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: All right.Welcome everyone to today's
Google Webmaster Central OfficeHours Hangouts.My name is John Mueller.I am a Webmaster Trends Analyst
here at Google in Switzerland.And part of what I do is talk
with webmasters and publisherslike the folks here
in the Hangout,and also the people who
submitted questions, of course.As always, we have
a chance for someonewho is kind of new to these
hangouts to ask a questionbefore we get started
with the submitted ones.I think, Michael,
you had one that youwanted to throw our way.
MICHAEL: Yeah, sure.So regarding the Search
Console, specifically the searchanalytics.I've got a few clients
who are focusing on this,and saying, hey,
it says that I'mranking number one for this,
that, and that search term.And it's even in a lot of
cases-- I'm looking at one nowthat's showing
clicks, impressions,click-through rates.But yet when they click the
link that goes to the SERPs,there's nothing there.And I've tried replicating this
by forcing the search through.co, .uk, .com, .us,
even using a Tor browser.And I can't replicate
the results.So I'm just wondering
where the data comes from.
JOHN MUELLER: OK.So we don't make it up.It's essentially data that we've
seen when we show the searchresults to people.So it's not that this is like
a theoretical ranking, whereyour site would be ranking, or
the impressions that it wouldbe receiving, but
rather actuallywhat we showed at
search, and howit was ranking at that time.So that can include things like
personalization, of course,which could play a role there.It can include things
like the locationof the user, the search settings
that they used to search there.One thing that I sometimes
use to try to figure outis this a general ranking,
or is this a-- maybekind of exceptional
personalization-type issue,is to look at the number of
impressions that I see there.And if it's a query where I
think the number of impressionsshould be high,
but Search Consolesays it's a very low
number of impressions,then probably it's something
around personalization.So if it's like
a one-word query,where I imagine the number of
queries should be millions,and I see a couple
hundred impressions,then maybe it's just
from personalizationthat we showed it there.The other thing that
also plays a rolethere is the universal search
results, where we show thingslike the images on top.And if your site
has an image that'sshown in the universal
search results on top,then we would count that
for the web results as well.Because it's not just
limited to lookingat the rankings of the
web pages themselves,but rather in this kind of--
not specific to images or videoor whatever, your
site was showing upin one of these high positions
for that specific query.
MICHAEL: OK.That's help clear it up for me.Because I was trying to
create a relationshipbetween the analytics SERPs,
and the Webmaster SERPs.And for me, it was more or
less what you were saying.It was-- I didn't use
the term personalization.But I used localized settings.So if somebody was on a mobile,
and they were close to thatcompany and doing a search,
chances are they'd turn upat the top of the results,
especially if they're connectedin some way to that company
through Google+ or through someother means.But the images isn't
something I considered.So thanks for that.
JOHN MUELLER: OK.Great.Let's do another quick
question from any of youall, if there is something
still on your mind.
MALE SPEAKER: Hi, John.It's [INAUDIBLE].I'm here in Toronto.I just wanted to ask
a quick question.Did you have a chance
to look at my mystery?
JOHN MUELLER: Which mystery?Oh, the mobile mystery.Yeah.We talked with the mobile
search team about that.But I don't know if I have an
answer that I can give you.
MALE SPEAKER: OK.Because I just wanted
to let you know,I checked it with Keynote.I checked it with Splunk.I checked it with
pretty much everything.Everything seems to be OK.And then there's no
severity in the mobile area,regarding robots.txt.Everything's fine.Like--
JOHN MUELLER: Yeah.Yeah.So just to give some
context to everyone elseyou're seeing essentially
a site showing up normallyin rankings on desktop,
and very low on mobile.And from our point of view,
sometimes that can happen.But it's always good to kind
of look into the detailsto see is everything working
as expected on our side,or do we have to
tweak something there.So I've been talking
with the mobile teamabout that specific
case to see if there'ssomething we can do
to kind of recognizethese things a
little bit better.
MALE SPEAKER: And so
any-- approximatelydo you know when it
will return, you think?
JOHN MUELLER: I can't make
any promises [INAUDIBLE].But I'll double-check
with the teamto see what I can let
you know about that.
MALE SPEAKER: I appreciate it.Thank you so much.
JOHN MUELLER: Sure.
MALE SPEAKER: I think--
LYNN: Hi, John.Lynn from the US.I have a question.
JOHN MUELLER: Sure.Go for it.
LYNN: Thank you.So I'm with a
company that providesreviews and question-and-answer
services to other companies.And our question is
around syndicated content.Because we have that as
an option that we provide.And in the past they have
thought that they were-- well,they were avoiding the duplicate
content issues by the factthat they were loading
it with AJAX call.But as we know now that Google's
changed that, and can seeand index AJAX calls.Then they're concerned
about what the right wayto handle this is when you don't
have access to the meta tags.And so we're trying to come up
with the right, best policy,and explanations
to our customers,and that kind of thing.And so what I wasn't sure about
was the X-Robots-Tag behavior.If you were to make
that call, I mean,if you were to include
the X-Robots-Tag headersin the result of just the
AJAX call that brings backthe-- essentially the
partial duplicate content-- Imean the part of the page.Not the entire page is
necessarily duplicate content,but the syndicated part
that we're bringing in,done by an AJAX call.So if we made that AJAX call
have those header responses,like a no-index, does that
no-index the entire page,or does it just tell Google
not to include that AJAX call'sworth of duplicate content in
the page that it would index?
JOHN MUELLER: I'd have to
double-check how we currentlyhandle this.In the past we've
essentially ignoredthat, if it's a part
of the AJAX response,and not something that's
embedded within the pagedirectly.So one thing that-- I
also modify the head elements.So you could add a row
well, when we render the page.If it's just for a
part of the page,then I don't think that
would really change anythingif you included an X-Robots-Tag
still ignore that.If you want to exclude
a part of a pagefrom being used for
crawling and indexing,then the best way to
do that is probablyjust using the robots
text file, and kindof blocking that response
or that part of the pagefrom actually being
crawled itself.But I don't know if that
would even make sensein your specific case.
LYNN: Yeah.OK.It seems like I've seen that
that's the last piece of advicethat I've read that they don't
want you to do is block it.
JOHN MUELLER: So, yeah.So are these like
blocks of textsthat are included within
a normal web page?Or--
LYNN: Yeah.So well the typical scenario
would be a product detailspage on a retail site.So-- and if they've had reviews
and questions and answerswith their own
customers have provided,then those would be
in the page first.So there would be a product
description and pictures,and add-to-your-cart buttons,
and that's kind of things,and then reviews and/or
questions and answers.And some of that content is
their own from their users.But then if they're
just new or spinning up,and they don't have
much of that built upyet, then we would also bring
in from a brand partner,like if there's a brand site
for that kind of thing--that particular
product, and that sitehas accumulated questions
and answers and reviews,we would pull in those questions
and answers or reviews.
JOHN MUELLER: OK.
LYNN: And so-- there
could be-- that's whereit would be-- that's why I
say it's part of the page thathas the content.And that part, the
syndicated part,we had done with an AJAX
call, even though it stillcan be embedded in
the page afterward.But I mean not an overlay
or whatever, but-- and priorto Google indexing
AJAX that would hide itfrom indexing content, right?
JOHN MUELLER: Yeah.So I guess, in a
situation like that,I would probably tend towards
maybe moving that contentinto a separate directory
within your site,where you're aggregating this.So that that specific response
could be blocked by robots.txt.So that maybe a request
for their own contentwouldn't be blocked,
but a requestfor the aggregated content
would be blocked by robots.txt,just to kind of prevent the
situation where it startslooking like they're
auto-generating content thatactually doesn't
exist on their site.
LYNN: Mm-hmm.OK.But if we dynamic--
so if we dynamicallyadjust the meta tags on the
page like to add no-index,then we're just no-indexing
the entire page, right?
JOHN MUELLER: Exactly.Yes.
LYNN: And you think--
but my question, I guess,is still on the HTTP header tag
from just the call on the AJAXcontent.You're not sure whether that
affects-- whether it says,hey Google don't index the
entire page, you know, or justdon't index this content
coming from this AJAX call?
JOHN MUELLER: It
definitely wouldn'tno-index the entire page.But I don't think that we would
use it for that specific AJAXcall.Because what we try
to do is essentiallyrender the page as it
would within a browser,and look at the final
results, and usethe final results for search,
for crawling and indexing.So if one of those requests
has a header attached to itthat essentially is
invisible in a browserwhen you look at
the final result,then we kind of ignore that.Whereas if one of the
results-- one of the kind of--
LYNN: You'd ignore the header.You don't mean you'd
ignore the content.You mean you'd
ignore the header?
JOHN MUELLER: We'd ignore
the header, exactly.
LYNN: Right.Right, OK.
JOHN MUELLER: So if you
want to no-index or relcanonical the whole
page or something else,you could do that
that if you just wanta part of the page removed.I think for the most
part, we'll probablyjust recognize that this is
a block of duplicate content,and ignore it.But if you really
want to make surethat it's not taking
into account at all,then roboting the request
on your side probablywould be the easiest approach.
JOHN MUELLER: It's a
tricky situation, though.
LYNN: Yeah.Yeah, right.Because it seems to us,
a legitimate use case.Obviously they're
not really, I mean,they are trying to have--
add value to the customer.But still offer reviews if
they haven't got their own yet.
JOHN MUELLER: Definitely.Or if you can send
me some examples,maybe privately on Google+, I
can double-check with the teamto really make sure that it's
working the way you want itto work, and the way
that we want it to work.And maybe I can refine
that into somethingthat we can put out
publicly as well.
LYNN: OK.Well, one thing in
my research that Iwas interested in though,
that I came acrossfrom someone who wrote up
what they found with tryingthis Java-- this Google index
up, this particular personwas talking to Gary Illyes,
is it-- from Google?But anyhow that
Google doesn't alwaysindex the result of
Java-- doesn't alwayschoose to crawl the
it seemed like his--we can't just assume that
all-- so the announcementis Google can crawl AJAX
now, but are they always?Is it still hit and miss?We may crawl your
AJAX and we may not.
JOHN MUELLER: I'd say at the
moment it's still on and off.But it's definitely
heading in the directionthat we're doing
it more and more.So if you need to rely
on it not being indexed,then I wouldn't
rely on Google kindof forgetting to crawl the
LYNN: OK.But from the standpoint of
trying to dynamically putin tags from your
depending on if Googlehas chosen to-- huh?
JOHN MUELLER: I mean
from our point of view,when we see tests like that, we
see that more as a bug, ratherthan a feature.So if we see that
we're not picking upthe heading tags properly
the team would say,well, we have to work on making
this actually work betterand more consistently.So it's not that
we say, well, thisis a way you can kind of
sneak content to the user thatisn't being indexed.We essentially want to get
all of that content indexedas well.
LYNN: All right.Thanks so much.
JOHN MUELLER: Sure.All right.Let's run through some of the
questions that were submitted.And if any of you all
have questions or commentsin between, feel
free to jump on in.Let's start here.A good website related
to our industryhas invited us to produce
an article for them.In exchange they said
to give us creditby means of a text mentioned
for the company name, as opposedto a link.Can this still give
us a ranking benefit,even though there's no link?From our point of view, this
wouldn't pass page rank.So this isn't
something where youwould see any kind
of ranking changewith regards to
just a text mention.That said, even if
there were a link there,then that seems like a
kind of a tricky situation.Because you're exchanging
an article for a link,which would be
seen as being kindof problematic from
the web spam team.So it's something where I
think a text mention is great.A link in there with a
no-follow would be a great way.And it kind of encourages people
to go to your website directly.And from that, maybe you'll
see an indirect effect.But at least there's
no direct effectwith regards to ranking there.
There was a companythat I know around somewhere,
and they were requestinglike $200, you know, and
then the reporter wouldhave linked it somewhere.But you guys nicely found
that and took care of that.So it's no longer-- people
from that specific areano longer do it.
JOHN MUELLER: Great.Yeah.I mean the web spam team
watches out for these thingsall the time.So reports or comments when
you're seeing somethingproblematic definitely helps.I noticed our SEL agency
previously disavowed a linkto our site, which
was a genuine mentionfrom an non-industry
related forum.Is that a bad thing?Should I remove it
from the disavow file,as it's a valid link?The forum mentions help.If you think that this
is a legitimate link,I would definitely remove
it from the disavow file.Just because it's from
different industrydoesn't mean it's a bad link.Or just because
it's from a forumdoesn't mean it's
necessarily a bad link.So from that point of
view, I would definitelykeep those kind of
links in general.The other thing kind
of to keep in mindis if you're looking at it,
from a link-by-link basis,and you're saying, oh, I
accidentally put this one linkin my disavow file, then
removing that from the disavowfile is unlikely
to change anythingfrom a practical point of view.Because it's just we're looking
at so many different thingsfor your website,
one individual linkisn't really going to sway
things this way or that way.Does the ranking
benefit to changefrom HTTPS outweigh the loss
you would get from a 301with regards to link juice?Yes.So essentially when you're
moving from HTTP to HTTPS,with a redirect like this,
you don't lose any patron.It's not something you really
need to worry about there.On the other hand, the
ranking boosts for HTTPSis more like a tiebreaker in the
sense that we look at the URLs,and if they're
kind of equivalent,then we'll swap them
with the HTTPS version.So it's not that it
would be a reallyheavyweight ranking
factor anyway.But even a lightweight
factor is more thanessentially a null factor.Right?Notice some e-commerce
sites, the iButtonLinksare no-follow, even though
the link leads straightto the product.Is that something
we should do or not?If this is a bi-link within
your website, then and no-followthere, it wouldn't
really be necessary.One thing the
no-follow does thoughis kind of prevent patrons
from passing through that link.So if that's a link
to a part of your sitethat you don't necessarily
need to have crawled,or you don't want to have
Google's crawlers spendtoo much time on, then maybe
a no-follow would be fine.In general, I wouldn't
recommend putting no-followswithin your website unless
it's really problematicthat Google actually
crawls in that direction.Just because adding
no-follows to someplaces might mean that
you forget about leavingthe no-follows there.And then suddenly your site
becomes really hard for usto crawl properly.Does it cause any harm to
have a canonical tap pointingto the canonicalized URL itself?No.That's perfectly fine.That's something you can
definitely do by default even.
DANIEL PICKEN: Can I just
ask on the canonical?
JOHN MUELLER: Sure.
Should you canonicalto the forward slash version
or the non-forward slash?The reason why I ask, if you
have a non-forward slash insay a browser, and
you copy and pasteinto somewhere like Excel,
it adds the forward slash.So I think in my mind I just
add the forward slash anyway.because when I'm copying
out of a browser it's there.So should you add
that to a canonical?Or should that be your URL,
therefore your canonical URL?
JOHN MUELLER: It depends.So there are two different
things to kind of differentiatethere.On the one hand, if you
just copy the host name,then the slash is kind of
implicitly there already.It's like you copy the
host name without a slash,then actually that means it's
the root of the website anyway.So the slash that's added there
automatically by some systemsis perfectly normal.And you can canonicalize
with or without the slash.We'll see it as the same URL.On the other hand, if there's
a path after the host name,like directory
name or file name,then the slash is
essentially a different URL.So we will treat those
as being different URLs.And if you want to choose one
or the for canonicalization,then you should
explicitly mention that.
DANIEL PICKEN: OK.
JOHN MUELLER: So if there's
no path, do whatever you want.If there is a path, make
sure you're consistent.
DANIEL PICKEN: OK.So as long we're
consistent, that's the key.
JOHN MUELLER: Yeah.
DANIEL PICKEN: OK.Thank you.
JOHN MUELLER: Do hreflang
tags work on a blog level?For example, can my Australian
and Canadian websitesshare the same blog
posts and use hreflangto avoid duplicate content?Sure.You can definitely use
hreflang across blogs.It doesn't matter what
type of site it is.If these pages are equivalent
and you have different versionsfor different languages
or different regions,then you can use hreflang.Can a page layout, text
font sizes, et cetera,have an impact on rankings?For example an e-commerce
site with a product pageshowing description
prices, et cetera,but the prices are made more
prominent in the font size.Is there any situation
where it would?I don't think we would
really use that for rankings.Except for the part
where we do tryto understand when a page has
primarily ads above the fold.Then that's the
kind of situationwhere we might react to
things like the font sizeor the layout of a page.But in general, if you have
this content on your pageand it's visible, and
one is like a biggerfont and the other
is a smaller font,then I don't think
you'd see any differencewith regards to ranking there.Now what's the difference
between HTTPS and HTTP/2?Why should one go
for HTTP/2 whenthey've already moved to HTTPS?So HTTP/2 is essentially
a different wayof accessing your
server that makesit possible to request multiple
things at the same time,very simplified.I'm sure there are lots of
details I'm skipping over.Whereas HTTPS is
essentially a secure channelto your website's content.So they're essentially
different things,looking at different parts.From a practical point
of view, most browsersrequire that you
use HTTPS in orderto fully implement HTTP/2.So it's almost like
moving to HTTPSis a good step for those of
you who are planning on movingto HTTP/2 at some point.Another thing to keep
in mind is your servercan support HTTP/2
transparently in the sensethat requests will automatically
be upgraded to HTTP/2.It's not something that
Google search reallyneeds to know about.It's not something
users really recognize.It's not like a site move.It's essentially a change in
your server's infrastructurethat users probably
wouldn't evennotice, other than that
suddenly things go a lot faster.Does Google consider
SSL as a ranking signalfor local businesses' websites?So SSL is actually
the old protocol.The new one is TLS.And that's used for HTTPS pages.And we use that for
all kinds of websites,so not specific to
local business websites.Do you know when Google
will be marketingnon-SSL sites as insecure?I don't know.I don't think that's
Google, in general,but rather Chrome specifically.And there are some posts from
the Chrome team about that.So I'd double-check with
the Chrome team on that.Why does Google Webmaster
Tools have lots of links,and links to my
site but-- and alsoon other link analysis tools.But a search over Google,
then nothing is found.Whereas we know that most
of them are do-follow.So I think there are different
things being mixed up here.On one hand, if you do a
link query in Google search,and you'll just see
a sample of the linksthat we show to your site.The links that we
show in Search Consoleis essentially a bigger sample
of links that we show there.And whether or not
they're no-followis essentially irrelevant for
both of those, in that we'llshow those links anyway.So do-follow or
no-follow doesn't reallymatter for those links reports.But you can't easily compare the
link report from a link queryto the link information
you'd get in Search Console.And usually external tools find
other kinds of links as well.So these are different
sources, different waysof looking for links, and
not easily comparable.
ROBB YOUNG: John, can I
ask an analytics question?I don't know if you'll know it.
JOHN MUELLER: Maybe.I can try.
ROBB YOUNG: Can I share
an image with you?Can I just do that in the chat?Is that how I do it?Does that work?
JOHN MUELLER: I don't
ROBB YOUNG: I'll try
and explain it then.
JOHN MUELLER: OK.
ROBB YOUNG: Or share my
screen, but basically whenwe're looking in
analytics, we havelet's say, 1,000 impressions
on some things, 300 or 400on others, and they're resulting
in a handful of click-throughs.We have one of the search
queries is the whole domain, sowww.experiencedays.com.It's coming up with
roughly 1,500 impressions.So someone will be searching
specifically for that.It's getting 10 clicks,
even though it's rankingan average of position one.
JOHN MUELLER: OK.
ROBB YOUNG: So does that mean
that it's-- is that somethingwhere it's the browser?So let's say in
Chrome, for examplethese days, if you
go to Google and youstart typing in a query,
it moves it to the header.It moves it into
the normal search.So it's kind of performing
a query, and kind of not.Does it have
anything to do that?I mean because
normally, if you'regetting 1500 people
searching for an exact domainwith the www, you're going
to get more than 10 clicksout 1500, particularly if
its average position 1.1.So I'm wondering
how it treats the--what you're searching to
versus what moves to the top.And if they go straight there,
or if Google is actuallyperforming a search
at that point,but still takes you directly
to the domain anyway.
JOHN MUELLER: I think we
only count it as a searchif we really show the
Search Results page.So if it's someone,
like typing the URL in,and they always
have it complete.And we show an
auto-complete type thing,then that wouldn't
count as a search.Whereas if they
actually do a search,and we show the
search results page,then that would
count as a search.My suspicion, without looking at
what your search results wouldlook like, is that
they're probablyclicking on something
else, somethingthat we're not counting.Which could be
something like an ad,if you had an ad
for your website.It could be something like-- I
don't know-- a knowledge graphentry on the side, or
something like that.
ROBB YOUNG: I just shared
my screen if that helps.
JOHN MUELLER: Yeah.
ROBB YOUNG: So this
entry here, the--
JOHN MUELLER: Yeah.What I would do there is
try to reproduce that query,and see what all is shown.Maybe kind of
double-check that you'relooking in the same locale,
the same country settings.And just double-check to see
what's actually shown there.And if there are any, let's say
nonstandard search UI featuresthat are visible there,
then perhaps they'reclicking on that.And we're not counting
that for Search Console.
ROBB YOUNG: Then it seems like
a very standard set of results.There's no targeting.I've not put loads of
parameters in there.I've just gone straight
there and said, show mewhat queries are doing what.
JOHN MUELLER: Yeah.But I'd double-check
in the search results.Just kind of see what's
actually being shown there.My guess is we're
showing your site.And they're clicking
through to go to your site.Because I mean, what else would
they be doing with the queryessentially with the URL?But maybe we're just
not counting thatas a click in Search Console.
ROBB YOUNG: I mean how do you
even do that-- say that again.
MALE SPEAKER: Maybe they're
clicking on the site links.Because usually when
you're searchingfor a brand name,
or the exact URL,you were being shown
a really a high numberof site links, [INAUDIBLE],
and so maybe they'reclicking directly on
one of the site links.
JOHN MUELLER: I don't know.I mean, what I
know we don't countis the knowledge graph sidebar.So if I search for
your company name,then it has a link
in the sidebar thatgoes to your website as well.So that's something we wouldn't
count in Search Console.But I think the site links
we should be counting.Otherwise I'd see that as a bug.But I'd kind of double-check to
see what search UI features arebeing shown to
people, and kind ofguess where people might click.And if you think that we're just
not counting something that weshould be counting, then--
ROBB YOUNG: Well,
you're definitelynot counting something that
you should be counting.I think we know that with the--
JOHN MUELLER: OK.I'll double-check with
the team to see what'sactually happening there.I mean we shouldn't be
counted things like ad clicks.We shouldn't be
counting probablythings like the
knowledge graph sidebar.But site links, for example, we
should definitely be counting.
ROBB YOUNG: Right.So how does Google
handle that in Chrome,if you start googling?It goes straight
to the address bar.If you're doing
a www, you've gotno choice but to go
straight to the website.You can't even google
that domain from Chromewithout disabling
certain features.Because it'll just
take you to the domain.There's no search involved.Because if you go to
google.com and starttyping in the search
box it'll take youstraight to the address bar.
JOHN MUELLER: Really?OK.I haven't actually
tried that for a while.Because I always search from
the address bar directly.
ROBB YOUNG: Right.So it's a little
bit frustrating.Because particularly from
our side, or SEO's side,when you're trying to google a
whole domain, because you wantto see the results
for that, you almostcan't do it without
disabling certain features.Because it just moves
it straight up to the--
JOHN MUELLER: Yeah.So what I usually do is I enter
a question mark in the addressbar, and then the
domain name, and thenthat will be used as a query.
ROBB YOUNG: Yeah.But if your-- let's give you
the benefit of the doubt,and say you're a power user,
and not an average Chrome user,if normal out of
those 1,500 people,if a normal person
just starts typing,they're not even searching.
JOHN MUELLER: I don't know.
ROBB YOUNG: They're
just going there.
JOHN MUELLER: I don't know how
that's handled at the moment.
ROBB YOUNG: Because, I guess
if it goes straight there,is somewhere in the
background Googleperforming a search for that,
and just doesn't bother.Because you're obviously
looking for domain.Or is it no search at all?It just goes straight there.
JOHN MUELLER: I don't know.I mean when I tried here with
Google Switzerland, whichcomes up-- maybe
it's different here--if I enter a domain name
in the search box there,it does a query for
that domain name.So it doesn't go
directly to the website.But maybe it's
different in the UK.I don't know.Some of these things are subtly
different from [INAUDIBLE]versions.
ROBB YOUNG: Right.OK.All right.
JOHN MUELLER: OK.We're looking at implementing
an onload model lightboxfor new users, the site the
background of the lightboxis 0.5.Will Googlebot see this overlay
as blocking content behind it,subsequently discounting the
copy for ranking purposes?That's always a
tricky situation.But essentially, we
probably see thatas a kind of an
interstitial almost.You can double-check to see
how Googlebot would see thatby using fetch and
render, in Search Console.So I'd double-check there to
see what it actually looks like.But also maybe
double-check your analyticsto see that this is
really something that'shelpful for your users,
rather than just sending themaway again.How are impressions calculated
if a page ranks anywherein Google?Is that also calculated?No.It's really based on
the search resultsthat were actually shown.So if a site ranks on page
2, and the users only searchon page 1, then we wouldn't
count that as an impression.When committing a brand
search on a mobile phone,both our iPhone iPad apps appear
in the search results pagewith our logo.Could we influence it to
only show one of them?I don't know.I haven't seen that.If you could send me
a query like that,maybe on my Google+ directly,
then I can double-check withthe team to see what's
actually happening there.Maybe both of these apps are
marked up with app indexing,and they're seen
as separate apps,and we show them separately.But maybe we can do a better job
of showing the right version,or at least showing
one canonical version.On an aggregate review with
perhaps 300 user reviews,would you display all of these
reviews on the same page?If not, should subsequent
pages be rel canonical or relnext and indexed?This is essentially up to you.So you can put them
all on one pageif you think that makes sense,
if you can put them maybein a scrolling box.That might be an option.If you want to put the separate
pages that's an option as well.With regards to
the rel canonicalthat's kind of up to you.If you think the reviews are
really important content,then maybe they should have
their own canonical URL.If you think that these
additional reviews are kindof less important
content, and you reallywant people to focus on
the main product pageand you can put a rel
canonical to that.If you do want to have them
indexed separately, then usingrel next and rel
previous is a good wayof letting us know
about the connectionbetween those different pages.Why is Google giving
preferences to OTA websitesin ranking for our
brand keywords,while our websites
are actual brand,for example, a
query from the UK?I have to double-check
what's actuallyhappening with
that specific queryto see if there's something
that we can do there,or that we need to do there.But in general, it's something
where our algorithms lookat a lot of different factors.And sometimes it makes sense
to show the original websitefirst.Sometimes we just see that
other websites are justas relevant or more
relevant to users,and we show those
in the top ranking.So that's something where we
don't explicitly make a calland say, well, we should
always show aggregators,or we should never show
aggregators in search.We essentially treat these
websites as separate websites,and try to rank
them appropriately.I have website that's more
than a few years on the air.The problem is all tools say
that we have implemented schemacorrectly.But we don't see any
information in search.So this is, I think
specific to rich snippets.And for rich snippets we have
essentially three criteriathat we watch out for.On the one hand, they have
to be technically implementedcorrectly.So it sounds like
that's the case here,if you're using
testing tools, and theysay this is technically
implemented correctly.On the other hand, it has be
implemented in a way that'saligned with our policies.So when someone looks
at these, for example,if they review them manually
based on maybe a spamreport, for example,
then the waythat the structured data
markup is used on a pagehas to be aligned
with our policies.So for example if you have
reviews on a product page,those reviews have to be
for that specific product,or for that primary piece
of content on the page.They shouldn't be
general reviews.And finally we also look at the
overall quality of a website.So if the quality of the
website isn't that great,then maybe we'll be more
cautious with regardsto rich snippets,
and say, well, Idon't know if we
really want to showthese in the search results.Because we aren't sure we can
actually trust this websiteto do everything properly.So those are the three things
I'd kind of watch out for.If you're sure that technically
they're implemented correctly,if from a policy point of
view you feel that everythingis fine, then I would
work on improvingthe quality of the website,
so that it's really clearthat this is a
high-quality website thatuses the structured
data markup correctly.Therefore it would
be a loss for Googleif we didn't show it in search.How much time is-- hi.
MALE SPEAKER: Yeah, John,
regarding [INAUDIBLE] snippet,I had--
JOHN MUELLER: Oh,
I can't hear you.Your voice has disappeared.
MALE SPEAKER: John, could
I ask a quick question?
JOHN MUELLER: Sure.Go for it.
MALE SPEAKER: It's with
respect to Google WebmasterTools in the back end.I have a situation-- we've
talked about it before--with the vacations
website out of Costa Rica.You may remember.It's a case very similar
to what Robb Young has.And my question
revolves around whyis it in the back end
with-- I'd love to--can I send you a screenshot,
like a link to a screenshot?
JOHN MUELLER: Sure.
MALE SPEAKER: OK.Let me get it here.
JOHN MUELLER: Or if it's
specific to a website maybe youcan just send me a
quick Google+ post,and I can take a look
at that specifically.
MALE SPEAKER: I'll just
put it in the chat.It's just a link
to a screen shot.And then inside of
that screenshot,you're going to see a real
big difference between whereGoogle shows us ranking
and the actual impressions.We get no impressions
for a phrase thatgets 22,000 searches per day.Yet Google sometimes will
show us ranking on page 1.This is a 15-year-old domain.To give you a real elevator
pitch on this domain,it's 15 years old.Originally you said it was
hit with a historic Penguin.Then you told me it was fixed.Then we had a site
review on it on here.And it got slammed pretty hard
with a lot of Vietnamese linksafter I appeared on a
site review on here.
JOHN MUELLER: Oh, no.
MALE SPEAKER: And then
we disavowed all that.And we cleaned it up.And the last time I had
a private chat with you,really short, I
suggested that wewere moving in the direction of
putting up a brand new website.Well six months, five
months after our chat,we released our new
website on December 14th.It's the same effect.It's the same.I can't-- I guess this is a
really loaded question witha bunch of questions.So I guess let's
stick to the point.Why is it that Google is
not showing this website?There's no impressions at
all despite a high ranking,according to Webmaster Tools.It's the only place that
shows a high ranking.Everywhere else it's 200-300.
JOHN MUELLER: I don't know.I'd have to double-check
with the team on that.But let me copy
it onto the site,so I can remember to
actually double-check.I'll have to look at that.
MALE SPEAKER: OK.Could you-- the Christmas
email that I sent you was--
JOHN MUELLER: I'll
try to find that.
MALE SPEAKER: I'll
send you another one.But could you expand
on how, why-- whythere is such a discrepancy
between the ranking that GoogleWebmaster Tools shows and
the actual impressionsof the website?
JOHN MUELLER: I imagine
that's similar to whatwe looked at way in the
beginning of the Hangout.In that with personalization,
what can happenis that we show a fairly high
ranking in the Search Console,because we've shown it
at that position before.But you'll see a low
impression numberfor a query that
probably should beshowing a lot of impressions.So if you see kind of the
combination of a querythat you expect to
have a lot of queriesor a lot of traffic for, and low
impressions and high ranking,then probably that's just
from personalization.In that a small number of
people are seeing the siteon the first page in
the search results.But most people
aren't seeing the siteon the first page in
the search results.So that's kind of that
tricky combinationthere, where we say, well
we've seen it on page 1,but actually only just for a
couple hundred impressions.And that is a number we'll
still show in Search Console.But you kind of have to
interpret that yourself,and recognize, oh well.This query should
be getting thousandsof impressions every day.I got 100 impressions
over 30 days.Therefore my site was probably
just sporadically showing upfor that query, not regularly.
MALE SPEAKER: So is
there a particular reasonthat you could talk about
why Google would repressa site like that in impressions,
specifically impressions,without any kind
of manual review?Or is there something that
you can see on your screen,like you did before
that this site is stillbeing slammed by something?
JOHN MUELLER: I'd
have to take a look.It's really hard to
double-check these things live.Some things are
really obvious, and Imight be able to say
something really quickly.But for most things when
it comes to ranking,it's not yes or no type thing.It's very-- lots of
shades in between.
ROBB YOUNG: Don,
are you saying it'sa new site or a new domain?Is it just a new design
code or a new domain?You've hidden the domain
from the screenshot.
MALE SPEAKER: It's a
15-year-old domain, actually.And--
ROBB YOUNG: So
it's just new code?Or a new design?
MALE SPEAKER: It's
just a new website.Yeah, because the last
session I had with Johnit was suggested that
it would be a good ideato redo the website.I've tried canonical.I've done hreflang.I've done, like I said, I
fixed the historic Penguin thatwas attached to this domain.But everything I do-- if
you see in that screenshot,you're going to see
three distinct spikes.Those spikes are when I
fetch as Google, that's it.The only time I get a
click, the only time Iget a high level of
impressions, and Iwouldn't call it very
high for a phrase thatgets 22,000 search a day.But those spikes are definitely
fetch as Google and that's it.Otherwise Google represses that
site from all search it seems.
JOHN MUELLER: But that's
for that specific query.
MALE SPEAKER: For that
specific front page.I'm only focused on that query,
because it's 22,000 a day.So it's an easy target.
JOHN MUELLER: Yeah.I'd probably have to take
a look at this in detailto see what
specifically we can sayor what might be
the problem there.
MALE SPEAKER: But if you want to
see the domain, it's very easy.Just add .com to the main
phrase that's shown in thatscreenshot.That's the domain.I'll send another email, John.Thank you.
JOHN MUELLER: Let me try to run
through some questions here.And then maybe we'll
have a few minutes timefor some more open questions.How much time does
it take Googleto de-index pages that
are blocked by robots.txt?So robots.txt doesn't
block indexing.It just blocks crawling.So maybe we'll keep
them indexed foreverif they're blocked
by robots.txt.If you want to remove
them from the index,make sure you use an no-index
meta tag and allow crawling.Is it necessary to have
AMP pages if we alreadyhave a responsive website?AMP or Accelerated Mobile
Pages are essentiallyseparate from mobile website
in that the mobile website issomething that
people will see whenthey're using mobile phones.And AMP pages are the things
that we can embed directlyin search in the
news carousel on top.That should be showing up
towards end of February.So I'd keep my
eyes open for thatto understand how these pages
are being shown in search,and to make a judgment
call on whether or notit makes sense for
you to also providethese kind of AMP pages.
MALE SPEAKER: OK, perfect.Because one of my new
sites, I looked atand, yeah, it's
still not showing.
JOHN MUELLER: Oh, yeah.I mean it's not launched yet.So you wouldn't
be seen that yet.But what I'd double-check
with regards to AMPis you need to make sure
that these pages arereally error-free.Where as soon as there is any
kind of error within the AMPpage, then we won't be using
that for the AMP carousel.So in Search Console you
have kind of aggregated errorinformation.That gives you information
on, in general,how error-free are my AMP pages.But you can also double-check
them manually individually.And when checking
them individually youneed to double-check,
on the one hand,the technical
implementation of AMP,which is by adding hash
development equals 1,and checking the
structured data informationfor the AMP pages, which you can
check with the structured datatesting tool.
MALE SPEAKER: OK.Thanks.
JOHN MUELLER: All right.We just have five minutes left.So I'll open it up to you.We still have tons
of questions left.So it seems like
maybe we should doanother one of these
hangouts at some point.
MALE SPEAKER: Just mobile
friendly, I talked to youabout that about a
year ago, regardingon how errors still show.I mean there's basically
a delay in clearingthe mobile usability
JOHN MUELLER: That's
always the case.Essentially we have to
re-crawl and re-indexthe pages themselves.So that always has some
kind of a delay there.
MALE SPEAKER: So you guys aren't
going to be speeding it upeventually?Mobile is so critical now.
JOHN MUELLER: We're always
trying to speed up things.But when it to crawling
and re-indexing,we also need to get adopt to
what the server can provide.And that's on your side, right?So if we can't crawl your
website that quickly,we can't update our
index that quickly.And especially with
regards to mobile pages,if it's not a
responsive page, wehave to re-crawl it with both
mobile and desktop user agentsagain, to double-check
the conceptthat we're seeing there.
MALE SPEAKER: John, I try to
stay within milliseconds, John,milliseconds.
JOHN MUELLER: OK.OK, that's a high
goal. that's good.
DANIEL PICKEN: John, can I just
ask a quick question, please,around tile tags?
JOHN MUELLER: Sure.
DANIEL PICKEN: I've just
shared an example in the chat.The example, it's a
URL of Google search.I search for a brand name.The brand name is two
words with a space.However it retains
the top result,the title tag with no space.However, in the title
tag itself on the site,It's got the spacer.So I'm just wondering why
Google would alter a brand nameor a particular [INAUDIBLE].I'm looking for
[INAUDIBLE], space, i.e.a brand name.I have that on the site.So I'm not sure where
it's picking that up,and why it's taking
the space out.
JOHN MUELLER: Probably because
it's like that in the URL.
DANIEL PICKEN: Right.OK.That's interesting.
JOHN MUELLER: I'm just guessing.I mean the tricky part here
is you call it a brand name.But actually it's
just like two words,which could be seen as
just two connected words.So sometimes what will
happen is we'll say,well, this isn't specifically a
combination that needs a space,but rather we find
a lot of URLs,for example, that don't
have a space in there.So we'll show the title in the
shortened version like that.One thing that you
can generally do--I don't know if that's the case
in this specific situation--is just make sure that you
really have unique title tagsacross your site, and that
you're not using the same titleon lots of different pages.Because if we see something
that makes it look like,well this title is not
specific to this page,then we'll try to invent
a new title for this page.Whereas if we can recognize
that this title isunique for this page and we
can trust that we believewe can show this in
the search results,then we'll try to reuse what
you've provided on that page.So the better your titles
are directly, the more likelywe'll actually show them
in the search results.
DANIEL PICKEN: And just one
more question, quick one.302 redirects, you
said the past pagerank, how does that
work in a search engine?Because a 302 is a
temporary redirect.So are you still looking
at the original contentand indexing that?Surely you have to index
both pages as well.So I just want
some clarificationaround that, please.
JOHN MUELLER: Oh, man.These 302 redirect questions
will never go away.So essentially from
our point of view,what happens with
redirect is eitherwe index the destination
URL, so whereyou're redirecting
to, or we indexthe URL where it's
redirecting from.So in both of those
cases, we essentiallykeep the page rank normally.So either we pass
it through the URLwhere you're redirecting to.Or we keep it on the URL
that you're redirecting from.And with a 302 redirect,
technically you'resaying you want the URL, you're
redirecting from to be indexed.From our point of view, if
you use a rel canonical,you can specify
either one anyway.So it's not that the
page rank evaporates.It's either on the
URL that you'reredirecting from, or on the
URL that it's redirecting to.
DANIEL PICKEN: And
which content--would you take the original
content, or the new content?
JOHN MUELLER: Well, we'd only
see the final destinationcontent.If you have a server
side redirect,then we wouldn't see what
else would be on the pageif it didn't redirect.Because it's just redirected
us to the other page.So we'd always use
the final content.But we'd show it under either
the old or the new URL.
DANIEL PICKEN: Right.OK.OK, thank you.
JOHN MUELLER: All right.
ROBB YOUNG: John?John, can I ask a
very good question?Or you can even have
time for another one.I was just trying to Mihai.In regards to my
previous question,is it one search result
with 10 site links?Is that 11 impressions
or one impression?
JOHN MUELLER: It's per
search results page.It counts as one impression.So if you have multiple
URLs from your sitein the same [INAUDIBLE],
that would be one impression.
MIHAI APERGHIS: Are you sure
that's true for site linksas well?Because I was looking at
Robb's Webmaster Search Consoleanalytics, and all
of the site linkshave the classic
number of impressions.
JOHN MUELLER: So let
me be more elaborate.If you're looking at it from
a per query point of view,then it's one impression
per search resultspage for that query.So for a specific query,
we would count thatas one impression, even if
you have multiple pages.If you're looking at
it on a per-page basis,so if you take the filter
in search analyticsand show the page, then we show
those impressions separately.So if you have [INAUDIBLE],
if you've gotten impressions,for example, if you look
at it on a per-query basiswe see well we showed your site
once with these search results.
ROBB YOUNG: I thought
Mihai was a genius.
JOHN MUELLER: So you can't
compare those numbers.I guess that's the point there.But it's kind of
tricky in that wetry to show things that
are relevant to you.But at the same time,
if you think too far,then you're like, well
this doesn't add up.So--
ROBB YOUNG: I guess
you can't reallyshow 11 impressions there.And then how would that
affect the average position?Is it one?I mean-- 11 impressions
in position one, then?
JOHN MUELLER: For
the average positionwe use the average top position.So if you have
multiple URLs thatare ranking for the same
query, we take the top one.And we use that as an
average for different queriesacross different searches.So there's a lot
of math behind it.
ROBB YOUNG: In the
future, I'm justgoing to stick to asking
for a solution to our site.Because these technical things
are too difficult for me.
MALE SPEAKER: Great.
JOHN MUELLER: I think
[INAUDIBLE] would be easier.Sure.I like it.All right.So let's take a break here.Thank you all for coming.I wish you all a great weekend.And I hope you don't
have to go back to bedand Kenon said, get some early
work done if you're in the US.Hope to see you again in
one of the future Hangouts.
DANIEL PICKEN: Have
a good weekend, John.Thank you.