It's all about speed – no one wants to wait for a page to load – and Google has been saying for ages that they want a super fast internet.
Basically Page Speed can be simply stated as “the amount of time that it takes for a webpage to load.”
Sounds simple, but getting a page loading fast takes a lot of work.
This includes having a fast host / server, optimizing page sizes and images and ctilizing a great CDN.
Google has put together a channel called SEO Mythbusting. This helps webmasters find out what Google Wants.
Below is a video direct from Google about page speed Below this is a transcript so you can dig in either further.
In the third episode of SEO Mythbusting season 2, Martin Splitt (Developer Advocate, Google) and Eric Enge (General Manager of Digital, Perficient) discuss the most common SEO questions and myths around page speed.
MARTIN SPLITT: What
do you think are
misconceptions about page
speeds and especially page
speed and ranking?
ERIC ENGE: Well, a
lot of people think
that it's a big ranking factor.
In fact, I was literally
looking at a document
that a company had produced.
This document actually
talked about SEO,
and it had a section
on SEO which is good.
At least they're
thinking about it.
But the first thing they
listed was page speed.
And they were actually quite
insistent in the write up
that it was the most
important ranking factor.
MARTIN SPLITT: Oh, no.
ERIC ENGE: And I was like, OK.
I've got to find the
right way to tell them
that I want them to deal
with this because it's
really important.
And it clearly impacts user
engagement and conversion.
No, it doesn't mean
you're going to move up
three spots in the results.
MARTIN SPLITT: Right.
Yeah.
[MUSIC PLAYING]
MARTIN SPLITT: Hello and
welcome to another episode
of SEO Mythbusting.
With me today is Eric Enge.
And would you like to
introduce yourself?
Because you're
doing so much stuff.
What is it that you're doing?
ERIC ENGE: Well, you
know, I'm General Manager
of part of the digital marketing
team at Proficient Digital.
And altogether, we do SEO,
content creation, content
marketing, pay per click,
analytics, conversion rate
optimization–
MARTIN SPLITT: Trainings,
Twitter, conference speaking.
ERIC ENGE: Yeah.
That's a fair amount of
stuff to keep us busy.
MARTIN SPLITT: A fair amount
of stuff to keep us busy.
But today we're going to get
busy talking about page speed.
ERIC ENGE: It's a great topic.
Because so many
people get it wrong.
MARTIN SPLITT: Oh. yeah.
It's quite a deep topic as well.
ERIC ENGE: Yes.
MARTIN SPLITT: So
what kind of questions
do you have around ranking,
factor, trade speed,
and page speed in general?
ERIC ENGE: So let's
actually start in general
and just talk about why
page speed is important.
How's that sound?
MARTIN SPLITT: Sounds fantastic.
I think if you
look at what you're
trying to accomplish
as you're trying
to accomplish that,
you're building
a good website for your users.
Right?
ERIC ENGE: Right.
MARTIN SPLITT: So
now, how many times
have you been on the metro
or in the car or somewhere
in the countryside where
you didn't have fantastic
reception on your mobile phone?
And you were basically just
like really quickly trying
to find something out and it
just took ages for the content
to actually show up.
That's painful, isn't it?
ERIC ENGE: It is painful.
MARTIN SPLITT: And in
fact, on some sites that
can happen when you are
in a place where you've
a perfectly strong signal.
MARTIN SPLITT: That's
actually true Yeah.
Yeah.
ERIC ENGE: And that's not–
that's so frustrating.
MARTIN SPLITT: Right.
And you don't want to
frustrate your users.
ERIC ENGE: Right.
MARTIN SPLITT: And
we as a search engine
do not want to have
users frustrated
when they see content.
So for us it makes sense
to consider fast web
sites a little more helpful to
the users than very slow web
sites.
Right?
ERIC ENGE: It does make sense.
And I guess my thought
process in this
has always been
that well, yes, it's
likely that you're using at
some levels a ranking factor.
But you can't make it such
a strong ranking factor
that you won't show the
most relevant content.
MARTIN SPLITT: Oh, yeah.
Absolutely.
If you have bad content,
if you are the fastest
website out there but
the content is not great,
then that's not helping you.
ERIC ENGE: Right.
Right.
I mean, to get the content
you don't want quickly
is probably not what
the user's looking for.
MARTIN SPLITT: Exactly.
Like, I have a blank website.
It's the fastest website ever.
What's the point?
ERIC ENGE: Yeah.
Well, yes.
Exactly.
But it does make
sense to consider it
at least at some level.
And there's actually a
fun pair of statistics I
think they're both from Google.
One is that something
like 53% of sessions
are abandoned if it takes longer
than three seconds for the page
to load.
And then the companion
statistic is,
and I think it's
a little bit old
but still, the average page
takes 15.3 seconds to load.
What a frightening combination.
MARTIN SPLITT: It
is frightening.
It's frightening.
And it's so many
different factors.
Right?
Sometimes it's slow servers.
But sometimes it's just like the
server responds really quickly
but then there's a
ton of JavaScript
that has to be processed first.
And JavaScript is a
very expensive resource
because it has to be fully
downloaded and then parsed
and then executed.
But, yeah.
So we keep seeing this.
And everyone knows this.
Anecdotal evidence
is there as well.
You have studies.
You have the anecdotal evidence
of you sitting in front
of a website going like, ugh.
And Just imagine being
on a metered connection
where you actually pay
by megabyte when you fly
or something.
It's like you can
buy 20 megabytes
for 10 euros or something.
And you're like, oh, OK.
Open one website.
You said, what
was it 15 megabyte
is the average or something?
ERIC ENGE: Well 15.3 seconds
is what I'm was talking about.
MARTIN SPLITT: Oh,
sorry, 15.3 seconds.
So you can just
imagine how much data
you were pulling in
these 15 seconds.
ERIC ENGE: Yeah.
In fact, I did see–
I really was looking
at this just yesterday.
There is this data
from Think with Google
where by market sector it shows
the average web page size.
And they're all in the
megabytes in every market.
And I think your recommendation
is 500 k-bytes or less,
if I'm not mistaken.
MARTIN SPLITT: Yeah,
the fewer, the better.
The fewer, the better, really.
And just think about it.
Like I grew up with entire
video games on like two or three
floppy disks which each fit
like a megabyte and a half
or something.
So why are we doing
this on the web now?
ERIC ENGE: Hm.
What a great idea.
Well, maybe we should help
people speed their sites up.
What do you think?
MARTIN SPLITT: That's the thing.
And that's why
ranking these by speed
is also an important factor.
But as you say, like
content still is king.
Like there's no
question about that.
ERIC ENGE: Right.
Absolutely.
MARTIN SPLITT: How do you think
people are thinking about page
speed as a ranking factors?
Like what are they
trying to do when they
are trying to optimize for it?
ERIC ENGE: Well, in terms
of what they try to do,
I think there's a few
things that people
are really good at thinking
about related to page speed.
So I think almost everybody
recognizes that images
are a potential issue.
And certainly,
pre-sizing the image
rather than making the
browser do it, for example,
and things like that.
And so they get to that
first level of optimization.
But I think there's
other things that they
find a lot more
difficult. So for example,
the idea of not loading
the content below the fold
until the content above
the fold is present,
of course, that's a little
harder to implement.
MARTIN SPLITT: We have native
lazy loading images for now.
So that's something, at least.
ERIC ENGE: Yes.
It is something.
And then I think
another thing that they
have trouble with is–
and you actually mentioned
it a moment ago–
the idea that the way you're
hosted and the way your CDN
is set up can be big factors
if those aren't actually
set up properly.
First of all, they
might not have the CDN.
But they may have
it, and it may not
be properly configured as well.
MARTIN SPLITT: Configured
with caching and stuff.
We've seen all of this.
ERIC ENGE: Yeah, exactly.
And then it could be as
simple as, I need more memory
in my web server.
Or a dedicated server, when I'm
on a shared server connection.
MARTIN SPLITT: All of
that sounds pretty solid.
But is there any
misconceptions or myths
that are going
around where like,
what's happening here, where is
this coming from, is that true?
ERIC ENGE: So I do
think, and maybe I
could state the myth almost
as an inverse, is they
are too focused on just a
few surface level factors.
And they don't realize there are
other layers to this problem.
MARTIN SPLITT: There's
layers to this, yes.
ERIC ENGE: Although
there's another thing
I can suggest actually
as a myth, if you will.
Which is if I go into and
get my Lighthouse tools
report on a page, and I see
it says, oh, this will cut six
seconds out of the load time.
And then they do that
thing and the page
didn't speed up by six seconds.
And I don't think people realize
that some of these things
are threaded.
MARTIN SPLITT: Oh, yeah.
ERIC ENGE: So yes, I
did something good.
But I have four other problems
that also need to be fixed.
MARTIN SPLITT: Yes.
ERIC ENGE: So I do see a lot
of people getting tripped up
on that.
MARTIN SPLITT: That's
an interesting one.
Yeah, and Lighthouse is a
tricky one to begin with.
Because people are getting
confused by the idea
that what they are
seeing in Lighthouse
is what users are seeing.
And that's not the case.
Because you are literally
testing from your machine,
from your browser, from
your internet connection,
and not necessarily what real
people are experiencing when
they're on their mobile phones,
on their spotty connection
out there.
So I think it's
important to remember
Lighthouse is lab data.
And it makes predictions
on what you can improve.
But that doesn't necessarily
mean that, oh, now you're
all doing fine.
Do you also think that people
are paying too much attention
to the scores itself?
Because I hear that quite a lot.
So like the myth
is like, oh, we're
using the Lighthouse
score for ranking.
That's not happening.
That's not what we're doing.
ERIC ENGE: Right.
No.
Exactly.
In fact, they get too
attached to that score.
And sometimes it
misleads them to thinking
that they are doing just
fine when they actually still
have problems.
MARTIN SPLITT: Yeah.
ERIC ENGE: And another area that
I see people running into is it
works fine from my
phone, but the user
doesn't have such a nice phone.
So you have to remember that
there's different devices.
MARTIN SPLITT: And you could
see that in Google Analytics.
You can actually figure
out what kind of devices
you were seeing on your site.
And then you can
specifically try
to understand–
best way would be
to buy one of the phones that
is most prevalent on your site.
ERIC ENGE: Yes.
MARTIN SPLITT: And
[? I can ?] have a look.
ERIC ENGE: A very
interesting idea.
And I actually shared a slide
in one of my presentations
recently which
showed data actually
for CNN.com processing.
And it was around three seconds
for the high speed phone.
But by the time you get
to a user with a less
than $100 phone, it
was 15 seconds to load.
And you just really
need to remember
that the users have all
different [INAUDIBLE] devices.
And you probably
want to do a good job
by the great majority of them.
MARTIN SPLITT: Absolutely.
And you want to be aware that a
slow phone on a slow connection
is like the worst situation
you can probably run
into in this kind of situation.
And you can use things
like web page test
to get a better feeling
for how that would feel.
Like you can test from different
locations and different network
connections.
I would definitely
recommend doing that.
There's so much more
that you can do.
And also, if you
have a website that
is listed in Chrome
User Experience Report
or [? CRUX, ?] then
definitely use that as well.
And I think not many
people are trying that out.
ERIC ENGE: Right.
Well, it's good to
get real world data.
MARTIN SPLITT: Real world data,
real user metrics, absolutely.
ERIC ENGE: Yeah.
Absolutely.
In fact, you could broaden
that piece of advice
well beyond the page being
conversation, by the way.
Like it relates to all
manner of aspects and things
around mobile, for
example, because we
have everybody who
designs for a desktop
and then has to slam that down
into a mobile phone format.
Maybe designed for the
mobile and then it's
kind of easy to figure out how
to run [INAUDIBLE] desktop.
MARTIN SPLITT: Exactly.
You have more
[INAUDIBLE] so yeah.
ERIC ENGE: Yeah.
Exactly.
But for the page speed
conversation, absolutely.
You just have to do that.
MARTIN SPLITT: Definitely.
Right?
And yeah, I mean, it's
such an important thing.
And people– do you remember
the entire controversy on people
like AMP is a ranking factor?
ERIC ENGE: Oh my.
Yes.
MARTIN SPLITT: It's not.
And then people
are like, but, it–
and page speed pastes
into that as well.
Right?
AMP gives you a
certain expectation
that you can have for your
sites in such results.
And I've seen good fast
web sites rank higher
than the end equivalent.
So like, maybe it is not the
most important ranking factor.
But it's definitely
an important one
as in like page speed is an
important ranking factor.
AMP, not so much.
AMP is just like
this little batch
that gives the
user an expectation
that they can have about it.
But page speed does
matter for your users.
And it does matter for your
conversions, as you said.
Sometimes it's configuring
your CDN– getting a CDN,
configuring your
CDN, making sure
that caching is done
right, and making sure
that you architect your
websites and web apps in the way
that they are fast by default.
If you can do that without
AMP, then that's fantastic.
AMP is a fantastic
tool kit to help you
do that if you don't know how.
ERIC ENGE: Yeah.
And you could go with
Progressive Web Apps
as well, by the way, which
are very nice because
of their ability
to preload content
into the cache on your phone.
So by the time the
user requests the page,
it's [? continuous ?]
[INAUDIBLE]..
MARTIN SPLITT: Yeah.
That's true.
ERIC ENGE: And it's
another way to skin a cat.
No.
I'm not supposed to say that.
Because that's really
uncomfortable for cats.
MARTIN SPLITT: It's really
uncomfortable for cats.
ERIC ENGE: So take it
the way I meant it.
MARTIN SPLITT: I get it.
I get it.
So anything else around page
speed where you're like,
what's happening there?
Any questions you
have on page speed?
ERIC ENGE: I mean,
really, I guess
it's reasonable to
presume that there's not
any prospect of Google
dialing up that ranking notch.
It's basically, you're kind
of set with what you've done.
I mean I know that algorithms
change all the time.
MARTIN SPLITT: Algorithms
change all the time.
ERIC ENGE: But just from
the logic perspective,
the issue that we talked about
already between the relevance
of the content being–
well, content being king.
It's still going to be king.
MARTIN SPLITT: Absolutely.
Absolutely.
ERIC ENGE: Have to
deliver the right result.
MARTIN SPLITT: You want
the relevant content first.
ERIC ENGE: If you had
five right results
and maybe it nudges
something up.
MARTIN SPLITT: Like if you have
two results that are basically
doing fine content
wise, we would probably
get the one that is faster,
more prominence in the search
results.
And also, I think it's important
to understand that we're not
doing it by Score or Lighthouse
or something like that.
It is more we're
bucketing pages into
like this is a
programmatically slow one.
This is an OK one.
And this is a fast one.
You see that in the speed report
as well, in the Search Console.
So I think people
need to just like
figure out if they have
really slows pages,
how to make them faster.
And probably if they're
in the middle bit,
you also want to
go to the fast bit.
But it doesn't
matter if you have
a Lighthouse score of 90 or 95.
That doesn't really
make a difference.
All right, Eric.
Thank you so much for being
here and talking all things page
speed with me.
That was amazing.
And I hope that everyone liked
it and leave comments and likes
with us.
And thank you very much.
ERIC ENGE: Hope
you all enjoyed it.
MARTIN SPLITT: Bye.
Hey, everyone.
So next episode is going to be
with my fantastic guest Rachel
Costello.
And Rachel, what have
you brought for us?
RACHEL COSTELLO:
We're going to be
talking about canonicalization
and URL de-duplication.
MARTIN SPLITT:
Sounds really cool.
Don't miss it.
RACHEL COSTELLO: See you then.