Transcript Of The Office Hours Hangout
Click on any line of text to go to that point in the video
JOHN MUELLER: Welcome, everyone,
to today's Google Webmaster
Central Office Hours Hangout.
My name is John Mueller.
I am a webmaster trends analyst
here at Google in Switzerland.
And a part of what I do is talk
with webmasters and publishers
like you all.
It seems I have
some new faces here
in the Hangout,
which is fantastic.
Do any of you who joined
these for the first time
have any questions that
you'd like to have answered
or anything that's been unclear
that's been on your mind?
MALE SPEAKER 1: Hey,
John, I've got a question.
JOHN MUELLER: OK.
MALE SPEAKER 1:
We've got a blog.
And we've got about
800 blog posts,
and those are
ranking quite well.
And I'm 100% sure
that those blogs
don't have any algorithmic
or manual issues with Google.
So what happened was
last weekend suddenly
17 of those 800 blog
posts have disappeared.
So the first thing
I did was check
if those blog posts are
crawlable for Googlebot.
So there's no
no-follow, no-index tag.
There's nothing in
the robots tags.
And I am quite confused
because those blog posts were
ranking quite well and driving
quite a lot of traffic,
and suddenly, it's
just disappeared.
So one our developers were
mentioning about [? jQuery ?]
comments and one
[? jQuery ?] comment.
They just forget to close a
comment in the [? jQuery ?]
in the header section.
So we were just asking if
does that confuse Google bot?
Do they index, so to
say, the blog posts?
JOHN MUELLER: That
shouldn't be the case.
So essentially
what you're saying
is you kind of had some broken
HTML on these pages, right?
It's like the plug-in
generated some HTML that
wasn't well-formed,
something around that.
That shouldn't be a
reason why it would
remove a page from the index.
So usually what we'd look
for there is something
like a no-index robots tag, or
if it's maybe returning a 404,
or maybe it's a 500 error,
or something like that.
Sometimes what we've
seen is that the server
returns this just to Googlebot
for some weird reason.
So maybe they were trying
to do something else
that was kind of special
there, and somehow that broke.
And now Googlebot is seeing
an error or no-index,
and normal users are
seeing the full content.
So that might be something
that's happening there.
I guess that's more
of a technical issue.
One thing I try to
differentiate is whether or not
this page is actually completely
gone from our search results,
from the index, or if
it's just not ranking.
MALE SPEAKER 1: No,
it's quite gone.
So you cannot just
even find any site.
You cannot find it
in a Google database,
which is quite confusing.
JOHN MUELLER: So like a site
query would say no results?
MALE SPEAKER 1:
Yeah, no results.
JOHN MUELLER: OK,
in a case like that,
either the Web Spam team took
action on it, which usually we
send you an email
through Webmaster Tools
so you'd see that there.
That's the first thing
I'd double check there.
The other thing that
might be happening
is especially if this
happened from one day
to the next, that's
something where maybe there
is a URL removal request
that was submitted.
MALE SPEAKER 1: We double
checked everything.
So we don't have any
requests, URL removal request
and the manual action messages.
So it's a kind of
a mysterious foe.
JOHN MUELLER: Yeah,
so one thing you
need to watch for with
the URL removal request
is we remove things regardless.
So if there are kind
of like similar URLs,
then we will remove that, which
means that you have the HTTP,
HTTPS versions www, non-www,
and you have one site removal
for one of those versions,
then that will apply
for all of those versions.
So, for example, if you try
to remove let's say the HTTPS
versions from the search
results because you wanted
to focus on the HTTP ones,
then the site removal
request there would
actually apply
to both of those versions.
MALE SPEAKER 1: So
we didn't actually
do any remove requests.
And as you say, if
it's a manual action,
you should actually
receive an email.
So the reason is--
there is no reason
to get it manually removed
by Google because that's just
completely normal
unique blog posts.
JOHN MUELLER: One
thing you could do
is maybe send me some
of those sample URLs,
and I can take a
look afterwards.
You can send them
to me on Google+,
and I will take a
look afterwards.
I can't guarantee that
there's something specific
I can get back to you.
Maybe what can also
happen is that there's
some kind of a
glitch on our side
where, for whatever reason,
these URLs dropped out briefly,
and they'll drop back in as
soon as we reprocess them.
But I can double
check on our side.
MALE SPEAKER 1:
OK, that's great.
JOHN MUELLER: Sure.
All right more questions?
What can we start off with?
MALE SPEAKER 2: Hey,
John, can start with one?
JOHN MUELLER: All right.
MALE SPEAKER 2:
Thank you very much.
A pretty short one.
A product page has additional
information structured in tabs,
because otherwise, we'll
render the page too long,
and we'll lose the page speed.
Well, I know there has been
talk a lot on this method
like adding that
info on the same page
and transforming the tabs into
expanding menus and so on.
What I'm interested
in it's what's
your latest advice on getting
that content indexed to
for adding a plus value?
JOHN MUELLER: The
latest advice is still
the same as the previous
advice, I'd say there.
So essentially if it's critical
information for your page--
and make sure it's visible.
If it's critical information
that isn't visible now,
maybe it makes sense
to set up separate URLs
for that information where
that tab is open by default,
and it's visible.
If it's just
additional information,
then it's fine to keep it
in a hidden tab like that.
MALE SPEAKER 2: OK, thank you.
JOHN MUELLER: All
right, I think, Darcy,
you had something as well?
DARCY: Yeah, I did leave
a question in the chat.
But I'll ask it anyway.
Specifically to
automotive industry,
I'm just wondering,
if when it makes
sense to combine multiple
businesses into one website.
So multiple businesses
have websites.
They're not exactly the same.
But they want to have the
Walmart-type structure where
it's here's the brand, and then
it's your Toronto, Mississauga,
whatever, your different
city-based brands.
When does that make sense?
What needs to happen
for that to make sense
from an SEO standpoint?
JOHN MUELLER: Essentially
that's up to you.
So from my point of view, if you
have more than a small number
of sites that are essentially
targeting something similar,
I'd aim to combine
those more because what
will happen there is you'll have
a much stronger central site
for all of this
information than you would
if you had to maintain these
individual websites, which
probably would have trouble
ranking on their own.
So that's something where if
you have different locations
or different niches
of the same business
that you're active
in, then that's
something where as soon as
you go past maybe two or three
sites, I'd aim to really try to
combine that into one website.
And to combine that, you
essentially just redirect URLs
to that central place and
see that as a site move.
But essentially
it's something where
there's no hard line on our
side where we say, well,
if you have four websites,
you must do exactly this,
or if you have websites that are
different in 50% of the words
on a page or some--
I don't know,
some other metric,
then you must do it
like this, that's not something
where we'd have a hard line on.
That's more something
where you should
try to think about it from
a marketing point of view,
like do you want to
present your business?
Is this essentially
one business that you
want to present to users
as something really strong?
Or do you really, really want
to separate these essentially
parts of your
business completely
and make sure that users really
identify them as something
completely separate.
So that might be if you like
targeting completely different
audiences, if you
have something that's
for consumers and something
for other businesses,
maybe it makes sense to keep
that completely separated.
DARCY: Yeah, my concern
was just over the types
of inventory and products.
One site might not carry
the same site specific
to automotive, so the
same car as the other one.
So that's why it's not sort
of the whole Walmart thing
where you can say
it's out of stock.
They may never even
carry that item.
So that's where I
got-- I wasn't sure.
Do they need to share
the same inventory
for that to make sense?
JOHN MUELLER: That's
essentially up to you.
You can do it either way.
I've seen sites that focus
on specific locations
where you say, in this location,
we have the set of inventory,
and it's kind of
structured that way.
And other sites take
focus on the inventory
and say this is what
we sell, and it's
available in these
and these locations.
So that's not
something where I'd
say from an SEO
point of view you
need to do it like
this or like that.
It's really up to
you how you want
to present your content, your
products, your locations.
DARCY: Excellent, thank you.
BARUCH LABUNSKI: John, can
I ask you a quick question
about server errors?
JOHN MUELLER: Sure.
BARUCH LABUNSKI:
So suppose there's
about three to four servers for
a site in a period of one year.
Does that have any effect
on the rankings or no?
I know you were already
starting to answer this.
But I just wanted to know
because it sticks there
for about a month, two months,
until it disappears back
to zero.
JOHN MUELLER: That
generally wouldn't
have any effect on the
site unless those URLs were
affected by that server error.
So, of course, if an important
URL on your site has a 404,
then that will
drop out of search,
and that'll affect your website.
On the other hand, if it's
some random URL that actually
never existed, and
it's a 404, then that
has absolutely no effect.
BARUCH LABUNSKI: OK, thanks.
DON HALBERT: John?
JOHN MUELLER: Hi.
DON HALBERT: A question-- did
you get my message in Google+?
JOHN MUELLER: Possibly.
DON HALBERT: I sent it 15
minutes before it started.
JOHN MUELLER: Oh, then
I probably didn't .
DON HALBERT: Do
you remember-- OK,
I don't know how to
get you the website
because I'm really, really
nervous about putting it
into the chat,
because the last time
I did that, I ended
up with an attack,
and we still haven't
recovered from that.
So how can we do this?
Do you remember my
website by chance?
The one with the
historic Penguin?
JOHN MUELLER: Historic Penguin.
I don't know.
DON HALBERT: The vacations one?
OK, whatever.
I've been attacked enough times.
I'll put it into the chat.
JOHN MUELLER: There's
also a way you can send
a direct message in the chat.
DON HALBERT: Oh,
well, a little late.
The cat's out of the bag.
JOHN MUELLER: Actually,
the people here are nice.
DON HALBERT: All
right, in a nutshell,
the advice I've
gotten from you is A,
it was a historic Penguin
that originally hit it.
I had a site audit done with
the group here about a year
and a bit ago.
And for the most part,
everything looked good.
And some people were
suggesting, hey,
you need to start over
with a new domain.
But you were adamant
that the domain was fine.
Since then I pay close
attention to what is told to Rob
and what has been
told to Gary Lee
because they seem to be in
a very similar situation.
As of late, we've actually
gone down to a skeleton staff.
And we lost eight more people
because we can't-- we're having
a very hard time
keeping the doors open.
It's not your issue.
My question is I've
done a hreflang.
I've done an entire translation
of the website into Spanish.
So if you go, you'll see an EN.
You'll also see an ES.
But now I haven't
done a Fetch as Google
since February or whatever,
which is good product for me
because I was doing it weekly.
But I'm desperate
right now, John.
I need to know--
I was anticipating
a release of something
that will at some point--
algorithms hitting your website.
And I'm wondering has there
been any more talk [INAUDIBLE]
because right now we spent
thousands translating
it and applying the hreflang.
And right now we have
a sandbox website
with a complete
redesign of the site
because that
template was actually
based on a template
design company off of one
of the major theme sites.
But they're out of Moldova,
and I'm at my wit's end here.
I don't know what to do.
So we're actually sandboxing a
new complete different design,
and maybe that'll help it.
I don't know what's going
on with that website.
But Google hates us.
JOHN MUELLER: We
try not to hate you.
When did you put
up the new design?
It looks really nice.
So from my point of
view, it's not something
where you're waiting
on a Penguin refresh
or anything like that.
It's essentially just
our normal algorithms,
the way that they're
looking at the site.
And I think something like a
redesign, from what I remember,
it looked very
different in the past.
I could imagine that
this is something
that would have a significant
effect on your website
where we look at
this, and we say,
well, this looks really nice.
This looks like a really
high-quality trustworthy site
that we should be
showing more in search.
So from that point of
view, it's something
where if you just recently
put this redesign up,
I could imagine that you
would see some changes.
No?
DON HALBERT: No that's--
the sandbox site is still
in the sandbox.
I have not even come
close to finishing it.
But before I invest another
$10,000 in the redesign,
I wanted you to look.
That's the same design
[INAUDIBLE] and an ES version.
And the canonical is applied.
JOHN MUELLER: I'd probably have
to take a look at the details
there to see what's happening.
But in general, it's not that
there's something specific
that we're waiting for for your
website where we're saying,
well, you need to
do this, or you need
to wait this algorithm out.
It's essentially our
normal algorithms,
the way they'd look at the site
to try to make sure that we're
showing something that's
really high quality that people
actually want to see.
And I probably would have to
take a look at your newer site
if that's not the current
one that's live there.
DON HALBERT: Would you
be so kind as to tell me
whether or not I still
have a historic Penguin
on my back on this website?
JOHN MUELLER: You
don't have anything
where I would say you'd need
to focus on links there.
So that's something
where if you've
been focusing on links for
Penguin or other web spam
issues, that's not something
where at the moment
I'd say you have to
worry about that.
DON HALBERT: Thank
you very much.
That's very helpful.
MALE SPEAKER 3: John,
is it a good time for me
to ask for any kind of
update as well then?
Since I haven't asked
you for about six months,
I've laid off for a while.
And we have just finished
in the last two weeks.
JOHN MUELLER: But can you
copy and paste your URL
into the chat or somewhere?
MALE SPEAKER 3:
Well, this is the one
that we're currently on.
This was-- that was the old.
We have just changed, as
you said, the entire HTML.
It's completely responsive.
It's new.
It looks similar.
But now it's 100% different.
The content and the products and
everything is still the same.
But this site is now
different to the old one.
So the old one is the X.
JOHN MUELLER: I don't
see anything holding back
your new site.
So I think that's kind of
evolving as it organically
would be evolving.
So it's not--
MALE SPEAKER 3:
There's no hreflang.
There's no 301s.
There's no nothing now in terms
of being officially connected
to the old.
JOHN MUELLER: It
looks like the traffic
is going up as well, right?
MALE SPEAKER 3:
Yeah, but that tends
to happen whenever
we do anything.
So when we-- if we
do the hreflang,
it works for a week or two.
If we do something else,
it works for a week or two.
If we 301, it works
for a week or two.
When we unwind the 301, that
works for a week or two.
And then it's not
until six weeks
down the line after any
change that we actually
see what the genuine effect
is once the whole thing's been
spidered, and it's flushed
out and essentially
been caught again.
JOHN MUELLER: From
what I see, I think
your site is headed in
the right direction.
And things are looking up there.
It's obviously hard to
compare to the old site.
But the new site looks like
it's as it normally should.
MALE SPEAKER 3: Right, and
there's no connected penalty
or connected--
JOHN MUELLER: Not
that I see there, no.
MALE SPEAKER 3: --change that's
going to hold that one back?
JOHN MUELLER: No.
MALE SPEAKER 3: All right,
OK, I shall everyone else
get in before I pester
you with something else.
JOHN MUELLER: All right.
Let me go through some of the
questions that were submitted.
And we'll see how far
we can get and go back
to questions from all of
you here in the Hangout.
"How do I deal
with content that's
similar for multiple products.
For example, FAQ pages
for different products
where the only change is
the title, description, H1.
Will having one page limit
my ranking potential,
or should I create
multiple pages?"
Essentially that's up to you.
I try to focus on combining
those pages as much
as possible so that you
have one really strong
page rather than essentially
a diluted set of pages
that are attached to your site.
So as much as possible,
I'd try to do something
like a rel canonical pointing
at your preferred version
so that we can focus
all of our signals
on that canonical version so
that we can really show that as
highly as possible in search.
What will happen otherwise is
we will index these separately.
We'll crawl them separately.
And we'll try to figure
out which one of these
is most relevant in search.
We won't show all of them.
We'll pick one of them
and show it in search
because it's essentially
very similar to the rest.
And that might not be the
one that you want us to show.
And it might not
be as strong as you
want us to show it
because we essentially
have to balance between all
these different versions.
So if you know that this content
is essentially identical,
use ta rel canonical.
Point at the one that
you want to have indexed,
and we'll pick it up from there.
"We're under manual action
for unnatural links.
We eliminated 96% of
all unnatural links
and disavowed the remaining 4%.
Google rejected our last
request and provided us
three links whose domains
were already disavowed.
What's up there?"
I don't know.
I'd have to take a look.
So if you want, you
can send me that URL.
And I can take a look
with the Web Spam team
to see if there's
anything specific that we
can let you know about.
It's really hard to say without
taking a look at the details
there.
It might also be that you're
looking at the 100% unnatural
links that you
found are actually
just a part of the
unnatural links
that we consider
to be unnatural.
So that's something
where maybe it
makes sense to give
you some other example
URLs that you can focus on.
Maybe there's an
issue on our side
that we can pick up as well.
So having that URL
would really help.
'When optimizing for speed
and using PageSpeed Insights,
is there a number for
mobile and desktop
that we know we're good?
It's hard to get 100% on
both mobile and desktop.
What are the ranking
effects there?"
Essentially, we look at
mostly mobile friendliness
which is part of that--
what you can test
with the mobile-friendly test.
And past that, the
numbers that you
see there are essentially more
of a guidance for you with
regards to what you
could be focusing on,
where you should be
focusing your energies,
where we or our systems
see some issues that you
might be able to resolve.
It's not the case that you need
to get 100% for some sites.
It doesn't even make
sense to get 100 there
because you know you might
have a perfectly fast site.
But it's doing something
that our systems are
picking up and saying,
well, theoretically
this could be a bad practice.
Maybe you should consider
doing it in a different way.
But if you really
know that you're
doing it the right
way or in a way that's
more efficient for your
website, then maybe that's fine.
So I'd focus on the
mobile-friendly test primarily,
which tells you if it's
mobile-friendly or not.
And the rest is
essentially just working
to make sure that you have
a really high-quality,
fast, and efficient website.
BARUCH LABUNSKI: But passing
the 80s is a good idea, right?
Because it was--
JOHN MUELLER: As high
as you can get, yeah.
It's always a good
idea to aim high.
With these kind of
tests, I wouldn't
try to be a minimalist.
If you're going to work on
this, there a lot of things
that give you a big
jump in those metrics
without actually
requiring a lot of work.
So maybe you can just copy
and paste some cache and code
into your HT access file.
And it'll
automatically take care
of a lot of server-side
caching for you
with maybe 10 minutes of work.
So that's something
where I wouldn't
aim at a specific number,
but rather figure out
what works well for
site, where you can go,
how far you can take it with
a relevant amount of work.
BARUCH LABUNSKI: OK, thanks.
JOHN MUELLER: It's also worth
mentioning that PageSpeed
Insights has an API.
So if you want to check
your client sites,
for example, you
could probably set up
in a small script that actually
go through all of those sites
and tests them for you so that
you know which clients you
should be focusing on more
and where you could see,
well, I put them on
this cheap server,
and now the test says that
the website is really slow.
Maybe I should I talk to them
about moving to better server
pr a different host or whatever.
"If from one domain
going only a single link
from a full article,
but Googlebot
finds this link on
a couple of subpages
from the original article and
tags sub pages, for example,
does the algorithm
count this as one link
only in the calculations?"
So in Webmaster Tools, we would
show that as multiple links
so that we would probably
give you one of those links
as a sample in Webmaster Tools.
But we count all of the
ones that we actually find.
The thing to keep in mind there
just because it's from-- it's
counted as multiple
links doesn't
mean that it's
automatically seen as higher
quality or higher value.
We'll try to figure out
how we need to evaluate
those links individually.
It's not that we
would say, well, this
is three links
from this website,
therefore, it's three
times as strong.
"Are you working on a Panda
update or a data refresh?
It's been a long time.
Everyone's frustrated."
We are working on updates there.
I don't have any time
frames at the moment.
But I know the team
is working on that.
I know it's
frustrating if you've
worked a lot on your website to
actually clean up these issues.
The same applies to Penguin
as well where maybe you've
cleaned up a lot
of web spam issues,
and you're just waiting for
things to open up again.
And that's something
we're definitely
working on to update
that again and to make
it a little bit faster.
BARUCH LABUNSKI: So when
that update happens,
it's going to be a manual like
everybody is saying, right?
I mean, everybody knows that.
It's going to be a
manual update, not
an algorithm update.
But at the same
time, everybody's
thinking it's
running-- it's running
with all the other stuff.
JOHN MUELLER: Yeah,
I think that was
a bit confusing from our side.
So it's definitely an algorithm
that runs all the time.
But we have to update the
data for this specific case.
So that's something where it's
kind of a mixture of both.
And I think we
probably explained
that in a little bit
of a confusing way,
an people picked that up
and thought that, oh, it's
completely like this or it's
completely like that when
actually it's a mixture.
And we just want to
apologize for-- I
don't know-- confusing
people about something.
"How can I filter queries
by multiple words, phrases
in Webmaster Tools,
search analytics?
Also, are key words from
the secure Google search
included in those reports?"
So, first of all,
the easy question
is, yes, keywords from
the secure Google search
are included in those
search query reports.
It also includes things that
Analytics wouldn't be showing.
Where you'd see the "not
provided in Google Analytics,"
that would also be included in
the search analytics reports.
Filtering queries
by multiple phrases,
I think that's something
you'd have to do individually.
So if you have
multiple phrases, you'd
have to test them
individually and maybe
download the data as a CSV
file or for Google Spreadsheets
and combine it on your side.
It's not something
that we would say
you would be able to search
for this word or this word
or this word and just giving
one aggregated table, at least
not at the moment.
FEMALE SPEAKER:
Gotcha, thanks, John.
I was actually reading one of
the help texts on your site,
and it said it was plural
so that confused me.
I thought maybe there was
just something I was missing.
But I appreciate
the clarification.
JOHN MUELLER: Glad
to help there.
DON HALBERT: John, on
that same note [INAUDIBLE]
or I don't think it
was overly top secret.
You just said it that in the
search analytics [INAUDIBLE]
consideration local [INAUDIBLE].
JOHN MUELLER: I can hardly hear
you, Don, you keep breaking up.
Or maybe you can
type it in the chat.
DON HALBERT: Can you hear me?
JOHN MUELLER: Every now and
then I hear individual words.
DON HALBERT: It
should be working now.
JOHN MUELLER: OK.
DON HALBERT: OK, in
Search Analytics,
if it includes local [INAUDIBLE]
page one-- page one for 20--
the term "exact match" receives
22,000 searches per day.
So if you're on
page one, you would
expect that if Google
[INAUDIBLE] zero.
And when I talked to you
before, you said that it is true
if a site is punished
by an algorithm,
it can still rank in
the local results.
And I've seen that.
It pops up into
the local results,
which screws up my average.
But if you go over
the last 90 days,
it still shows an
average of page one.
And I get almost
no traffic from it.
So if that data is
accurate, what's happening?
JOHN MUELLER: So
probably what's happening
is we're not showing
your site often.
But when we are showing it,
we're showing it on page one.
And people maybe
aren't clicking through
to page three, or four,
or five or wherever
your site is showing up.
So we're not counting
those views as something
into the average.
So essentially we're
only counting the ones
that we actually show in search.
And if these are just individual
users who maybe are local
or have your site ranking
number one for other reasons
for personalization,
for example,
then that's something
where we count those ones
because that's-- page one is
where we're seeing your site.
And we wouldn't count the places
where theoretically your site
would also be ranking because
nobody's really looking
that far into the search
results to actually find those.
So from that point
of view, it's tricky
because we're showing
you what we actually
showed in search, which is I
think the most correct thing.
But it doesn't mean that this
is where your site is always
ranking.
So what I'd look at there is the
number of impressions and use
that as kind of a
qualifier for the actual
rankings that you're seeing.
So if you're seeing the
number of impressions
is really, really low and the
rankings is really, really
high, then chances
are we're just
occasionally showing your site.
But showing it in the higher
search results, but essentially
not a lot of people
are seeing that.
So we had that, I believe, on
our blog a couple weeks back
where we would rank for the
query 'Google.' And we would be
showing up for the query
'Google' for the blog,
which obviously is kind of
specific and not something that
people would generally want to
look at when they're looking
for the query 'Google.' And
that kind of made those Search
Analytics numbers look
really weird in that we had
to interpret what does
this actually mean?
We see a ton of impressions.
Nobody is clicking
on our blog anymore.
The clickthrough rate
overall is really miserable.
But when you look
at the details,
you see, well, this is because
our site was ranking where
it shouldn't be
ranking or accidentally
ranking in some way that
skews the numbers there.
So you have to work on
interpreting those numbers as
well instead of just taking
down as they show up.
DON HALBERT: OK,
thank you very much.
MIHAI APERGHIS: John,
I think Don was also
asking if like a website is
showing up in local results,
but not in the organic search
because it has been penalized
or whatever, would
that still show
in Webmaster Tools as
being on the first page
because local results are
just on the first page?
JOHN MUELLER: As far as
I know, they would, yes.
I would have to double check.
But as far as I know,
I think the last time
I checked with the team, they
definitely are included there.
MIHAI APERGHIS:
That's interesting.
So that means that
for certain queries,
page one might mean 15, 17
results because you have
the seven local and the 10.
JOHN MUELLER: It's
possible, yeah.
MIHAI APERGHIS: And
what if a website
appears both in
local and in organic?
JOHN MUELLER: We take
the-- what is it?
The average top position.
So if you appear once-- like
if your website appears once
with one URL at number
two and another time
with a different URL at number
seven or something like that,
we take the top URL
and average that
across the different queries.
MIHAI APERGHIS: What
if it's the main page
and you appear with the
main page in organic
and with the main page in local?
JOHN MUELLER: We take
the top ranking one.
MIHAI APERGHIS: OK.
That's interesting.
Have you always done that?
I don't think before Search
Analytics was launched
you were counting the local.
JOHN MUELLER: We we were
doing it different before.
But I think we changed,
not for Search Analytics,
but sometime before that where
in the way they were getting,
we would take the average
of the search results
and use that as
the ranking number.
And now we take the top one.
The main thing that changed
with Search Analytics
is on the impression side, where
previously if you were ranking
with two URLs on one
search results page,
we would count that
as two impressions.
And now with Search Analytics,
we count that as one impression
because your site is
showing one-- essentially
in one search query.
And that might be multiple URLs,
but it's one search query, so
one impression.
MIHAI APERGHIS: And is
there any possibility
that you would be able to
add a local filter sometime
in the future just
to see local results?
JOHN MUELLER: There's lots
of possibilities, sure.
I can definitely pass
that onto our team
to see what we can do there.
I think at the moment they
have some other things they
are working on.
So I'm sure you'll see some
interesting things coming out
in the near future.
We're also working on an API for
the Search Analytics feature.
So if any of you are interested
in giving that a spin,
then signing an NDA with us
and trying things out with API,
feel free to contact me as well.
I see someone waving.
BARUCH LABUNSKI:
How big is the NDA?
MIHAI APERGHIS: I worked
at using the Google Apps
Script to kind of parse the
data automatically using
all [INAUDIBLE].
So using the API will
be pretty interesting.
JOHN MUELLER: OK, yeah,
send me a quick note,
and I can put you on the list.
I don't know how soon that
will be ready for testing.
But I know they're
feverishly working on it
to make that more available.
MIHAI APERGHIS: Cool.
JOHN MUELLER: "We recently got
hit with your so-called Phantom
II update.
Our traffic went down.
We would like to ask
some help and advice
why our site got hit.
I started asking other
webmasters in the forum.
It didn't bring much help.
Our site is this URL."
So essentially this is something
where we're not-- I don't know,
calling this anything specific.
This is essentially just
a normal algorithm update
as we make them all the time.
And sometimes this is something
that affects more sites.
Sometimes it's something
that affects fewer sites.
But essentially, we're
working on trying
to increase the relevance and
the quality of search results,
and that's essentially
just a normal update
that was happening there,
nothing really specific.
So if you're seeing changes
within your site's traffic,
within the impressions that
are coming from Search,
I think that's something
where you can work
on your website in general.
For most cases,
it's not something
where there's any
technical issue
that you need to focus on.
You would see that in
Webmaster Tools, obviously.
But it's not the case that
there's this one line of HTML
that you need to swap
out these two letters,
and then suddenly, it'll
rank much higher again.
We do these algorithm
updates all the time.
And for some sites, it goes up.
For the other
sites, it goes down.
Our general goal is to make sure
that the quality of the search
results is really
as high as possible.
So if you would like to be
more visible in these kind
of updates, focusing on
your site, making it really
as best as it could possibly
be is always a good strategy.
I realize that's not an easy
answer and not something
that you can just
say-- take back
and say, OK, I will increase
the quality of my website by 10%
by Friday and hope
that'll change things.
It's more something
where you have
to take the long-term
approach and look
at for maybe a little bit,
take a little bit of distance
to look at your
site and figure out
what you could be doing
in general to increase
the quality overall.
Maybe there are things you
could do with your users, chat
with them to figure out where
maybe they're getting stuck,
what you could be doing in
general to really increase
the quality of your website.
"Does there exist
another algorithm
similar to Panda
or Penguin which
only works all the time now?
If yes, is there any possibility
to refresh before the big--
to recover before
the big refresh?"
So we have lots and
lots of algorithms
that are running all the time.
And some of them run frequently.
Some of them run
less frequently.
So I guess the answer is yes.
There are lots of
algorithms that
are similar to the existing
ones that are known.
And, yes, of course, you
can recover from algorithms
when they run again.
So it's-- I don't know.
There's nothing
exotic essentially
that I can announce to
you and say, well, we
have a new animal in
our farm that we'd
like to announce to
you at the moment.
"Having added the rel canonical
tag to some potential duplicate
pages that are now
pointing at a primary URL,
I've noticed that these two
pages are still indexed.
How long should I wait before
these pages are de-indexed.
Will they ever get de-indexed?"
The rel canonical
tag is something
that's on the page
itself, which means
we have to crawl and index those
pages to actually process that.
So in a sense, in
order for us to keep
using that rel
canonical tag, we have
to have that page on our index.
We have to know about that page.
We have to crawl it
from time to time
to see what's
actually on that page.
So if you do something
like a site query,
it's very possible
that you'll still
continue to see
these pages there
even though we're essentially
passing all the signals that
go to these pages to
your preferred canonical.
So this is something
where I wouldn't
focus on things like a
site query or blindly focus
on these individual
URLs, but instead assume
that when these pages crawled
and processed for indexing
and have a rel canonical,
then those signals actually
do get forwarded,
even if we still
index those pages individually.
So I wouldn't assume that
they'll drop out of the index
completely.
It's not something
that you artificially
need to push out of a index
by putting a no-index on it
or something like that.
They essentially won't show
up in normal search results
because we have
passed everything
to your preferred canonical.
But they might still show
up in something artificial
like a site query.
MALE SPEAKER 4: John,
I have a quick question
about disavowing.
JOHN MUELLER: Sure.
MALE SPEAKER 4: We've been hit,
as you know, multiple times
by negative SEO link attacks.
Do we need to ever
disavow a no-follow link?
JOHN MUELLER: No.
MALE SPEAKER 4: So
then, in other words,
no-follow links cannot hurt
us or any site in any way?
JOHN MUELLER: They don't
pass any page rank.
They don't pass
any signals there.
So you can essentially
ignore those completely.
MALE SPEAKER 4: Easy answer.
Thank you.
JOHN MUELLER: "How
can we help algorithm
to link brand mentions
with the domain?"
I think this goes back to the
question of if someone randomly
mentions your brand
in a text somewhere,
is that's something
we'd pick up as a link?
And the answer there is, no, we
don't pick that up as a link.
We don't pass any kind
of page rank to a website
just because someone
mentions your brand.
If you have a URL,
of course, that's
something that people can link
to if they want to do that.
But if someone has just
mentioned your brand,
that's not something we'd
pick up on as a link.
I imagine there might be
indirect effects where
if people are talking
about your brand,
then that's something that
might bring traffic indirectly
to your website, which
indirectly could bring
more links to your website.
But it's not the
case that we would
use that as a primary ranking
factor in saying people
are talking about this brand.
Therefore, it must
be a fantastic brand
to rank it number one.
MIHAI APERGHIS:
By the way, John,
do you still use that
maybe to understand more
about that specific website
because you associate the brand
with a certain website.
You see the brand mentioned
together with a certain topic
and maybe you use it
to understand that,
OK, so this brand is-- or this
website is about this or that?
Maybe, OK.
JOHN MUELLER: I could
imagine that maybe there
are some algorithms that do
something exotic like that.
But I don't think
that's anything
you would see any
kind of visible change
in search results from that.
It might help us to understand
things a little bit more.
But you're talking about
this tiny little thing
where we can't
really say this is
like a primary ranking
factor because it's such
a vague element around that.
It's like what do they mean
by mentioning your brand.
Are they saying that this
is a terrible website?
Are they saying this
is a great website?
How should we take
that into account?
It's really tricky to kind of--
BARUCH LABUNSKI: I was
listening to Mihai's question.
Are you guys taking
any from what
Yandex has done from
their experimenting,
removing the links and
bringing the links back.
I think we had a
really-- I don't
know if you remember
when [? Spock ?] was here
at that Hangout.
Yeah, so regarding
the links, are you
taking that as a
learning experience
for what they've done?
I mean, they brought
the links back
because the results
were not that great.
Does that help you guys?
JOHN MUELLER: I think they just
have a different search engine.
So it's not something
where we could say, well,
what they did in our
search results is something
that we need to do or
we need to learn from.
It's very different
search engines.
It's really hard for us to take
that as something-- some kind
of useful feedback for
algorithms for our engineers
where essentially what you're
saying is some other company
tried this specific setup.
And it didn't work for them.
So they went back
to a different one.
Does that mean no company
should try that specific set?
Probably not.
I mean, maybe there are ways
that it can be made to work.
Maybe there are
different configurations
where it does make sense.
I imagine if it didn't
make sense at all for them,
they wouldn't have tried it
out for such a long time.
So, obviously, some part had
some positive impact there.
But it's-- I don't know--
I think you can assume that
our engineers in the
Search Quality team,
they're not all playing pool.
They are actually trying
new stuff out all the time
and working on
improving the algorithm.
So some of that is
probably around links
and trying to understand links
better and thinking about ways
we can pick that up without
it easily being gamed
or without causing any
problems if someone
does accidentally game
it or game it on purpose.
I would totally expect
changes to keep happening.
Otherwise, if I was the
only person working here
at this company, that
would be awkward, I guess.
BARUCH LABUNSKI: But it was
a good learning experience,
I guess, for anyone that it's--
without links it cannot work,
right?
I mean, to really
deliver the 100%.
JOHN MUELLER: I don't know
if you could say that.
It's really hard to say.
I don't know what
specifically they changed
and what specifically
they were looking at.
From what I heard, it was just
a part of the search results.
It wasn't even the case
that they removed it
from all of the search results.
So I don't know.
It's always
interesting to see what
other people are doing there.
MALE SPEAKER 2: John, can
I ask you another question?
JOHN MUELLER: Sure.
MALE SPEAKER 2: We have
an on-page JavaScript
variable starting with the
URL column followed by an HTTP
and so on.
Google finds that
specific variable
within the JavaScript
on PageScript
and tries to use the
get method to crawl it
which returns a lot
of server errors
as we use the parse
method to render it.
How can we block specific
JavaScript syntaxes
from being crawled by
Google, keeping in mind
that these JavaScripts
are not separate resource
files to be blocked
through robots.txt there,
but they are their
own page elements?
JOHN MUELLER: You can't.
So essentially we'll
try to crawl these URLs
if we find them.
But if we see errors or any
problems from those URLs,
that's completely fine for us.
So that's something where in
general that's not something
you need to worry
about that our bot runs
across a bunch of things in
JavaScript that look like URLs.
We'll try them out.
And if they don't work,
that's completely fine.
That's not something
you need to block.
That's not something you need
to obfuscate in the JavaScript.
If those specific URLs are
problematic in that they cause
problems on your server,
maybe changing the JavaScript
so that it doesn't
recognize that makes sense.
Maybe using robots.txt
makes sense,
but essentially it's not
something where in general
if we run across any random
URL, we try it and see an error,
that that would be anything
that you'd need to worry about.
MALE SPEAKER 2: OK,
because the only way
we worried about eating
up our resources,
not because of the errors,
because Google doesn't
show them in Webmaster Tools.
But we have a bunch of
them, because in each page,
Google tries to do the same.
So it's eating up
of our resources.
So we will try to
rewrite that syntax.
JOHN MUELLER: That's
something that
often helps because
we try to look
for these links in various ways.
We probably have like
these very simple link
searchers that essentially
look for anything
that looks like a URL
and try to crawl that.
And that's probably what
we're picking up on there.
And the more complicated ones
where we render the JavaScript
and try to see what the content
is, what the links are actually
being shown to the user, that's
probably working out well
because it's leading to links
that are actually working.
So if you essentially
just obfuscate
that link a little
bit in JavaScript,
then chances are we'll
pick up something
maybe slightly different
and try to crawl that.
But if it's something that
leads to an obvious 404
and doesn't cause
problems on your side,
maybe that's preferred for you.
MALE SPEAKER 2: OK,
thank you very much.
JOHN MUELLER: All
right, let's grab
some more of the questions that
were submitted since people
took time to send those in.
"Some of our mobile pages
have an accordion menu,
and the content appears when
you press the accordion menu.
Is that an OK practice?
Or is that similar to
hidden text on the desktop?"
That's essentially fine.
A lot of sites use this as
a way to make their content
mobile-friendly or make
the menu mobile-friendly.
That's definitely not
something I'd worry about.
"This may have been
answered already.
But is it possible
to set up filters
to only show user
sessions and page views
from Google image results?"
So the sessions will be
something you probably
see more in Analytics.
In Webmaster Tools,
you can limit
it to get results if you want
to look at those specifically.
And that you should be
able to see the queries
you should be able to see the
pages from image search results
separately in Webmaster
Tools in the Search Analytics
feature at least.
"I have added a new client
to my Webmaster Tools page
and immediately got flagged
for security issues.
I've gone through the server,
and everything looks good.
I clicked Security Issues.
And it all looks good.
Now, how can I remove the flag?"
If this is something that's
visible in Webmaster Tools,
then generally that
would be either malware
or your site got hacked
in a specific way.
And if you cleaned
that up, then you
can send a review request in
general in Webmaster Tools.
But there are other issues
that we essentially just
pick up when we
recrawl the pages
and we see, well, this
hacking is removed now.
So we can essentially
remove that.
And sometimes there's
just a lag between what
we remove in search
for where we say, well,
this is no longer hacked.
This is normal.
And what we show
in Webmaster Tools
where maybe there's a delay
of a day or so at most.
So that's something where maybe
you're just seeing that delay,
and essentially
everything is OK now.
If you're seeing this for a
longer time where Webmaster
Tools is flagging something
as a security issue
that you can't actually
request a review for,
then maybe that'd
be something you
could post about in the forum.
And we can pick it up
from there and see what
actually is happening there.
Maybe there's something
else on your site
that's still being
flagged that you could
work on cleaning up as well.
"Will Google take manual
or algorithmic penalty
for links acquiring only for
brand or brandless keywords,
assume all branded links
are follow attributes
and follow-- and links from
sites, nature of content,
or similar to the main site."
We try to recognize
unnatural links
regardless of the anchor text.
So that's something
where just because it's
branded anchor text,
it doesn't necessarily
make that link natural, right?
So we try to apply a
variety of different methods
to figure out which
ones are natural,
which ones are unnatural links.
"Does Googlebot understand
and index web pages
built with Polymer
without any issues?"
I'd say without any issues
is probably exaggerating.
But we are pretty good
at crawling and indexing
JavaScript content.
And if you have your
pages set up in a way
that there are URLs that
we can pick up for indexing
and the content is crawlable,
so meaning the JavaScript is
crawlable, the server
responses are crawlable,
then we do a pretty good
job of being able to index
JavaScript-based sites.
There was recently a blog post
I think on "Search Engine Land"
that reviewed a bunch of
our JavaScript indexing
capabilities.
And it actually
ended up pretty good,
I imagine it'll
continue to get better.
So if you're using Polymer or
other JavaScript frameworks
to build sites,
that's something where
you can test that out on your
site to see how far we can go.
And if you're
still seeing issues
that we're not
picking up everything,
I'd love to get feedback
in the Help forums.
And what you can
also do, of course,
is the Java Ajax crawling scheme
to make those pages crawlable
by default if you
wanted to do that.
"Since March 25th, our search
traffic has dropped by 70%.
We have lots of pages
indexed in Google.
And I'm sure users like them.
We tried several actions to
solve this without success.
Can you tell us what's wrong?"
I probably would have to take
a look at this in detail.
My recommendation in general
for something like this
is to also get help
in the Help Forum.
So kind of make sure
that you're covering
all the technical issues by
double checking with peers
in the Help forums.
But also make sure that
the quality of your site
is really the highest
it can possibly be.
And get feedback from peers,
even if it's harsh feedback,
even if it's something
you don't necessarily
want to hear because sometimes
that's really good feedback.
So that's kind of
direction I'd head there.
"Here's something specific
to a set of sites.
When searching for
a site in Google IE,
we see a lot of
our UK page titles.
And the company's also called Le
Boat in the UK and Emerald Star
in Ireland.
Why does Google use UK
titles for IE domain?
The IE page titles should
just say Emerald Star."
In a case like
that, where you're
seeing a mix-up between
different versions,
I'd definitely take a look
at the hreflang markup
so that you can really tell
us which version of your page
is relevant in which
locales and let
us know that this is different
content, not exactly the same.
So if we can really crawl
these pages and index them
separately, we'll try to show
the right titles, right pages.
And with the hreflang markup,
we can get that better
for individual
countries as well.
So that's something
I'd definitely
take a look at there.
"Is there any difference
ratios if we use
capital letters in the URLs?"
Which is better?
Small letters or
capital letters?"
You can use whatever you want.
Keep in mind that most of the
internet is case sensitive.
So we're also case sensitive
on our site with Googlebot.
So if you pick one upper or
lowercase version for a URL,
make sure you use
it consistently
within your website and
not that you link once
with a capitalized version
once with a lowercase version.
Really pick one
of those versions
and stay with it so
that we can consistently
call, crawl and index the
same URL over and over.
All right, just a few
minutes left so that.
let's switch back to you all.
MIHAI APERGHIS: I'm
going to go ahead and ask
a website-specific question.
I talked about, I
think two Hangouts ago,
of a client of mine who has
multiple websites regarding
ISO standards.
They offer implementation and
training for any specific ISO,
or at least some of them.
And they have a website
for each ISO standard.
And the idea is to combine
them into a single brand
because they've already
acquired quite a few websites.
So it's now time to best
combine them in a single site.
The problem is that while
we can do the redirect,
the new website will have
each ISO as a subdirectory.
And I know that Google
doesn't-- Google Webmaster Tools
doesn't offer on option
for change of address
to a subdirectory specifically.
So what do you think how we
should deal with that best?
JOHN MUELLER: Just redirect.
MIHAI APERGHIS: Because in
the example in the Help page
for the change of address,
they only give the example--
you cannot move it
from example.com
to example.com/directory.
But they don't mention
example.com to another
example.com/directory.
So I am guessing that's
not possible either.
JOHN MUELLER:
Yeah, I don't think
that's possible [INAUDIBLE].
I would use page-by-page
redirects for a case like that.
MIHAI APERGHIS: And just let
Google figure everything out.
JOHN MUELLER: Yeah.
MIHAI APERGHIS: OK, so Webmaster
Tools is out of the question
when it comes to a subdirectory.
JOHN MUELLER: Especially
when you're combining sites,
that's not something where
Webmaster Tools would really
help.
It really makes sense
from our point of view.
You're moving everything from
one side to another side.
And you're essentially
replacing the destination site
with a new one.
So in those cases, the
Webmasters Tools setting helps.
But for combining things,
it's not really right.
MIHAI APERGHIS:
OK, and should we
add subdirectories
as new entities
in Webmaster Tools for each?
JOHN MUELLER: If
you want to, sure.
That's up to you.
What you'll see is things like
Search Analytics is separated
then, which might
make sense for you.
MIHAI APERGHIS: OK,
yeah, that would.
OK , thanks.
JOHN MUELLER: Sure.
MALE SPEAKER 2: John,
another old question.
How much weight,
if any, does Google
put on HTML elements which
help visually impaired visitors
navigate the website, like
let's say link title attributes.
I have posted in
the chat an example.
Maybe you can take a look
to see exactly what I mean.
I'm interested if Google
gives any weights to this.
JOHN MUELLER: I don't
think we use that at all.
I think at the moment
we use the Alt text.
But we don't use the link title.
MALE SPEAKER 2: I understand.
If there are any other--
using rank, anything
who helps visually
impaired visitors or not?
JOHN MUELLER: Well,
text on the page?
I mean, if it's-- yeah,
if it's text on the page,
then that's something that
works well for screen readers,
and that's something we
can definitely pick up on.
If it's just like a link
attribute like that,
I don't think we'd pick up on
that because if the content is
really only in there,
then for the most part,
users, when they
go to that page,
they won't see it by default.
It'll just be visible there
if you hover over it or if
you have a screen reader that
reads that out.
So that's something where
if its text on the page,
we can definitely use
that for everything.
MALE SPEAKER 2: I understand.
OK, thank you.
MALE SPEAKER 3: John, can I
ask one followup to ours again?
JOHN MUELLER: All right.
MALE SPEAKER 3: If we-- now that
we've safely moved and changed
the HTML and done
everything, is there any risk
based on our previous
situation to us starting
to contact the old-- any
sites that are linked to us
and ask to link to the new site?
Or is the links the thing
that might cause an issue?
JOHN MUELLER: I think
that would be fine, yeah.
MALE SPEAKER 3:
OK, because we've
got a load of old
very high ranking.
I mean, obviously a load of
the crap we'll just ignore.
It will be a natural
disavow effectively.
So that's fine.
JOHN MUELLER: That sounds
like a good idea, yeah.
MALE SPEAKER 3: OK, thank you.
JOHN MUELLER: All right, with
that, let's take a break here.
I'd just like to point
out the Hangout on Friday
will be with someone
from the Webmaster team
from Google who has worked on
lot of the responsive sites
here.
So if you have any questions
about responsive web design
or if you want to talk
about responsive web design,
be sure to join
that Hangout and be
sure to submit your
questions there.
If you've submitted
questions for Friday that
are about ranking and
changes in our algorithms,
then you probably
want to move those
to one of the future Hangouts
that I'll set up as well.
So I wish you all a great week.
Thank you all for joining again.
And thanks for all the questions
that were submitted here
live in the Hangout.
Hopefully we'll see you again
in one of the future Hangouts.
MALE SPEAKER 4: Thanks, John.
MIHAI APERGHIS: Thanks, John.
MALE SPEAKER 2: Thank you,
John, have a good day.
JOHN MUELLER: Bye, everyone.