All questions have Show Video links that will fast forward to the appropriate place in the video.
Transcript Of The Office Hours Hangout
Click on any line of text to go to that point in the video
JOHN MUELLER: OK,
welcome, everyone,to today's Google Webmasters
Central office hours hangout.My name is John Mueller.I'm the webmaster trends analyst
here at Google in Switzerland.And part of what we do is talk
with webmasters and publishers,like the ones here
in the Hangout,and the ones that
submitted lot of questions.As always, we have
a chance for thosewho are kind of new to these
Hangouts to ask questions,before we get started
with the submitted ones.Is there anyone here who wants
to go ahead and take a shot?No?OK, well, we'll still have
room for more questionstoward the end.So I'll just go through the
questions that were submitted.I see one of them here has
lots of different parts, whichmakes it kind of hard to read.It usually helps to keep things
really short and to the pointso that I can go through
one of these questionsand figure out what
it's actually about.So let me see where
this one starts.We have multiple brands, and
they're on separate domains,and we want to move them
to, I think, subdomains,but still, of course, rank
number one, like everyone else.Is it a problem to go
to subdomain or for usto be on separate domains?From my point of view, you can
definitely move to subdomains.If that works better
for you, go for it.That's not something
where I'd say you'dbe worse off on subdomains.We do try to understand the
structure of these sites.So if you have one
normal website spread outacross a lot of subdomain,
which we sometimessee if people have wild
card subdomains set up,then we'll probably
recognize that as one site.Whereas if we recognize that
there are multiple sites usingsubdomains, we will recognize
that as multiple sites.So it's not necessarily
something that alwaysgoes one way or the other.Would it be possible
to provide more querydata in Search Console?Right now, a lot of
queries slip throughthe cracks as only the top
1,000 queries are shown, or 999.So the 999 comes from
one query being filtered,where essentially we don't
have enough query informationfor that query.We don't know what we
can really show there.So that's why some
people have 1,000.Some people have 999.Theoretically, lots of
things are possible.But as always, we have to
balance the different thingsto work on when it comes
to features like SearchConsole, Search Analytics.We have to understand is
it worth putting more querydata in here?Or would it be better
to add a new featurefor some other feature or to
add something else really fancythat people have always wanted?So these are the kind of
trade-offs we have to make.And I believe Zineb has
been asking on Twitterfor more information on what
people will do with moredata in Search Analytics.So I'd check out
her Twitter accountand give her feedback
if there is somethingthat you feel you
would do differently,if you had this kind of data.Another option is, of
course, to use the APIand to try to automatically
filter down more informationso that you can get
some of these queriesthat you might be missing.Another idea might be
to separate your siteinto sections.If you have really
logically distinct sectionsof your website that you
want to look at separately,then it makes sense to
just look at a subdirectoryor a subdomain of
your whole site.Cycling query results-- what
order are they displayed in?This is pretty much
no specific order.So it's not that
you can pull outany kind of secret information
by looking at the site queryresults.Usually you will have
something like the home pageas the first one, but
that's not always the case.And that's not necessarily
a sign of a problemif your home page isn't the
first one for site query.So from our point of view, site
query is an artificial query.It's not something where
someone is specificallylooking for anything specific.So the order, the relevance
of the results there,isn't something where we'd say
there is a well-defined orderthat we need to provide.We know HTML sitemaps
can be useful for users,but is there any
SEO value in them?If we're already
doing XML sitemapsand submitting that to Search
Console, any SEO value in HTMLsitemaps?I don't know.Sometimes.Sometimes it can
definitely make senseto have these kind of HTML
sitemaps, which are essentiallya mapping of your category
and your detail pages,especially if we can't crawl
a website normally otherwise.So if you have a really
complicated navigationstructure, maybe if you have
pages that are almost connectedjust through search forms
rather than a logical structure,then at least having one
place where we understandthe structure of the
site, based on the links,that can really help us.
MALE SPEAKER: So if the owner
doesn't want to add the sitemap in Search
Console, so Googlebotwill recognize the HTML
sitemap and go from there?
JOHN MUELLER: Yeah.We use HTML sitemaps to
understand the structureof the site to find the URLs.But there is obviously
a lot more informationin XML sitemaps, things
like the change date,for example, that tell us that
these pages actually changed,which, if you have a larger
website, then lettingus know that this
section of your siteor this specific detail
page changed recently,it makes it a lot easier
for us to crawl thereand to actually update our
index, based on that change.So that's something you wouldn't
be able to do with an HTMLsitemap.It's almost like an
HTML sitemap couldbe comparable to a
normal site's navigation.And if you already have
a good site navigation,then you probably don't
need an HTML site mapfor search engines, at least.
MALE SPEAKER: But Bot is still
amazing at recognizing siteswithout the HTML page
like sitemaps as well.
JOHN MUELLER: Sure, yeah.If you have a normal
site structure,a normal navigation within your
website, that's perfectly fine.We can find all of those pages.If you already have a
website that you knowis bad with the navigation,
then an HTML sitemapcan a help a little bit.But on the other
hand, if you knowthat your website is
bad with the navigation,then you might as well
just fix that, too,to tackle the
problem at the root.All right, it's a common rule
to just use one h1 tag per page.Is it a problem for Google to
use multiple h2, or h3, tagson a page?That's perfectly fine.That's definitely not a
problem from our point of view.You could also, if you wanted
to, have multiple h1 tags.And we'd probably be able
to deal with that, as well.So that's something
where the headings helpus to better understand
the structure of the page,the context of the text,
and the images on the page.So if you can structure
your page in a waythat we can understand that
this block belongs togetherand this block belongs
together, then that helps us.And it's not
necessarily the casethat you have to limit
yourself to one h1or always have this clean
structure of h1, h2, h3.Some pages just have
h3 tags on there,and that helps us to understand
the structure as well.
MALE SPEAKER: Yes, so, John?
JOHN MUELLER: Yes?Oh, you're muted.
MALE SPEAKER: Sorry, sorry.So John, it means keywords
being in the heading tagor in simple content
doesn't make any differenceto search engine?The content should be good.
JOHN MUELLER: Well, obviously
having some informationwhat the sections of the page
are about with the keywords,for example, that helps us.So it's not the
case that you wouldwant to just put an
image in the headingand not tell us anything about
what the section is about.On the other hand, just
stuffing keywords in a headingdoesn't really help
that much either.Or marking up the whole page
and saying the whole pageis actually one big
h1 tag and assumingthat Google give this
page more weight, that'snot going to happen.
MALE SPEAKER: OK.
JOHN MUELLER: These are things
that everyone has tried out,probably.
MALE SPEAKER: So
call the classic SEO.
JOHN MUELLER: Yeah.
KREASON GOVENDER: I just have a
follow-up question then, John.
JOHN MUELLER: OK.
KREASON GOVENDER: Does it
matter at which point do you putthe most important keywords?So should your most relevant
keyword be on the h1 rightat the top of the page, or can
it be lower down on the pageand have a less relevant
[INAUDIBLE] higher up?Does it make a difference what
order it's in on the page?
JOHN MUELLER: We
essentially wantto understand what
this page is about.And the clearer you can make
that for us, the easier itis for us to pick up
and, of course also,the easier it is for
users to understandthat this is the page that
they're actually looking for.So I would definitely not put
the most important keywordsor terms that you're trying
to rate for somewhere wayon the bottom in the footer
in small font somewhere.But rather, make it really
clear that this pageis about this topic.And yo probably have
that in the headings.You might have
that in the title.You might have that
in multiple places.
OK, thanks, John.
MIHAI APERGHIS: By
the way, I noticedthat you can use
anchors on the headings.And then you can link them.And Google will pick that up.So it will show them
in the search results,under the results,
under the resultitself, under the
metadescription,it will show the anchors that
you marked up, or the headings.
JOHN MUELLER: Sometimes, yeah.Sometimes we do that.
MIHAI APERGHIS: And that can
help with click-through rates.So that's really strong.
JOHN MUELLER: Yeah,
if it's a longer page,and we can understand
the structure properly,and you have
these-- I don't knowwhat they're called-- jump marks
or anchors within the page,where you can go directly to
a specific part of the page,sometimes we do show that as a
type of almost like side linksbelow the page.OK, here is a question
about redirects.Everybody loves
redirects, 301 or 302.We should change our
opinion from time to time,just to throw people off.Or maybe not.Redirect chain-- a 302 to a
301 to a 200, or a 301 to a 200directly-- which one passes the
most link juice or page rank?
MALE SPEAKER: Page rank.
JOHN MUELLER: From
our point of view,they essentially
do the same thing.So you're forwarding users
from one page to another.And depending on which of these
pages we keep in the index,that'll be the one that
collects all of these signals.So if we decide to index
the destination page,then that one will
be the one thatcollects the page rank and the
other signals from that page.If we decide to index
the redirecting page,then that will be the one
that gets all of these.So it's not the case that
a 302 kills any page rankor that you need to do a
301, and if you combine a 301with a 302, then you get 50%.Essentially we try to pick
out which of these URLswe want to keep indexed,
nd that's the one that'sgoing to get signals.
MIHAI APERGHIS: John, is it
the case, though, with 302'sthat maybe you wait
a bit more to be surewhether you should
treat it as a 301?
JOHN MUELLER: Probably.So essentially, that's
a hard question.That's a really hard
problem, deciding which URLto actually keep indexed.So it's essentially the
question of canonicalization.You have two URLs.There is some signals between
those URLs that you know about.And the question is,
which one of theseshould you show in
the search results?Which one should be
the canonical URL?And if you have a
redirect, then thathelps us to understand the
situation a little bit better.With a 302, you're saying
that the redirectingURL is the one that you prefer.And a 301, you're saying
the destination oneis the one that you prefer.So a 302 is like saying,
well, this is one better.And a 301 would be saying
that one is better.But we don't look at
just the redirects.We look at other things as well.As if there's a canonical
tag on this page,how the internal links
within the websitepoint at these
pages, what you havein your sitemap file,
external links as well,all of these things add up.And they say, well, overall,
the signals point at this URL,or overall the signals
point at that URL.So it's not so much a
case that you leave a 302in place for a while and then
it magically turns into a 301.It's more the case that over
time, these signals change.And if everyone starts linking
to the destination page,but there's a 302
redirect, then probably weshould be indexing the
destination page instead,because that's the
everyone is using.So it's really a hard problem,
I'd say, with search enginesto pick the right canonical
URL, and this is somethingthat we have teams that
are still working on this.So it's not a solved problem,
even though we've beenat it for such a long time.
MIHAI APERGHIS: OK,
so it's the casethat the more signals
you've had that confirma certain scenario,
the better are youat understanding whether
you should treat itlike this or that?
JOHN MUELLER: Exactly.That's why we say you
should be consistent.Because if you have
a redirect in placeand you tell us a little bit
that maybe you prefer this onebut some of your site
says, prefer the other one,then we're torn.And we don't really know.And it could go either way.But if you're really
consistent, and say everythingshould point at this URL.I have internally all
my links at this URL.There's a canonical setup clean.The redirects are clean.The sitemap is clean.Everything is
pointing at this URL.Then we'll say, well,
it's no question.The webmaster really,
really wants this one.So we'll pick that one
for the index instead.
MALE SPEAKER: Because
AMP is live now,are you going to make
that a ranking signal?
JOHN MUELLER: AMP.AMP a ranking signal--
MALE SPEAKER: Is
JOHN MUELLER: At the moment,
it's not a ranking signal.So it's obviously one
way to make reallygood mobile friendly pages.So that might be an
option, where I've alreadyseen some sites that have moved
their whole website to the AMPformat.And obviously, that's a
mobile friendly setup.So that gets at
mobile-friendly boost.But just AMP itself
is not somethingthat we have a ranking
signal at the moment.
MALE SPEAKER: Like they
also updated their plug-inin [INAUDIBLE].So it's pretty easy.
JOHN MUELLER: Yeah.I even put it on my blog.Now I just need to generate
some newsworthy contentand hide all of those
MALE SPEAKER: And put it
in live on Google News.
JOHN MUELLER: I don't know if
I could get it in Google News.I probably wouldn't
pass the criteria.Like posting once every
eight years is kind of--I don't know not a
perfect news source.
ROBB YOUNG: You could
answer my question.That might be newsworthy.
JOHN MUELLER: Your question.Oh, your general question.I don't know.I don't know.Now it's getting tricky.Let's continue with the
questions that were submitted.If I have a website-- page.com--
and the content in English,and I want to create
the same in Spanish,is it better to create a
separate page, like page.es,or include a translator
on the original page,and redirect it to .com?So in general, for
multilingual pages,we recommend making the
language clear per URL.So you have one page in
English, one page in Spanish.And the whole page
is in Spanish,and the whole page
is in English.And with that, it's
a lot easier for usto recognize this
is the English page,and we should redirect
people to the English page.And we don't have this
situation where the page ispart English, part Spanish.And we don't really know
which language this pageshould be ranking for.So from the general
question here--should I make separate URLs
or to use the same URL--I would definitely
use separate URLs.You don't have to put it on
a separate domain, though.When it comes to
language, essentially youcan use any unique
URL, per language.That could be with a
parameter at the end.That could be a separate
subdirectory or a subdomain.But it doesn't have to provide
any structural informationfor a specific language.When it comes to
geotargeting for countries,there we do need to understand
the structure of the site,and say, this whole part of the
website is for that country.But for languages, you can use
whatever structure you want.In Search Console,
a website ranksposition one for most keywords.We discovered these positions
reflect the presence on the mapabove the organic results.But it there a way to see the
organic listings rather thanthe map listings,
in Search Console?I don't think we differentiate
that within Search Console.So we would essentially look at
the normal web search resultsthat we'd show if
you just go to Googleand search for something.And if your site ranks in one
of those blocks on top, whichmight be-- what is it called--
the featured snippet on top,or with a map entry,
or in a news entry,or in the app
carousel, then that'swhat we count as the ranking.So where the rest of your site
ranks within the organic searchresults is not something that
we take into account there.And specifically
for Search Console,we use the average
position of the top mostresult for your website.So if your website ranks number
one, number seven, and numbernine, for example, in
one search results page,then we take that
number one rankingand use that for Search Console.We wouldn't reflect the lower
rankings of the other pagesthere.What you can do, however, is
look at it on a per URL basis.And that does give you a little
bit more information there.So for your home
page, obviously, itwill be tricky in
this situation.Because your home
page will be the onethat's probably connected
to your map entry.For other URLs, that we are also
ranking in the search results,you can, however,
see this specific URLis ranking in this position
for these keywords.So that's one way to get more
detailed information there.The important part here
is also to keep in mindthat when you're
looking at it per query,we aggregate that
across your site.Like I mentioned, the
topmost result from your sitewe use that.And when you're looking at it
per URL, we look at it per URL.So the numbers there won't
necessarily add up completely.
ROBB YOUNG: And John,
are they supposedto roughly agree from
Analytics to Search Console?Because you remember the thing
that Mihai and I both mentionedon Tuesday about our
www dot appearingas one of the queries?In Analytics, over
the last 90 days,that has something
like 4,000 impressions.In Console, it has
JOHN MUELLER: OK.
MALE SPEAKER: [INAUDIBLE].
JOHN MUELLER: Yeah.Usually I'd expect trends,
at least, to line up, but--
ROBB YOUNG: I added that that
in there to the chat as well,so you could see it again,
so Mihai could grab it again.We're just trying to get to
the bottom of whether someoneis genuinely searching
for that URL,or whether it's something
they're just typing inand that somehow their browser
or toolbar or somethingis triggering a
search, and it's not--like there's so many searches
for a www dot and no clicks,that it's just weird.
Yeah, I don't know.I'll double check what you
posted in chat afterwards.
ROBB YOUNG: But in general,
they should roughlymatch where possible?
JOHN MUELLER: They won't
be exactly the same,because of the way Search
Console counts, and the wayAnalytics counts. [INAUDIBLE].So you shouldn't need
like 400 versus 4,000,that you mentioned.
ROBB YOUNG: Right.Yeah, there's a lot
that are similar, justfor the two named experience,
say, one is 2,500.One is 5,000.They're way off.Now that's not to say
that our is the bestexample to use for
results that everyone elsecan pin their hopes on.But I'm assuming whatever
problems we have,we could at least still
measure the results accurately.
JOHN MUELLER: Yeah, yeah.
ROBB YOUNG: So as to know
whether they're good or bad.
JOHN MUELLER: Yeah, exactly.
ROBB YOUNG: OK.Well, I've posted
the image again.
JOHN MUELLER: I'll take
a look at that again.OK.All right, if a project is,
for example a sports eventsite with events, which
naturally expire-- for example,a cricket match-- would you add
the expires after a header tagand then set the response
to 410 after the expirationtime and date?Probably.If you're sure that these are
URLs that wouldn't be relevantafterwards, then
you can definitelyuse the expires after meta tag
and use the 410 or the 400.404 would work
just as well to letus know that this is
not relevant anymoreand that we can drop it
out of the search results.Of course, some types of events
you might want to keep longer.So it's not
necessarily somethingwhere I'd say any event
that has already taken placeshould be deleted
from the index.It really depends
on the site itself.For some sites, it
definitely makes senseto weed out all of the
old stuff so that youcan focus on the new ones.For other types of sites,
maybe the old eventsare really the ones that
provide value for your website.Does a site-wide
internal link gettreated any different from
a footer site-wide link?And would the
amount of page rankthat's passed be less
than having the linkon, say, the best pages only?It's a tricky question, I guess.Within a normal
website, I wouldn'tworry about where you
link to your pages.And essentially we try to figure
out the structure of the siteand understand how these
pages belong togetheror what the context is
of these individual linksand how we can understand
one page a little bitdifferently from the other page.If you're talking about linking
from one site to another site,that sounds an awful lot
like a natural links,where you're optimizing
the position of the linkto your site on
someone else's site.And that's really something
I'd recommend not going downthe path of.So again, if this is just
internal, within your website,do whatever makes
sense for the usersso that we can still find
the link to that page,and we can try to
understand the context.If you're placing links
on other people's sites,then I'd try to avoid that.In 30,000 unique URLs on a
site, if the average crawl rateis 5,000 per day by Google,
is that a good or bad sign?That can be perfectly fine.So what's worth keeping
in mind with the crawlrate of a website is
we're not crawlingrandom URLs from the website,
like random 5,000 URLsor lining all the URLs up and
going through those 5,000.We're crawling some pages
a lot more frequently,and other pages a little
bit less frequently.So within those
5,000, you'll probablyhave a lot of pages that
get crawled maybe dailyand a bunch of pages that get
crawled maybe once a week,and then a bunch of other
pages that might getcrawled once a month or
even less frequently.So it's a mix there.You can't say, well,
5,000 a day, 30,000 total,therefore it'll
take-- I don't know--six days for the whole
website to be crawled.That's not how it works.But 5,000 a day for a
website of that sizeseems like a good thing.Another thing maybe to keep
in mind with the crawl rateis crawling more
doesn't necessarilymean that your website will
be seen as more relevantor rank higher.So there is no need to
artificially push a highercrawl rate if we're
already picking upall of the content
that you're providing.Will AMP affect
the search results?Should I start implementing
it now or wait for a fewmore months and look forward
to Google's further guidanceas there are not many proper
AMP plug-ins available yet.So there is an official
AMP plug-in for WordPressavailable.I'd definitely take
a look at that.It seems to be
working fairly well.We do show AMP being kind of
a news carousel in the searchresults, on mobile.So that's one place
it's definitely visible.I suspect it'll become
more visible over time.I know the AMP people are
extremely fast and extremelypassionate about
improving the web.So that's not something
that I'd expect to go away.Also, I expect other
sites also to implementthe kind of AMP furor.It's open source.Anyone can essentially
implement it on their site.I think Twitter, and Pinterest,
and maybe some othershave said that they're looking
into doing this as well.So it's not just for search.It's not just for news.It's something that
maybe it makes senseto dig into a little bit more
regardless of your website.And if you work with
multiple websites,if you're consulting for
them, if you're an SEO agency,I'd definitely
understand how AMPworks so that you can inform
your clients a little bitbetter.And when they come to you and
say should I be doing AMP?You can say, well,
you should do it now.Or you should keep
it in mind for later.Or at the moment, this
doesn't seem to make sensewith your specific website.
MALE SPEAKER: But what if the
CDN is really strong, though?Because there are
certain CDNs thathave servers all
around the world,one of them being a very popular
one that everybody knows thatstarts with a C.
JOHN MUELLER: C?I don't know which
one starts with a C.[?
MALE SPEAKER: CloudFlare. ?]
MALE SPEAKER: OK, you said it.
JOHN MUELLER: Well, it's fine to
mention names as long as you'renot calling them out.
MALE SPEAKER: [INAUDIBLE] your
own club thinks so it doesn't--
JOHN MUELLER: Oh, yeah.The neat thing about AMP is
that it's not just a CDN.It's actually a way of letting
sites integrate your contentdirectly so that your
content gets a little bitmore visibility.
MALE SPEAKER: So let's say
if I'm searching from India,and my website is located in
the UK, yeah, it's a bit faster.Because the actual
library's located in India.So this way, it serves
the user the libraryfrom the actual location, right?
JOHN MUELLER: Yeah.With AMP, it basically uses--
so within Google search results,it uses Google's
cache for AMP pages.So you essentially have
Google's infrastructurebehind your website should
it become really popular.And I think the
interesting part with AMPis really the combination of
all of these different factorsin that by being able to
understand your pages betterby knowing that it is restricted
to a very well-defined format,we can integrate those
directly in the search results.And people can go, really
quickly, to your pages,to your content within
the search results.So that's something where, even
if you have a really fast sitenow, if you already
use a CDN, if you havea great mobile
friendly version, thensometimes it does
still make senseto create these AMP
pages so that you can domore with your content as well.
MALE SPEAKER: So now
[INAUDIBLE] shouldbe like in the
JOHN MUELLER: Well, you
could test the AMP pageswith-- what is it--
webpagetest.org,to see how they work.I think the Google
AMP cache URLs allhave the same URL structure
with the hosting in frontand then the website
URL afterwards.So that's something you
could compare pages and seewhat happens.I think that might make
an interesting blog post.I suspect the team
is still goingto be refining the whole
setup there since it justgot launched two days ago.So I wouldn't expect
the current statusto remain one-to-one
in place forever.And if you find issues where
there are really speed problemsor where there's a big speed
difference with the waythat we host it and the way
other people might host it,then that's definitely worth
looking into on our side.
MALE SPEAKER: All
right, sounds good.
JOHN MUELLER: Sounds like
an interesting blog postto do to compare different
website setups with the AMPpages.
MALE SPEAKER: What
do you think--[INTERPOSING VOICES]
MALE SPEAKER: I normally
use a site like Pingdom.What do you think of that?Or do you trust more--
JOHN MUELLER: I don't know
specifically what Pingdom does.But if you're just
pinging the host,then that probably doesn't
give you much insight on howit actually serves content.
MALE SPEAKER: No,
they do website tests.
JOHN MUELLER: OK.Well, if they check like the
whole web page, then sure.Why not?When are you going
to add more businesstypes to schema.org markup?Currently we have to use local
business markup as opposedto tool rental, which
is our business.So I think schema.org is
a public organization thathandles all of this markup.And I'd double check where
they discuss proposals and makesure to get your proposal
in there as well.And if there is
something that youthink makes sense for
your specific businessthat would be valuable within
schema.org, by all meansgo reach out to them and see
how they pick up your proposal.We have rel=publisher tag
on every page on our site.I've heard you don't
use this anymore.Should we remove it?Does it cause any harm?I don't actually know
if we still use that.Good question.It's essentially meant
to link your Google+ pageto your website.And I don't know what Google+
is doing with that specificmarkup.So from a web search point
of view, that's not somethingwe would use directly.Because it's more
for the Google+ side.So from my point
of view, you cando whatever you want with that.I am sure it won't cause
any problems in web search,because we essentially
don't use it for web search.Whether or not it's still
important for the Google+ sideis something you probably
want to check up with them.Why, if we choose a 302
redirect for our home pagewith a different local
version, sometimes the URLshown by Google is not the
original URL but ratherthe local version instead?That can happen if
we don't understandthe relationship between
the different URLs properly.So with the href flag
markup, you can definitelylet us know about that
kind of collection of URLs,where what you would do
is, say, the home pageis an [? xdefault ?]
file page, whichmeans it is a default
version that redirectsbased on the user's location.And the individual
local versionswould be indexed separately
with their individual local hrefflag tags.So you'd have this cluster
of the local URLs and the one[? xdefault ?] URL.And if you link those
properly, with href flag tags,then that's something we should
be able to understand and showa little bit better
in the search results.Now, does every folder
in a URL need to exist?And if it doesn't, will this
impact the flow of SEO equityacross a website, for example,
slash hats, slash cats hats,and slash hats alone
just returns a 404?So I don't know what kind
of hats cats would wear.That seems problematic
on its own.But from our point of view,
URLs are just identifiers.We don't require
that something thatlooks like a folder is
actually treated like a folder.So if you have one URL
that's really long,if the shorter version of
that URL doesn't exist,then that's perfectly fine.That's not something
that our systemswould count against you.That's what happens on
the web, in practice.And we have to deal
with that, too.
ROBB YOUNG: John,
so that reminds meabout one of the arguments
between a flat hierarchyversus a proper hierarchy,
within the URL structure.And one of the arguments
for a proper hierarchyis that if you consider one
of those files the URL, say,the category page URL in this
scenario with the hats, that'sconsidered to be like
a chapter of a book.And if you open that chapter,
it's got loads of other thingsinside it.And therefore, that folder
contains lots of content.So it's nice and important.Whereas, if you open it,
and it's essentially empty,then it's not.But are you saying
that's not the case?As a general resource,
if you open somethingand that's full
of lots of contentand links or subfolders
for other things,that looks like a genuine
resource for content.But if you open
it and it's empty,then anything underneath that
is also essentially useless.
JOHN MUELLER: I don't
think we do that.I think what we
usually do is justlook at the links
within the website.And if there are no links to
that specific subdirectory,then we probably
wouldn't even look at it.So I think, in the past,
when users were savvy enoughto understand what a
URL was, and to manuallytweak the URL in the browser,
that probably made a little bitmore sense, in that
they'd say, oh, I justwant to look at all the hats,
and then I'll change the URLand look at that.But nowadays, URLs aren't
even that visible in browsersanymore.In mobile phones,
you really haveto work hard to even
look at the URL.So it's not something that
I think users would do.And I don't think it's
something that search engineswould do at all.
ROBB YOUNG: No, but
users might expectto see a breadcrumb,
that would mirrora URL structure somewhat.
JOHN MUELLER: I don't
think a lot people would--[INTERPOSING VOICES]
ROBB YOUNG: If you land
on the cats hats page,you might think,
oh, actually, I'lllook up at all of the hats.And if you can't then--
JOHN MUELLER: But you could do
that with normal breadcrumbson the page, too.That's probably the
easier way to handle that.I don't think a lot of people
would navigate a site by URLs.
ROBB YOUNG: OK.
JOHN MUELLER: The one
exception that I know ofis the home page.So if your site's home page
disappears, it goes 404,or it has a parking
page that's justblocked from being
accessible, thenoftentimes we'll assume that
the whole website actuallyis broken.So that's the one
exception that I know of.But that's also the
case where we're alwayslinking to the home page,
and you're explicitly tellingus to look at this page.Whereas with a specific
folder within a website,I don't think we'd even care.Even if you linked
to that folderand it then turned out
to be a 404, that's life.We're not going to
say everything belowthat folder is broke.
ROBB YOUNG: OK.
JOHN MUELLER: All right, someone
has a quick question to ask.I'll let you go ahead.Michael?
MICHAEL: You have to
remember those keyboardshortcuts for unmuting--[PHONE RINGS]
MICHAEL: Let me
just cancel that.
JOHN MUELLER: Oh, no.Now he's frozen.That was probably the
wrong button to press.
ROBB YOUNG: Michael did ask a
question on the actual page,on the Hangout page.
JOHN MUELLER: OK.
ROBB YOUNG: It's on
there if you want to--
JOHN MUELLER: Whoa,
lots of questions.That's crazy.
ROBB YOUNG: I can paste it
into the chat if it helps.
JOHN MUELLER: OK.
ROBB YOUNG: Here.
JOHN MUELLER: Let me
take a look at the chat.Whoa, OK.We have a client that would like
to register a number of domainsfor various languages.The website will presented
in, rather than a 301,the primary domain
slash language.We would like to
mask the domains.However, are there any
caveats to be aware of,for example, required canonical.The intention would be that
the domain.es would redirectto domain.com/es, while the
other domain would redirectto /co.uk, for example.From my point of view, you
can definitely do that.So if you're moving from
[? CCTOV ?] structureto a generic [INAUDIBLE] domain
structure with subdirectories,you can definitely do that.I'd just make sure that
you use geotargetingwithin Search Console for those
individual subdirectories,so that we still have that
that geotargeting informationfor them separately.I'd also expect that this is
the kind of situation that'sa little bit harder
for us to process,because you're essentially
combining different sitesinto one big site.And that means we have to
recalculate all of the signalsthat we get for those
individual pagesand understand the new
site structure instead.So this is not
something where I'dsay it's like a trivial
domain move from onedomain to another,
but rather you'recombining multiple
sites, and we reallyhave to re-understand
everything first.So I would expect that you need
to be a bit patient for thingsto really settle down there.
ROBB YOUNG: And it
looks like they'retalking about getting
the country-level domain,just buying the domains
and redirecting themto the dot com slash country.
JOHN MUELLER: OK, so basically--
ROBB YOUNG: If I
read write it right.But then it looks like
if that's the case,then none of the header
tags would be read anyway.So none of the--
JOHN MUELLER: From
my point of view,if you're using these
for marketing purposes,like in offline ads or
for other types of ads,you could just do whatever
you want with that.That's not something that
would really bother usbecause we'd focus on the URLs
that actually have the contentand use those instead.
ROBB YOUNG: But
you wouldn't readany of the info, the
tags, canonicals,because the 301
would kick in before.
JOHN MUELLER: Exactly, exactly.So we wouldn't take
the country informationfrom that URL,
that's redirecting.We'd expect the country
information to be on the URLthat we actually index.All right, let me run through
two or three more questions.And then we can open things up.Would the content
within a blog postbe used in rankings when
that blog post is linkedto an internal page-- I mean
the rankings of the page it'sbeing linked to?Essentially, we'd look at
these pages on a per-page basisand use the content there.What we would take into
account from a linkis maybe the anchor text.So if within the
link to another pageyou give us some context
of that other page,then that's something
we'd take into account.Does Google give added weight to
blog posts which have a pictureto ones which don't?And how descriptive should
the file names be to help?No, I don't think
we give added weightto web pages that have
pictures, compared to pageswhich don't have pictures.Of course, if you have
pictures, then that'ssomething we could pick
up for image search.But that's separate from
the whole web search side.If the top three results in
e-commerce have AMP pages,then how will search engines
show them in search results?I know news websites, but what
about e-commerce websites,which use Schema for
e-commerce websites?So at the moment, we
only show news contentwithin the AMP carousel.So we wouldn't show your
e-commerce landing pages there.But of course, an
e-commerce sitecan also have a blog, can
also have a news section.And that content could be
shown within the AMP carousel.So that's something that might
give you some ideas on whatyou could do there, too.
MALE SPEAKER: Actually,
John, recently justyou said that if other website
also want, they can go for AMP.
JOHN MUELLER: Sure.
MALE SPEAKER: So I just
wanted to understandif any e-commerce website
goes for AMP and the top threeor four results are
AMP pages, so howGoogle has a plan to show them
that these are the AMP pages?
JOHN MUELLER: I don't know.We'll see when we get there.I suspect there are lots
of other types of contentthat will pick up
first for AMP and showthat in the search results.But I wouldn't be
surprised if, over time,we also got to
e-commerce content,and we're able to show
a little bit betteras AMP in the search results.But I'm not aware of any
specific plans there.
MALE SPEAKER: But for now,
it is good for news websitesand all informative kind
of websites, like blogs?
JOHN MUELLER: Yeah.
MALE SPEAKER: OK.
John, you mentionedif any web e-commerce website
has a blog or a resourcessection or something like
that, you could use AMP, true?
JOHN MUELLER: Sure, yeah.
So don't you needto be in Google News
to be able to be shown?
JOHN MUELLER: No, it
is purely for the web.Of course, some new sites are
also in the Google News partas well.But any website can be
shown in the AMP carousel.And that's similar to
the previous In The Newsblog that we used to have
on the search results,where any kind of newsy
content could be shown there.
OK, good to know.
JOHN MUELLER: Now that Google
has removed the right-hand PPCads, any insight on what will
be put in all the white space?
MALE SPEAKER: My commercials.My banners.
JOHN MUELLER: I
suggested cat photos.But I don't think they
took me seriously.
MALE SPEAKER: What
about YouTube videos?
JOHN MUELLER: YouTube
videos, well, sure.I don't know.So I don't know what the
specific plans are there.I know it's always
hard to figure outwhat the combination of the
elements in the search resultspage should be.And I know that
this is somethingthat they work really
hard on, doing experimentsto try to figure out how
it actually works out,what works out, how
we can make surethat the search results
stay relevant for users.
MALE SPEAKER: But this
is the longest testyou guys have done.Six years is a long time.You started in 2010.And it's not just overnight.
JOHN MUELLER: Well,
things change.I'm happy when I see
people do crazy experimentsand crazy changes.Because that tells
me that they'restill passionate about this and
still trying to figure out whatthe right combination is.And sometimes the
right combinationchanges, over time,
where we see usersgot used to this for a long
time and worked really well.But actually, the web has moved
on, or users have moved on.And now everyone is browsing
the internet on their watch.And we have to figure out
how to remain relevant there.It's just the same
as all of you.It's not like your
website works once,and you leave it
alone for 20 years.You have to keep working
on it to make surethat you don't lose touch
with your user base.
MALE SPEAKER: Absolutely.
ROBB YOUNG: This also isn't
a universal change, is it?There's right-hand side ads,
still, for a lot of things.
JOHN MUELLER: I
have no idea about--
MALE SPEAKER: No, it's gone
pretty much for everything.But there's still
some cache left.They said that they
were tweaking that.
JOHN MUELLER: I have no
idea what the ad side does.We try to separate
that fairly well.
MALE SPEAKER: John, is
there a ranking issue,if you have issues
with JSON-LDs?Like if some JSON-LDs don't
work on certain pages,[INAUDIBLE] a problem?
JOHN MUELLER: If you're using
JSON-LD to provide structuredata, then we wouldn't be able
to pick up that structure data.So that aspect there-- and
if that structure data wasimportant your page, if we're
using that for rich snippets,for example, that were visible
in the search results, thenobviously we wouldn't be able
to show those rich snippets.But rich snippets
aren't a ranking factor.So it would just affect how your
site is shown in search, notin which position.
MALE SPEAKER: OK, thanks.
JOHN MUELLER: Let me just grab
two more questions here maybe.Link rel=canonical to
HTTPS page with colon 443.I guess it's a port thing.Is that bad for SEO,
or is that fine?So 443 is the default
port for HTTPS,which means we treat it
as equivalent to the URL,without the port number.And it would be the same.So that's not
necessarily a bad thing.From my point of
view, it's probablycleaner without the port.It's less likely to
make a mistake there.But it's not going to
cause any problems.And similarly, for HTTP,
the default port is 80.So if you have colon 80
at the end of the URL,then it's really the same thing
as just without that colon 80.Did you change the way
you read robots.txtand interpret the wild card
and the disallow option?We didn't actually change this.I think we updated our
documentation a while back,just to reflect what
the current state was .It was like that for a long
time, where essentiallythe change is that
you have to havea slash as the first character.And after a disallow
directive, if youwere specifying a path
in the robots.txt file,it can't start with an
asterisk, or with a dot,or anything else.It needs to start with a slash.And we also added that
to Search Console,to the Robots.txt
Testing Tool, whichis why some people are seeing
this flagged in Search Consoleas well.And essentially we were trying
to treat it the right way,anyway.But we want to flag
that in Search Consoleso that people are aware
that, theoretically, thisis the wrong way to
submit that line.OK, let's open it up for
questions from you all.What else do we have left?
$350,000 sponsorshipthat you guys did
for Let's Encrypt.But just forget the buying
links and all that thing.No, honestly, this
is without criticism.I just wanted to find
out so you attend,let's say, a half a million
dollar sponsorship event.And you didn't place that link.You guys never placed that link.And it's a natural thing.You guys are attending it.It's not done intentionally.Just oh, we got a link.So is that still
against your guidelines?Because in your
guidelines, if youread it, and let even
a lawyer analyze that,you never placed that link.They just added you, because
you're a part of the sponsors.They have a page that's
dedicated to [INAUDIBLE].So how is--
JOHN MUELLER: I'd say it's
kind of a weird situation whenyou look at it from our
side because people mightassume that we were
supporting Let's Encrypt, just[INAUDIBLE].And that's definitely
not the case.So essentially, what
the webspam teamdoes when they look at
things like this is theytry to understand the intent
of these individual links,and they try to understand
how important are theywithin the general site itself.Is this a very common thing?Or is this something
that just happenedto happen for a couple things?So if we see that
overall, this websiteis well supported,
naturally, externally,and there are some
problematic links,then that's going to cause any
problems from the webspam side.And that's the same for
websites on our side asfor any random other
website externally.So if we see these kind
of sponsorship links,and we can recognize that
the intent isn't reallyto manipulate the
search results,that's not something that's
going to cause any problems.And I think, in some cases,
what the webspam teammight do is just
say, OK, we will justdiscount these links
in the back endand treat them like
disavowed linksso that they don't
pass any page rank.But we understand that there
is no harm meant here that'snot with the intent
to cause any problems.So we'll treat that fine.This is the same as if
you're a local business,and you're active in local
charities, or local events--
MALE SPEAKER: Right, exactly.
--then they're goingto put links on their website.And you're not going to have
any control over what's--
MALE SPEAKER: Yeah, you don't
have any control of thatbecause it's a
part of the event.I mean, they just want to
put you in the spotlight.
JOHN MUELLER: Yeah, these are
things we have to deal with.This is like on
the web we're notgoing to go out and penalize
all of these local charitiesand say, oh, you let
people buy these linksand tackle it from
that side because thatwould be the wrong approach.On the other hand, if we
recognize that individual sitesare going out and
saying, oh, well,I can donate $5 this charity.And then I can get this
linked with free casinopoker, or whatever,
to my website,and we see that this is all of
the links from that website,then we might assume that
actually these donationsthat they're doing are not
just to support that charity,but maybe there's
something more behind it.So we try to look into
it from the contextof the whole website
and to understand what'sactually happening there.
ROBB YOUNG: Are half million
dollar paid links a big issue?Are there many of
those out there?
JOHN MUELLER: I don't know.I'm really happy that we're
supporting Let's Encrypt.And I noticed they changed
the link to No Follow for us.So it's something
where we try to holdourselves to have a higher bar.But at the same time, looking
at the overall picture,this wouldn't change the ranking
of the Google Chrome home pageanyway.So it's not something where
I'd say the webspam team wouldbe sent out and
say, oh, you needto penalize this website for
this one individual link here.Or maybe there are five.I don't know.I'm sure the Chrome
team supports lotsof other good causes as well.But that's something
we do for any website.It's not the case that we
treat our websites differently.If anything, we'd
hold our websites upto a higher standard.All right, one last question
before I have to run off.
MALE SPEAKER: Yeah, John.Can you confirm us
if Google has takenany action against
reviews in search schema,in search result page?
JOHN MUELLER: Reviews--
MALE SPEAKER: [INAUDIBLE].
JOHN MUELLER: Yeah,
this has come upin a couple places,
where people have askedand said oh, the
review stars havebeen shown less frequently.And we double checked
with the team on that,actually, just to make sure.Because usually, it's normal
that the structured datamark-- the rich snippets
come and go, dependingon changes on our side.But they double checked.And they said there was actually
a bug and it should be fixed.So it should be
getting fixed soon,so that these will
show up again,a little bit more visibly.
MALE SPEAKER: It's back.It's back.
JOHN MUELLER: It's back.OK.I'm never sure if that's a
good thing or a bad thing.
MALE SPEAKER: It's a good thing.
JOHN MUELLER: But it's good
to know that the turnaroundtime is pretty good.
MALE SPEAKER: Or your
JOHN MUELLER: All
right, perfect.OK, so with that, let's
take a break here.I will be joining the AMP
Hangout later today as wellif you want to join in there.That's with the Google project
manager for the AMP team,which I'm sure will be
interesting as well.All right, so otherwise, I
wish you a great weekend.And maybe I'll see you again
in one of the future Hangouts.