problem is that they rank for each pages. What I suggested is to try to keep one of the pages, drop the other, the redirect 3100, redirect the old one to the one of old one, to the keeping old one and just match the article, the content. Now, another option somebody else came with was to create a whole new page and redirect 3101, redirect both old pages to a new one so it can have a new URL. So something like if one page has six reasons for doing something and whoever has four questions about doing something, it can be a new error, can be 10 things about. So what do you think is best to drop both pages and redirect to a new URL or just keep one of them as I suggested? I think both work. Sometimes one is easier than the other but both are the kind of reasonable strategies. I think anytime you're making this kind of a change, especially across bigger parts of the site, you will see some fluctuations. So that's something where you shouldn't kind of freak out. It's like after a week, suddenly your site ranks very different because it will just take a bit of time to set aside. Yeah, but if you make the whole new URL thing, you're not, on long term, you're not losing metrics for two pages. And there's no problem if you free one two pages to a new URL, is that it? That's perfect. I know it's not okay to free one 10 pages to one URL, but two pages is not no problem, no? I don't know where we would drop along. Like when it starts looking like a software. But not a two. Right, but two is the same. Okay, thank you a lot. I'm wondering about if we're using a reviews widget on our pages, if we put a canonical to a specific page, it is actually the reviews page. Is that okay? Can we have the reviews in multiple places as long as we're kind of saying the original content sits on one page? If you put the rel canonical on those pages, we'll probably just index the canonical page. So that probably isn't what you're looking for. But what you could do, things like this, if you have reviews for like the company, have the widget mark them up on one page and just make them visible on the other pages without having structured mark up the house. So that's not from there. You could like go really low tech and just take a screenshot of the widget output and just copy paste that as an image of what you want to. Okay, because I was concerned that it would look like duplicate content across the pages. If we had the same reviews in two places. That's no problem. So do the content there is not something I worry about. It's really a matter of if you have the same structure data mark up across these pages where you're essentially marking up the company instead of the individual products, then that would look kind of strong. And in terms of the markup, if there are reviews for a website, should that be like it's classified as an organization like in the structure data markup or what if it's a company that is not selling a product? So the new markup should be reflecting the primary content of the page. So it shouldn't be like you're reviewing the blog post or you're reviewing the website. It should be what is this website about? What is this blog post about? So that's kind of what we should be reflecting on. So it shouldn't be like review markup or the website itself, unless you're like a website that is reviewing other websites in which case obviously you're reviewing. Just to clarify, we want to avoid going out and then putting product reviews for every product that's in your commerce category on the category, too, which itself. Exactly. It should be like one, one. Can I transition into that question with widgets? If, for instance, suppose I made a widget called abc1.com, just that's what it's called. And then I go ahead and I developed this little tool. Now everybody is going nuts. They're placing it on their website and I've asked them as well, which I'm not supposed to, go ahead and put it on your website, blah, blah, blah. I know it does me good. But now it says, you know, ordered by abc1.com. And those are all my links. Those are all my links in my portfolio. Those are the only links I have. That's... Sounds like widget links, I don't know. But no, what if like you have now 12 million of those? Is that, is that a good number? Serious, though. Is that against your guidelines or not? So, widget links. Yeah, I would like to read that down on the help center we have with regards to, like... I'll have to send you a serious email and so you can see what I'm talking about and really understand because it's very upsetting and it's just when you have that amount and with an anchor tax... We recommend that people put no follow on links like that. I know, John, I know. It's just, I see the stuff, I go nuts. I mean, for large part what happens is we notice those things too, especially the things that are in larger scale and we just think the no follow there are. So essentially we probably site, we look, see these links, we say, oh, this is like all the same widget, we will just no follow all of these links. You'll notice, you mean automatically, right? Algorithmically. Yeah, we understand that webmasters don't really know what they're actually trying to do there and we will just say, okay, we think this is like widget link building and we will just no follow it. So just because they have it on those pages and just because they're ranking doesn't mean that they're ranking because of it. Did you guys think in 1996 that the word widget would just be caring on for 20 years? No, you're not, it's just widgets. John Badrat is done algorithmically, not manually, right? We do it on both sides. So whatever we think is necessary to make sure that kind of the quality and the relevance of the search results. And if done manually, would it be a penalty or just not counting them? Normally it would be a manual action. So we would notify the web sites about that, but that doesn't mean that we would remove that site from search or that we would demote it for this, we would, we would see these links were ignored. Yeah, but then you have these other penalties and then, but then you have these other penalties that the site might just go for something else that's maybe related to that. Cause then you have the separate widget penalty as well that you've introduced and it's on your blog, right? I mean, you wrote about it. So many widgets, yeah. So I mean, it really depends on what the issues are. All right, sounds good. All right, so I think we're out of time. So we close out. Can I ask a quick question? All right, really quick. All right, so we have a Google Webmaster Tools API connection where we have 123 new tool agency customers using them. We're consistently seeing lower numbers for the query data view through the API and what's provided in the Webmaster Tools UI. And I've seen a few comments online that other people are experiencing this. I was wondering if it's a known thing, if there's something we're doing wrong. So I think there are two aspects there. One, it could be a problem. So it'd be useful to kind of have more information about them. But for the most part, what the big difference is, is that in the search console in the UI we show the number for all the queries, all of the URLs, and in the query data view we aggregate essentially the queries that are not filled in. And in the API, we only show those. So if we add up those entries there, we would get a different number than the sum on top because some of these queries are actually there. So that's what you're seeing there. Especially if you go and kind of drill down to multiple parameters and try to really pick out individual queries, then that's something that would be more and more likely to happen. Okay, thank you. All right, so let me wrap up here. Thank you all for joining, and for coming by long. We're having visitors. Thanks for all of the questions that were submitted. We didn't go through too many of them, but we can complicate some of the next thing out and we'll try to run through those. Or maybe you put them in the helpful link that you can see here to kind of help out and answer some of these questions. Because a lot of this is something that people have seen before, where it's useful also to get input from peers rather than just Google official ads. Thank you for hosting us. Yeah, thanks for joining at Crazy Times, dogmas. Yes, thanks for joining us. All right, bye everyone. Thanks everyone.