 All right. Welcome, everyone, to today's Webmaster Central Office Hour Hangouts. My name is John Mueller. I'm a Webmaster Trends Analyst here Google in Switzerland, or, well, still in Switzerland, but not in the office at least. And part of what we do are these Office Hour Hangouts where people can join in and ask questions around their website and web search. And we can try to find answers, if that's possible. As always, a bunch of stuff was submitted already on YouTube. But if any of you want to get started with the first question, you're welcome to jump on in now. Hi, John. Hi. We have an issue with the meta description that is being displayed for our home page. So even though we have a meta description that is being implemented on that particular page, somehow in Google, when our website appears, the meta description is completely different. And in some cases, if we search for our company name plus the word UK, the meta description makes no sense whatsoever. It's just a bunch of words put together from various parts of the page. I know sometimes Google goes hunting for various things on the page if it cannot find relevant content for that particular region. So I guess my question is, because we have a lot of traffic that is coming up from branded searches, so it is important for us to have the correct meta descriptions showing up, what do we do to rectify the situation? It's hard to say without looking at the search results. So that's kind of the one part. Usually what happens is we need to have the description meta tag on the page. So that's kind of the first step. It sounds like you already have that set up. The other thing there is that we need to be able to, I guess, trust the meta description on the page so that it looks kind of reasonable. In particular, sometimes when we see a bunch of keywords that are just kind of collected in the meta description, then that's something that our systems might look at and say, well, this doesn't look that useful for users. So they'll try to rewrite something else. And most of the time when it tries to rewrite something, it's based on the content on the page itself. And the other thing, like you noticed, is the description can vary depending on the query that is used. So the first thing that I would do is just take the normal branded query that you use and double check that the description that you provide in the meta description is actually pretty useful and not too kind of spammy or overdone and then go from there essentially to figure out, is this something where Google always gets it wrong or is this something where sometimes Google's algorithms pick up something else on the page and get it wrong? If you feel that it's something that Google is always picking up incorrectly, then I'd love to have some examples like that. So things like the queries, maybe a screenshot of what you're seeing. If you could send me that or maybe post it here in the chat, that would be really useful. OK, cool. Yeah, thank you. Sure. All right. Other questions from anyone before we get started? Hi, John. I actually have a question around getting your website verified. We're trying to get our website verified through our analytics tag. We have the tag on the page. We actually have used the Google Chrome extension that detects if you have a tag already on the page and we're finding it in that. But when we actually go to the verification site and enter the URL for it, it's saying that it can't find the verification tag on the page, even though the user has all the edit access rights to that particular container that it's stored in. Do you have any tips for troubleshooting that? Hard to say. So sometimes this is something where the folks in the Webmaster Help Forum can help, because they can take a look at the common issues. One issue that I've seen a few times that has popped up is if the verification or if the Google Analytics code is not exactly the same as we would provide that within Google Analytics, then that might still work for analytics, because it's still proper JavaScript. But for Search Console, it would be something that we might not recognize immediately. So that's probably what I would watch out for there. Because if you're saying that it's showing it with the Chrome extension, if it's working for analytics, then it sounds like the JavaScript part was working well. It's working in a way that works well for analytics. But Search Console, when it tries to check that page, it just can't recognize that this is really that block of text that's used for verification. Sure. Great. Thank you. Sure. Hey, John. Hi. Hi, John. I am having a two questions on HL Flang Sitemap. The first question is, do we need to submit the same sitemap in all the language part? Let's take the example.com for the flash, F4, and the product name. And do we need to submit this the same HL Flang Sitemap in all the part of the language? And the second question is, now we are using the Brazilian Portuguese in our web pages. And what will be the HL Flang tag either? It's pt1ly, our ptbr. We are serving the language across the globe, not only the br, not only the Brazil. OK. So for the first one, in general, with regards to HL Flang, we need to see both parts. So we need to see kind of if you have two pages that are connected with HL Flang, we need to see the HL Flang markup on that first page. And we need to see the same markup on the second page. So if you have, say, an English page and a Spanish page, we need to be able to see the markup for both of them independently. So if that markup is within separate HL Flang sitemap files that you're using, then we would need to see all of those versions. Depending on the way you set up HL Flang within the sitemap file, it's possible that you also just have one sitemap file that has all versions. And in that case, that's also fine. But if you split up the language versions into different language sitemaps, then we would need to see all of them. So that's kind of the first one. The other one, with regards to Portuguese, with HL Flang, you can specify either a language alone or you can specify a language plus a country if you want to target a specific language within a country. So if you had different Portuguese versions, like Portuguese for Portugal and Portuguese for Brazil, then marking that up appropriately with the country and the language version makes sense. If you just have one Portuguese version and it's one of these two, then I would just use that language code and not the country code as well. OK, thank you, John. Sure. Hi, John. Hi. So we had a website and we extended it to multiple locations. It's a French website. And we forgot to put the HL Flang. So at some point, like Google de-indexed many pages because they're similar. They're all in French. But even if they're all in French, every page is for a country. So now we updated the HL Flang and we asked Google to re-index the page. The crawl happened, but the pages are still de-indexed. I can't get them indexed. Is it the same content just for different countries? Or what's the same content? But the title is different and the header one is different. But these are trainings. So it's just the content of the same training. But we have different headers and titles. OK. So practically what probably is happening here is we are recognizing that these pages are essentially the same and we're folding them together. So we're seeing them as duplicates of one page. And we'll try to make it easier and just index that one page. And with the HL Flang markup, essentially what happens is we don't change the indexing side. So we would still index these as one page. But if we recognize that the HL Flang markup is correct, then when a user searches in those individual countries, we would try to show the appropriate URL. So if you have a URL, say, I don't know, let's use English UK and a URL for English US. And it's the same content for whatever reasons. And we fold them together as a duplicate version. Then what would happen is we would index just one of those. And if someone is searching in the US, we would show the US URL with the snippet and the title from kind of the one version that we have indexed. And if someone is searching in the UK, we would show the UK URL. So the indexing side wouldn't change, which makes it really complicated to kind of debug. But in the search results, we would use HL Flang to swap out those URLs. And from our point of view, that's something that kind of makes sense as an optimization for indexing, because we just have to index one version and we have to keep crawling one version because it's all in that one version. But for the webmaster, it can be really tricky to monitor what is happening there. In particular, the index coverage report in Search Console will just show one of those versions as being indexed, even though we show both of those different URLs. And the search performance report in Search Console will also just report on one of those URLs, the canonical one that we choose. But you can see if you switch to the, I think, the individual countries that you can see in that report, you can still drill down and see how many users from this country saw that page and how many users from a different country saw the page. So it doesn't really fix the problem in that sense, that in that we still just index one version. But theoretically, we should be able to swap out the individual URLs. If you need to have both versions of these index independently, maybe for tracking purposes or maybe because you have a different rich result type, depending on the individual country, for example, then you would need to make sure that the content itself of these pages is significantly different so that we wouldn't assume that these could be considered duplicates. Yeah, but actually, I mean, it's the home page, like domain.com slash fr. I can't see it on Google, even in France. So even if I'm in France and I'm looking for the home page, I can't find it on Google. So the page isn't indexed at all? It isn't indexed at all, yeah. Google is off. If you search for the company name or? Yeah, the company name. Like domain name training, training center in France. And I can't find it. I can't find it. OK, but in that case, it wouldn't be a situation where we would pick one version as a canonical. So that wouldn't be a matter of the duplication in the URLs. But it sounds like that's really just something where we're not indexing that page or that site at all, which from a practical point of view, would usually come from three things. It could be a severe technical issue on the site from the start, essentially. It could be manual action from the web spam team, which is not something that would happen if this is just duplicate content. That's not something where the web spam team would take action on. Or the third one that sometimes happens is that someone used the URL removal tool, maybe incorrectly or accidentally, and accidentally removed the site from being shown in search results. Like Google Search Console is saying like they picked the canonical URL. They're just like indexing the Canadian one. But they're dropping all the other countries. Yeah, but in that case, we would still show the URL. OK. So that's something, if you're seeing that the canonical is chosen as one of these, then that's a situation where we would see this as being duplicate content, essentially. And we would pick one of those versions to index. But if you search for the company name, you should still see that result. It might be the Canadian URL. But essentially, we should still be able to show that. It would be different if it's really like the whole website is not shown at all for users in that country. Then that's more a matter of kind of these detailed technical issues. So the hreflang isn't given signals? The hreflang helps us to swap things out, but it doesn't change indexing. So if it's the same content on those two pages, that wouldn't be affected by hreflang. Because that would be penalizing for us, because a French guy would see like a Canadian URL in the best case. So he wouldn't click on it. If you have hreflang set up properly and we can recognize that across those pages, then we would swap out the URL against the local version. OK. Yeah. So it should. Yeah, it would take some time, maybe. Yeah. So it should look OK, even though from a technical point of view, the indexing side would still be indexing the other version. OK. OK. Perfect. Sure. Yeah, this kind of situation, especially with international websites and the same content in the same language, is really complicated sometimes. OK. Thanks. All right. Let me run through some of the submitted questions. And we can open things up to more questions from you all later on as well. We have some questions regarding the URL parameters tool and search console. In a significant number of cases, we've used the URLs with tracking parameters for interlinking. It's quite an amount of work to completely change this. So we decided to add parameters to the parameter tool. Now if we set the parameter to representative URL and to know this parameter doesn't affect page content, does this mean Google will only crawl the URL version without a parameter? Will it still consolidate signals between the parameterized versions but only crawl the URL without parameters? We don't want to negatively impact or interlinking because of these settings. So yes, this does essentially try to do the right thing in that it recognizes that we've seen all of these different URLs and we can fold them together into one kind of representative URL and use that one for indexing. So it consolidates all of those signals that we've seen and folds them into one version. With regards to crawling, if you choose the option representative URL, it doesn't mean that we would crawl without a parameter. It might be that we would drop the parameter. It might be that we pick one value with that parameter and just crawl and index that version. But from a practical point of view, both of those are essentially the same thing. And like I said, from kind of a ranking point of view, from a signals point of view, everything is folded together, so that would be fine. We noticed that half the parameters added to the URL parameter tools have disappeared. We had about 160 in January and now only 80. Is this problem with the dashboard displaying all the parameters, or should we add back the ones that are missing? It feels like if you have 160 different parameters that you're adding to the parameter handling tool, then that seems like quite a complex website where probably it would make sense to find ways to significantly reduce that number of parameters that you actually use. So parameter values is obviously something where you might have a lot of them. But individual parameter names feels like something where you should aim for a, I don't know, maybe 20 maximum, something like that, different parameter names. Because otherwise, it's really, really hard to understand how to crawl and index a website like this. So that's something where I don't quite know exactly what you're seeing with just the drop in parameters that you have listed there. But I'd really try to limit it to something that you can kind of watch over for a website. Since the last core update in June 3, 2019, our news website has disappeared in short top stories, AMP and desktop. We didn't make any technical changes, which could explain this. And we analyzed our whole domain. We don't see anything. What do you think could be affecting us in that sense? We review our news articles, and we don't see any issue about that. So the updates from last June are really long time ago. So it's really hard to say what specifically might have changed there. But in general, when it comes to core updates, we try to use these to essentially improve the relevance of our search results. And that's something where oftentimes, if you're seeing changes from there, it's not because of any specific technical issue on your site or anything specifically that you're doing wrong, but rather where our algorithms have recognized for whatever reason that maybe this website, this search result, wasn't as relevant to some of the queries that we had shown it for in the past. And we have a long blog post about different things that you can do to kind of take a deep look into your website and see what you could be doing to improve. So I don't want to kind of rehash the whole blog post. I'd kind of recommend taking a look at that. And if this is something where you're seeing those changes since last year, then that feels like something where it's not just kind of this one-off thing where it's like a little bit that our algorithms are deciding to change things a little bit, but probably things are still in kind of a shape where you might be able to improve them overall. I'm receiving a content injection security message in Search Console for some of my URLs. I checked those pages with every possible aspect, including Google Expert advice. I even watched the latest video regards to security issues and didn't find any issue related. And there are a whole bunch of URLs. So I don't know, I'll pass these URLs on to the team to double check. Usually the content injection message is something that we would show if our systems believe that there's content that is hosted on your website or served through your website, that might not be something that you created yourself. That might be something that maybe a spammer created or a hacker created, and they're hosting it on your website for some reason. Usually this is something where maybe if your website is about cars and then suddenly there's pharmaceutical content in there, then our systems might say, well, this feels like something where we need to be a bit more careful with. In general, these kind of messages apply to individual URLs. So it's not that your whole website would be demoted or shown differently in the search results, it's really just those individual URLs. But I'll pass these on to the team to double check to see if there's something maybe on our side that we're picking up wrong or maybe there is something that we can let you know about. Google is showing our old contact number as a featured snippet. Though we updated a new contact number, it's still showing the old one. Steps we've taken so far, we updated the number and submitted for re-indexing, applied for Google My Business page, which is still pending, removed the old number from the existing pages, and replaced it. So in general, these are the right things to do. If you're seeing things in a featured snippet, that would be due to essentially what we would show as a snippet on a page, which is based on the content itself. So updating the number on the page, removing that page or removing that number from existing pages, making sure that you have the new number in maybe description meta tag, if that's where it was shown, or any structured data that you have there, those are the right steps to take. With regards to updating the search results, sometimes this takes a little bit of time. So it's not always the case that you click the button for submitting to index and it updates it automatically. Sometimes this is a process that takes a little bit longer. So I give it a little bit more time there. If it's really critical that this number is not shown in Search, so for example, if it's a number that doesn't belong to you or a number that doesn't work at all, then what you can also use is the outdated content tool in Search Console, where you can tell us that there's specific content on this one page that has been removed. And you can tell us what that content is that was removed, the URL of that page. And then our tool will take a look at that and either suppress that snippet for the time being until it's been updated, or wait until that search result is such that that old information that you used to have there is no longer in there, and then it would show that again. So that might be something that you could do if this is really a critical issue. Recently, I requested outdated URL removal from Search Console. And I noticed that the site colon domain is still showing the actual number of pages that are live on Google to me after the URL removal. And Search Console indexing status shows more pages even after removals, including previous outdated URLs so the data doesn't match even after the removals. This creates confusion. So I think I'm not exactly sure what you're seeing, but probably what you're seeing is kind of a mismatch between the site colon search results and the data in Search Console specifically for the index coverage report. In general, the site colon number that you see in the search results, where it says of about so many thousand URLs that we know for the site, that number is optimized for speed and not for accuracy. So I would not use a site colon request for any kind of diagnostics purposes. Instead, I would use really the number that is shown in the index coverage report in Search Console. That's based on what we actually have indexed for the site. So that's kind of the one thing. The other part is specifically with regards to the URL removal tool in Search Console. The URL removal tool doesn't remove URLs from indexing. It essentially just hides them from search. So in practice, what can happen here is that the URL might not be shown at all or would not be shown at all in the search results, but would still be counted in the index coverage report for the time being until we've removed that URL completely from our index. So that's something where you might see a mismatch there as well. In particular, if you remove a large part of your URLs using the URL removal tool, maybe a whole subdirectory full or subdomain full, then that's something that could significantly affect the number of URLs shown in Search but wouldn't necessarily affect the number of URLs that we have indexed in Search. So we might have indexed them. We know about them, but we would not show them. If you want those removed from indexing as well, then you need to make sure that there's a noindex on those pages, that there's a 404 status code, all of those usual things that you would do to remove pages from the index. We have multiple location websites. The content of the pages is similar. So Google dropped the .com-slash-fr page and said the domain.com-slash-ca page is canonical. I think we talked about this briefly in the beginning. Skip this one. Google currently limits a number of URLs in a site map to 50,000 URLs. Does that include hreflang, alternate URLs, videos, and images? No, it specifically specifies the primary URL elements there. I don't know. I think the tag is the lock loc component tag that we use in the site map file. Not 100% sure. But essentially, it's the pages that is limited in the site map file. The attributes where you have extra information for those individual pages, like hreflang, alternates, videos, or images, those are not counted in the totals. In practice, they're kind of a part of the totals as well because there's also a limit on the file size of a site map file. So if you have a large number of other URLs that you supply in addition to those pages, then usually the file size would get large enough that it would also be too large and would need to be split into separate site map files. But you can create separate site map files. You can submit them individually. You can submit a site map index file, which points at a number of different site map files automatically. So all of that is possible to really exceed those limits by quite a bit. We've seen some websites being targeted by Blockhead SEO. They have consistent bad links pointed at them daily, and they've also seen some big drops. We're constantly having to block links. Do you still maintain that bad links don't matter? So this is certainly something that our systems work really hard on to make sure that any kind of negative SEO or any kind of irrelevant backlinks, unnatural backlinks, that as much as possible we're able to ignore them completely. Usually the cases where I see that something around negative SEO is happening are the cases where when you would look at them manually, you would say, well, this looks like maybe someone has built these links up over the past. And it's not really a competitor, but maybe an SEO that's been working for the company. So that's something where our web spam team, if they look at that and they see this kind of situation where it's really kind of improbable that a competitor has built up links for a website over the course of the last decade, then that might be something where the web spam team would say, well, maybe you should clean those links up. The second part is we've seen some websites that were hit by Google BERT update. And they still have not recovered. Was that update only about intent? Because it seems like it was about more. So BERT is essentially a way of better understanding text. It's not a ranking change in that sense. It's not an update, kind of algorithm update that suddenly we rank things differently. But it's really about understanding text. So it means that we work hard to understand when people enter queries into the search results, in particular, when these are long queries where we need to understand what is the context here, what is something that people are actually searching for within this query. And when it comes to pages themselves, we try to figure out what are those pages actually about and how do those pages map to those specific queries that we've got. So it's not a ranking change, per se. It's really about understanding the text on a page and the text that people enter in the queries. And from that point of view, it's not that websites get hit by this update. It's really that we're trying to understand what these pages are about. And if these pages are such that it's really hard to understand what they're about, then users will have trouble with them. And our search engine will have also trouble with them. Usually in the cases I've looked at where people say they were hit by this kind of update, it's more that there were just general changes in search over time that also took place. And we make changes in search all the time. So it's not necessarily the case that because Google understands the pages better, we suddenly decided to penalize a set of individual pages because we're trying to understand these pages better, not try to understand what people are doing things wrong. Image search results sometimes show images from a related product section on an e-commerce site with the title of the containing page, another product page. This could lead to results with the title product and an image of product B. Misleading users and creating false information. We saw it also on big sites like Etsy. Is there any way to signal to Google that these images are not describing the page but are describing the target page, similar to how an anchor text describes target pages? Or is there another way to solve this problem? So in general, we do try to understand where the primary images on a page are and to use those for image search primarily. So if you're embedding the main images in a visible way within a page, in particular, if you're also using the structured data to tell us about those images on a page, for example, if you have the article markup or something similar and you can really specify the primary images on a page, then that's something that makes it easier for us with regards to Google Images to understand which images we should be associating with this particular page. Obviously, if you have related images, kind of like thumbnails leading to other pages, and those other pages are not indexable, or the primary image on those other pages is not recognizable, then it can still happen that we would show those thumbnail images associated with that particular landing page as well. So that's something where if we don't have anything better, then we might choose something suboptimal. But if you can specify those images in an optimal way, then it's a lot easier for us to pick those up and to actually use them. Many times, due to website structure, we have long and weird anchor text. Price, number of reviews, review text, ratings are included in the anchor text. Is it OK for SEO, or should we fix it, or should we just make sure that important keywords are present in the initial part of the anchor text? So from our point of view, you can leave your anchor text as natural as you want. And if you're including more information in an anchor text, that's perfectly fine. It's not the case that we kind of count the words in an anchor text and only evaluate each word as a fraction of the total number of words in a piece of anchor text. If you're giving us more context about a specific page on your website through the anchor text, then that's fantastic. So for the most part, I wouldn't recommend doing anything particularly artificial for the anchor text just to make sure that only the keywords that you want to have associated with that page are visible in the anchor text. On one of our mobile domains, I'm seeing that level three depth category pages are not crawlable from level two category pages. We have breadcrumbs also. Will Google be able to make the relationship of level three level two by breadcrumbs present on level three? Or should we make all important subcategory pages crawlable for bots on the parent page? Currently, they're not AHFs, but the user can navigate. So I don't quite understand this question. But in general, if you want to make sure that we can crawl your whole website, then we need to be able to find links across that website, ideally from both directions. So in particular, if you have different category levels where you can go down one category, maybe like a submenu or a different subcategory section on your website, then we really need to be able to find links so that we can go down that path as well. And it's not enough to just find links leading us back up, but we really need to be able to find the way down as well. And ideally, we'd also be able to find some way across those different pages. So between individual items and maybe related items or other items that are within that same category as well. So that's something where I really recommend that you make sure that this kind of structure is crawlable on your website, that it's kind of a reasonable structure, that it's not too deep, that you just have lots and lots of categories, but also not too broad, that you put everything on the same level. So that's finding that balance is sometimes a bit tricky, but making sure that we can crawl in essentially any starting point and reach the rest of your website from there through clicking on links, that's essentially what you should be aiming for. And there are a bunch of third party crawlers that you can use to test your website in that regard. If you're unsure about this, and it sounds like you have some understanding already of what it takes to make a crawlable website, then I would strongly recommend using one of those crawlers to kind of start crawling from one URL and see, does it actually reach the other side of my website and kind of find all of the pages within my website, or does it get stuck somewhere? Maybe there are individual categories that don't have links to the next parts where you're using some kind of other JavaScript-based navigation that doesn't use links. So that's kind of the direction I would head there. Lately, I noticed new pages on my website are not being indexed, and there are no errors related to crawling or other things. The content is unique. It's been around for three weeks since a new page is indexed. What should I do? Is this normal? I have a new website. In the past, pages were indexed relatively quickly, but the process of indexing is getting slower and slower. So we don't promise to index all pages on the web. That's kind of the first thing there. It's not a guarantee that just because you create a page that Google will go off and crawl and index that page, even if you use the kind of inspect URL tool and submit it to indexing, it's possible that, especially with a newer website, if there's a lot of content suddenly on a new website, that our systems might be a bit cautious and say, is this really something we should be spending a lot of resources on to crawl and index all of these pages here on this website that we don't know about yet? Or should we be a little bit more cautious there and wait to see until we have better signals about that website to understand why that website is so absolutely fantastic that we must absolutely go forth and crawl all of the content on that website. So in particular, if this is a new website, working more on quality rather than quantity is something I would recommend. And working to make sure that your website is well embedded with the rest of the web also makes sense. So that's something where maybe it makes sense to promote your website a little bit to kind of get things going on your side, kind of how you would do that with a normal business. If you open a business in a new city somewhere, you don't just open your business and hope that people find you and come to your store and buy something, but rather you do something to kind of get things started. And over time, if you do things right, if you have fantastic content, then you'll have that kind of word-of-mouth effect that people will recommend your content. And over time, that's something that we would pick up for Search as well. I'm looking at a website that's approximately 100,000 pages in size. Approximately two years ago, they made a bunch of fairly significant changes to a website all at once that caused the algorithms a lot of confusion and haven't really recovered since then. They saw a drop in rankings and visibility across search results in general after changing domain name, framework, platform, content, et cetera. OK, that's a lot of changes. One of the key pages seems to rank OK for some of the commercial terms, but they don't appear for all of the other variation. This page doesn't even appear for the term in the top 250 results. Less relevant pages, like blog posts from the original or from their website, rank somewhere in for the term. It seems that the page itself is not a problem as it ranks for other popular pages. This particular keyword is also used in the footer and makes up 98.9% of the anchor text pointing to that page over the other terms. Have you seen other cases like this where pages penalized for a particular keyword because of over-optimization? So I don't know. The 98.9% sounds like a lot, but depending on a website, especially if you're talking about internal links, it is very common that individual pages have a lot of the same internal links. Depending on the CMS that you're using, it's possible, for instance, that you have a product and that product has a name. And it's always linked by exactly that same name. And then all of the internal links to that product use that product name. And that's perfectly fine. So that's something where it's not necessarily automatically the case that we would algorithmically ignore things just because they're used a lot in the anchor text. It's really hard to say specifically what is happening here. I think you sent me some examples. So I'll double-check those together with the team and see if there's something particular that I can pull out. But it can certainly happen that individual pages rank well for some popular queries, but don't rank that well for other popular queries. So that's not necessarily something where I'd say there's automatically something broken that needs to be fixed. I experienced the following situation. I published a new blog post. For the past two weeks, the URL was and still is listed under Discovered, but not indexed, in the coverage report in Search Console. Inspecting the URL, Search Console says the URL is on Google with the correct self-transfer and saying canonical tag. Looking into a search performance report, I discovered that the URL is indexed twice with different UTM parameters from Facebook and Twitter. Is this something that Google will fix by itself because the correct canonical is set up? Or do I have to take any action? I'm using a domain property in Search Console. Should I also add the URL prefix property and use the legacy URL parameter tool? So if you're using a lot of these parameters on your website and you want to simplify the tracking, then it's probably worthwhile to use the URL parameter handling tool for this because with that, you can more clearly specify which URLs you would like to have indexed. From a practical point of view, I don't think this page would be ranking differently in the search results because if we already have one version index, if we have the non-parameter version index as well, we would see that as a duplicate of these parameter versions. And we would essentially just fold those signals into that preferred version. So from a tracking point of view, it's a little bit trickier because you have these UTM parameters. But from a performance point of view, it should be essentially the same. So that's something where I wouldn't see this as a critical issue that you must fix immediately, but rather as something that just makes reporting a lot harder than it needs to be. So if there is a way that you can tell us that you want those parameter names excluded or ignored, then I would go ahead and do that. Otherwise, what should happen here is over time, if we see those URLs with the parameters with the rel canonical pointing to the non-parameter version, we'll pick that up as well and we'll just do that automatically. So if you wanted to give it more time, that's perfectly fine. That should just settle down. If you want to take things more into your own hands and using the URL parameter handling tool is an option there. With the parameter handling tool, it's still not the case that this will kind of immediately fold everything over. But in particular, with new URLs, as we discover them, we'll be able to ignore that parameter essentially from the start. In our old articles, they're photos that host on HTTP servers. Google Chrome says that we have mixed content. Is this really a problem for us? I read that Google Chrome automatically changed HTTP to HTTPS. And if the object is no longer available, then it would be ignored by the browser. So in practice, this is something where when talking with the Chrome team, they obviously don't like to see this. And it is something that is shown in the browser bar that you have mixed content on a particular page. The security effect here is that with mixed content, if the browser is requesting HTTP URLs, then you could be leaking information about that user session over HTTP, which theoretically someone along the way could be picking up or taking advantage of in other ways. So in practice, I'd recommend to try to clean this up. From a ranking point of view, it doesn't really have any effect. It's more a matter of for your users, it's worthwhile to clean this up. Sometimes there are ways that you can search and replace for things like old host names with HTTP and just replace it with the HTTPS version. So in particular, if your content is already accessible on HTTPS, then it might be an option to do it like that, to pull together some kind of SQL query to automatically swap those out, if that's something that is kind of on a database-based site. If these are old HTML pages that are individual files, then maybe there are different things that you'd need to do that. Let's see. I recently tried to QC one of my client's hreflang implementation, huge e-commerce site with more than 10 million pages. I found it extremely hard to QC or examine certain pages because they're using a sitemap hreflang, especially when I try to examine a specific page and I have to go back to the sitemaps to see if there are references. I already found a lot of errors in their hreflang sitemaps too. Should I recommend them to use HTML or HTTP method instead? I heard that HTML hreflang will increase page weight, but those codes are not render blocking, nor do they have any download or send HTTP requests. They're just in plain text, so it shouldn't affect page speed too much. What about HTTP hreflang? How would you deal with a huge e-commerce site hreflang setup? So we support hreflang through the meta tags on a page or through HTTP headers or the sitemap file. And from our point of view, all three of those are equivalent. So if there is a method that makes it easier for you to maintain or to track or implement hreflang, then I would try to use that one. If you're finding it really hard to maintain a particular implementation that you didn't add yourself, then maybe it makes sense to migrate that to an implementation that works better for you. With regards to the HTML version of the hreflang tag, so if you're using it in the header of a page, you're correct. This is essentially just static text. It compresses really well. So for the most part, it doesn't change a lot considering, assuming this is an e-commerce page with a lot of content on a page, it doesn't affect the payload significantly, depending on obviously how many hreflang variations you have. But this kind of content compresses really well, so it shouldn't affect the payload too much. This is something you can double check though. So you can add this kind of hreflang annotations to some test pages, use kind of static testing tools to determine the actual payload size and to determine the effect on theoretical page speed. When it comes to speed, I would recommend taking a look at the new web vitals, which we put together, I think, last week, or fairly recently, which includes some neat metrics that we think are fairly relevant for the web. So it might make sense to double check what those metrics say with these two variations. OK, kind of running low on time. Maybe we'll just switch over to live questions if there's anything still left from your side. Any more questions? Sure. John, can I? OK, just double-clicking on the first question about meta description. During April, we had updated one meta description from one client, adding during COVID-19 outbreak we are working because it's just an online business. And it has nothing to do with health or anything. So just to emphasize, we are working. And we didn't see it on the SERPs. So we updated again. We thought maybe it's a red flag for Google. Let's exchange COVID-19 for April. And still, it doesn't show on the SERPs. Now, with all-go update, it's not showing still. Is there anything wrong with it? Or what would you say about it? Hard to say. But I mean, with the meta description, if you're putting it into the meta description, on the one hand, there are limits to how much we would display from a meta description. So maybe that's quite a lot. Inside the limit. We consider the limit. Yeah, OK. The other thing is, if it looks like something that is kind of spammy, where it's more keyword stuffing related in the meta description, then that might be something where our systems try to determine a meta description to show ourselves. My guess is that it's less the case if you're adding kind of text, additional textual information. And the other thing is it's worthwhile to search in the way that users would search. So not by checking with a site query, but rather by searching for your company name or kind of the common queries that you would do. Because we do change the description that we show in the search results based on the query. And if it's a site query that you're checking, it might be that we're not really sure what we should be showing in the description part. Right. We work on that. Thanks. No. The other thing is also that sometimes this just takes a while to be updated. But if you're talking about April, then it feels like it's more than a couple of weeks already. So that feels like something where we should have updated that. Thanks. Sure. Hi, John. Can you hear me? Yes. Give us a mic. So something I just wanted to share and see if you have any ideas or thoughts. So we have a site where US doctors are answering patients' questions anonymously, just like Quora or Stack Overflow. So we have a several million long library. We recently did an infrastructure update where we had to change some of our URL structures. So we basically did a 301 redirect for many of our pages. And we submitted a new sitemap. Since then, the majority of our pages have been stuck in discovered not indexed. And pretty much like 90-plus percent of our pages. And it's also resulting in some significant traffic drop. So just trying to understand if there's something wrong that we did, or we should have done it in a different way. My guess is it's unrelated. Because with a redirect, what you're really telling us is that we should pick a different canonical URL, which means if we haven't processed the redirect yet, we would keep the old URL in search. If we have processed the redirect, we would use the new URL in search. It wouldn't be the case that the URL would drop out of the search results. So it's not that we would show less in search. It's really just, do we show this URL or do we show that URL? If you're seeing a traffic drop overall, then that feels like something that would be unrelated to this kind of site move. So the old URLs that are not in the new site map are not automatically being removed from the index? Exactly. No. So that sounds to me like it's more a matter of maybe our quality algorithms just not wanting to index as much from this website as before. It's not something where there's a technical issue with that redirect. Got it. OK. Thanks. All right. I think there's someone else that wanted to ask a question in the chat. Hey, John, this is my first time joining the chat. And I totally feel like the new kid that walked into a very, very advanced class. And I know my question is going to be really stupid. So please just bear with me. I don't know SEO. I know the Webmaster guidelines. I've been reading them. But if you had, let's say, for example, a website that sold, I don't know, let's say baseball hats. And they sold baseball hats for, let's just say, every single team. And there was different colors. This is all hypothetical. What on Earth do you put in the meta description besides like Yankee's blue hat, Yankee's red hat? You know what I mean? The meta description is kind of the description of the page. So if it's essentially like a Yankee's blue hat, then you might use something to describe that in a more, I don't know, readable way, which could be like, buy your Yankee's blue hat on our website or add something like free shipping or whatever kind of additional information that you might tell someone. If you were to meet them on the street and they're like, oh, I'd like to buy a hat, then it's like, what would you tell to them? That level information. So the different things across those different products, that would probably be like the product name or some attribute from the product, but the rest can definitely be the same thing. It can be like, buy your Yankee blue hat here or buy your Yankee black hat here, depending on what the individual products are. And you're talking about on the meta description of the individual products. It's shopped by, I don't think that makes a difference, but awesome. Thank you. Thanks for taking the time. Sure. All right. I need to jump off to another meeting, but it's been great having you all here. We have the next one lined up, I think, for Friday. If you all want to join, depending on time zone, maybe that works better or less well. And I'll set up the next meetings as well in two weeks again as well. All right. Thanks a lot for dropping by and hope to see you all again in one of the next Hangouts. Bye, everyone. Bye, John. Bye-bye.