Rand Fishkin - How Marketers Can Keep up with Google in 2016

Rand Fishkin – How Marketers Can Keep up with Google in 2016

Hello everybody, welcome back, I hope that
you enjoyed your coffee, because I didn’t have any time to get one. It’ so exciting for me to introduce Rand
Fishkin. When I started SEO in 2006-2007, SEOmoz was one of the best
inspiration sources of mine even whiteboard Fridays or friend articles,
he is of the best influencers in our industry and I’m really
proud to announce him to speak at Digitalzone- thank you so much Rand
for answering our invitation with a positive response, I really
appreciate that you accepted. Ladies and gentlemen Rand Fishkin please come to the stage.
Thank you so much, I think it's very impressive.. It’s incredibly impressive to see what you've
built here, you should be so proud that you're able to
put together a company, an agency that serves clients, a conference that
attracts hundreds of people and amazing speakers, amazing speakers and me.
We have a tremendous amount of information to get through today, so I'm going
to go quickly, but I've made sure to get the slides up ‘bit.ly /keepup2017’
I'll have this URL at the end of the presentation as well, in case you
need to get it you can also find it on Twitter, I put it on there as well.
We know that in the last two years Google has made a tremendous amount of
changes, I will admit the vast majority of these changes and the first mover
parts of Google always happen in English and they usually happen in the United
States or another English language market- that's because Google started in
California and Google's engineers, at least the majority of the most
experienced ones speak English as their first language and so we often see the
changes first and then they roll out to the rest of the world and that can be a
little frustrating as Judith pointed out in her presentation as well.
In the last two years we know about 17 major named or observed algorithm updates
there surely have been thousands more, we've also seen this move to machine learning
first, from google as an organization not just in their search engine we're seeing
the end of the 10 blue links era, certainly in the English language
markets and then rolling out to more and more countries as well.
We are seeing AdWords take away data like they're the
CIA that you know the American assholes who do all sorts of terrible things all
over the world. We're also seeing PPC growth leveling off at least in the
United States, this is the not that PPC is going down but that the growth
rate is finally leveling off and I don't think Google has a tremendous amount
more ability to increase what PPC can drive in terms of clicks- at least not on
desktop, maybe in mobile. Google's fighting back against this
trend of PPC by making the ads more and more subtle,
but what's the difference between this and this, it's
that little green (I'm gonna jump as high as I can, this is pathetic) but that right
there, that's all, that's the only thing that separates these two. What are they
going to do to make it more subtle in the future? I can't really
imagine it, only to have it say ‘A’? One of the things that Google has been
doing as they're taking away all this data is forcing marketers, software
makers, like Moz to go and get data from other places and one of the places
that we can get data from is Clickstream sources. Are you folks generally
familiar with Clickstream data? This is essentially lots of applications, plugins,
certain browsers, certain antivirus softwares, when they install on your
desktop machine or your mobile device they track all the clicks that you make
through your browser and then they anonymize that data and they sell it in
aggregate format to companies like Moz. We're able to buy this clickstream
data and we can see awesome things like keyword length of queries, this
is using clickstream data, this is for the United States only, but we can see
the percent distribution of keywords. You can see that one, two, three word
queries are still the most popular but the distribution has changed a lot from
the days of 2006-2007 when you saw vastly more in the three word, four
word and fewer in the six plus. We also know what search demand looks like, so I
created this cool search demand curve where you can see that the keywords that
are most popular, the million keywords that are most popular right here
represent only six percent, so if you took the top million keywords and nobody
ever searched for those you would still have ninety-four percent of all search
volume, that's pretty insane. For people on google, that long tail is still
enormous, it takes to the top 10 billion keywords,
before you get into that range. On desktop, at least the
average searcher is still doing about three searches per day. I think what's
actually happening as most people are doing one or two, but then there's some
people who are just crazy with searches- like when you have to go on vacation you
do a million searches in a day. We're also seeing some different distributions
of paid clicks. This is going to sound crazy, this is US data again, but on
Google.com (I know you, you want to film a
whiteboard Friday later? Thanks man you rock) You can see that this number
is mental, out of all the searches on Google only 1.19 percent
of searches that Google takes result in a click on an ad,
that means 99% of all the searches on Google bring Google no money. It's a pretty
amazing business model when you think about it. 51% of all the
clicks that do happen on Google search results go to organic, just the
10 blue links style, classic blue links. For the rest of this
stuff, 49% is going to.. well some of its going to Adwords, some
of its going to Google itself, some is going to maps to YouTube, to
other places and in fact lots and lots of searches result in no click at all.
On average 40% of all the searches that happen
on Google- no click. no one clicks any results. Because Google essentially
answers the query for them, by the way this type of query, weather queries, we
looked at these specifically, about 75% of all weather searches result in 0
click at all. I actually think that's pretty good, I thought it would have been
even higher. Search suggests in a lot of these cases we think is probably
biasing some of these behaviors. About 25% of all the
queries that come through desktop are coming through Chrome instant and many
of those that may load a result, are of course
resulting in no click, but did the searcher really mean to search for that
or did it just load as someone was typing slowly? It’s hard to say for certain.
In the US, these are the top 20 sites that receive traffic
from Google in order, so YouTube is getting the most traffic. It used to be
Wikipedia. Facebook, Amazon. Facebook is really
interesting because there's not actually that many pages on Facebook that rank
that well and drive traffic, it's just that so many people go to Google,
search Facebook and then click Facebook- because they don't know how to type in
the URL. I don't know
what number 18 is, that sounds real weird to me. This is the distribution,
we talked about distribution of clicks across traffic, this is distribution from
websites. You can see that there's actually a tremendous amount of bias to
the head of the website world. There’s a Moz scape, our web index crawls
many hundreds of millions of domains, we think maybe on the order of
three or four hundred million domains that are sort of active and real and
useful on the web and then there's a ton more that are spam or sketchy or have been
abandoned. You can see that the top ten thousand websites get 62%
of all the traffic. As an SEO, you better be here, you cannot be
below this, that's not going to work. This is a ton of change
that's happened in the last couple of years. What does it mean for us,
what are we going to do about it? I think there's five big things
that I'm going to talk about today, that we as marketers can do
actively to change our habits and our behaviors in order to better serve this
new world of Google. First one, we need to diversify where we get traffic from,
we cannot be purely Google reliant, it is too dangerous. I'll show you what I'm
talking about. Google knows where traffic flows on the web, they have a
Google Chrome, they have the new Google Hub for Wi-Fi, they have android, so
they know where a good 70-80% of all web visits are going.
And of course they can buy data from other places if they want to see the rest of
it. I think that we should be savvy to the fact that Google is going to know
where traffic starts, where it visits, how it gets there and where they end up.
If Google knows that your site is getting the vast overwhelming majority
of its traffic from Google themselves, that might be a sketchy signal to them.
That could be seriously problematic, I would be very scared
of a website that's getting 90%, 85% of traffic
from Google alone. This is Similar Web, Similar Web obviously collects a ton of
clickstream data, they're based in Israel, but they do a phenomenal job of getting
browser data for tens of millions of web users- then they look at where
traffic comes from, where does it go. These are the sources that send traffic
on the web. If you take a look here search is big, I mean Google is
like 95% of this number, they're around 24%
of all traffic. But if you're getting 85% or 90% of your
traffic and Google is only 24% of all web traffic, that's a
little sketchy. I would really look at direct, I would look at social, I would
look at referring links, type in, that's important stuff. This is YouTube,
YouTube and Facebook are kind of neck and neck on video,
in terms of where traffic's going on the web. But when it
comes to video, Facebook counts a video view at three seconds- ‘1,2,3, you scroll too slow that means we get a video view for Facebook’. For, YouTube you have to watch for 30 seconds before they count it as a view. For Facebook I would ignore that
number. YouTube are dominating in video, they're also the world's second
largest search engine. If you are thinking about where else should I be
getting traffic from- probably YouTube. DuckDuckGo at least the United States is
continuing some rapid growth, outside not so much. Amazon in much of Western Europe,
in South America, Central America, North America, doing very well, but obviously
here in Turkey, not as well. My advice would be, don't ignore a search
channel just because it's not Google, if you are a local small business
you might want to pay attention to something like Zomato or Four Square or
Yelp. You can look at where your competition gets its traffic,
this is inside of similar web pros platform and you can see here that I'm
looking at Warby Parker, they sell glasses in the United States.
They are getting a lot from direct Google search, Facebook, Shareasale.com, Lifehacker,
Pinterest, Reddit, these are all places I could consider going after. If my
competition is getting traffic from there, I should too. This is what a lot of
us do in link building research, but we should be thinking about it from a
traffic perspective too. I think it's important that we apply different
tactics to different sites. If you are targeting Amazon because you're in
the United Kingdom or the United States or France or Germany
where Amazon is very big and you're selling products, you can find Amazon's
ranking factors. I've actually linked to them here, so check out what's going on
there. I would also strongly encourage you to consider different content for
different engines. This is something unique that Moz does, that not a lot of
other folks do, but we have a method to the madness. Does anyone in the room
watch Whiteboard Friday? Okay a few of you.
Basically I wear ridiculous shirts like this one and then
stand in front of a camera like that one and film a video about some SEO topic.
Here's what we do, we will take the video and we will put it on our own website
first, then three months later we will upload that
same video to YouTube. Why? Because we want to rank for it in Google, we don't
want YouTube to rank for it in Google, we want to make sure we own the traffic,
especially those first 90 days, then we put it on YouTube because we know that
there's lots of search volume on YouTube itself- so we have both pieces
of content. Now you might say ‘wait is that duplicate content?’ no it's not.
Even if it was, YouTube is the one that's the copy, so I don't
care if google filters out Youtube. In fact, Google you can go ahead and throw a
penalty at YouTube, we've got a lot of duplicate content, I think you should
take a careful look at what YouTube is doing, there’s some sketchy SEOs working on that
site. Number four, we need to evolve our keyword targeting so
that we match Google's sophistication. Judith I very much appreciate your
presentation which teed me up perfectly for what I wanted to talk about.
This is a look at these instant answers or the featured snippet, so you can see
if I asked Google- do polarized sunglasses actually make a difference, do
I need to spend the extra hundred dollars to get my lenses polarized or
ten dollars whatever it is. You can see it on desktop, that's taking up a lot of
room. Google is actually doing this interesting thing where they're
minimizing the desktop search experience, they've made it this one column
layout so that it matches what you see on mobile and they're putting more and
more of these answers. This is a pathetic excuse for an answer, but
answers on top of the results and that is trying to help them provide you with
a very fast and quick experience on Google, especially on mobile but now on
desktop as well so that you don't get frustrated when having to go visit a site.
Of course it also takes away some traffic from us. At Moz we expected
that when Google put this answer up for ‘how long should title tags be’, because we
ranked number one for this, we thought when
Google put that up there it answered the question and we would lose traffic.
You know what happened, we actually gained traffic, we got more visits than we did
just ranking organically number one. Ranking number zero and number one,
beautiful. Number zero is terrific, that's not true for everything though.
Some of these answers like this carousel, it has taken a tremendous amount of
traffic away from all of these sites and we've talked to a number of the players
in here and they have been really hurt, especially in travel by Google's
carousel results. This is more like a 50% loss. When you are
choosing your keywords, I agree with Judith you should think about volume, you
should certainly think about intent, search for intent and topics. You
should think about competition, but you better think about click-through rate as
well because targeting Seattle Washington sites or targeting high-tech
gadgets, that is a very different click through rate that you are going to
earn than you would for a result that gives you 10 blue links.
100% of clicks on this result go to the
organic results. On this one, 60. This is again from Clickstream data, we
know which features appear, we can look at the averages of the click-through
rates and then get that. You can make these estimates yourself, you
can go in, you can say this is my Excel spreadsheet and it's
got my volume and my difficulty and I’m going to estimate my own click through
rate opportunity. Or you can use a tool like here at explorer which will do this
for you. I expect that many other tools like the ones Judith talked about will
soon use Clickstream data that they buy and will integrate this and then have
their own click through rate scores, so that we can all choose the right
keywords based on potential click-through rate. It is possible
to use SEO to get into those featured snippets. If you want to learn a
lot more about that Dr. Pete has a couple of great blog posts and I didn't
have time to put in this presentation but Rob Bucci from Stat in Vancouver
gave a great presentation at Search Love and at Mozcon.
You can check him out online. Another big problem in the data world
for keyword choice is volume. We know there's a bunch of weirdness in
here. When I say ‘weirdness’, what I mean is ‘dirty lies’.
These numbers that you see in average monthly searches, they are ranges, there
is no chance that 12,100 people searched for champagne
flutes, that is not what happened. What it actually means is a range. Russ
Jones from Moz did a bunch of research, looked at a bunch of actual adword
impression data, compared it to the google volume numbers and then showed
what those ranges actually mean. You can see those distributed in
his blog post if you're interested. We have generally found google trends to be-
I don't want to say more accurate. When Google Trends says something is
bigger than another thing, it usually is bigger than that other thing. Accurate is
not correct because you can see there's no axis on the left, meaning we
don't know what number Google is telling us.
AdWords is doing this conflating of volume numbers for similar keywords.
That's super frustrating, it has taken away our ability to see which
are the right keywords. We also know keyword planner in AdWords is hiding
lots of data, look at this with me- I just searched for fitness tracking tools
‘no ideas were returned for your query’. That doesn't seem right to me at all, Because look, I just searched in google and there's a whole
bunch of suggest ideas, then I take all those suggestions and I plug them into
AdWords. It looks like they have volume for these ‘dirty lies’.
The ‘no ideas’ thing doesn't mean anything and in fact there's lots
of times when they'll show you ideas and they have way more ideas that they
wouldn't show you that you have to find on your own. I would
be careful about using these volume numbers as comparisons, I would be very
careful about using AdWords as your only keyword source. I think trends is okay
for volume comparison and if you want some of that
clickstream data you can use a tool here at Explorer or many
other ones which will be coming out with similar stuff.
If you want the very best data about volume and conversions
and trends- you have to pay. You buy the
keyword, you see what the actual trend is, you see what the volume is, you
know what the actual impression count is, you know whether it converts-
this is no surprise. If you want the best data, this is
the way to go. If you are relying on keyword planner exclusively
let me strongly encourage you to try a few different sources.
These are suggest and related, they can certainly assist you, they’re free.
There's seven kinds of keyword research generally out there, amongst the
different tools- search suggest obviously. Answer the public basically
scrapes- I think only search suggests actually, not people search.
Many of the other tools will also include ‘people also search for’.
‘Similar pages rank for’, this is what SEM is brilliant at, absolutely
the best in that field. You can also check out Key Compete and Spyfu, they
do similar semantically connected terms- these would be things that
are related to each other, like the synonyms, topically related searches and
questions containing the keywords. Answer the Public, when they scrape
search suggests they always use the answer format, ‘where why when how’.
AdWords, which are very commercially focused but they do have
their own suggestions. This next slide, I feel dirty showing it because it
makes Moz look really good, but it's not actually the case. There's a
bunch of different tools and the discovery options that they have.
I think many of these will have all of them soon and certainly I expect SEM Rush, they are
the best at number three by a long shot. Number three- when we are creating content and
matching that content to the keywords we know we want to target, we need to
change our mindset to one that considers searcher intent very carefully and
considers SERP features very thoughtfully. One of my
favorite types of snippets is this disaster type snippet, the
nor'easter in there. Our data suggests- this Moz cast attracts all
the different kinds of snippets that you will see in Google. This is us again, but
you can track these in Semrush Stat, which will also show you SERP
features in any country that you do tracking on and so will Moz campaigns.
This data suggests that out of all the search results we track, only 3%
look like that. 97% have some
features of some kind, SERPS like these with lots of knowledge graphs,
‘people also ask’, videos and news results.
Those can hugely affect your click-through rate. We're estimating
for nor'easter storm here, probably 24% click-through rate on organic SERPS.
Frustratingly, if you want to do video, it used to be the case that you
could put video on your own site and use the video rich snippet markup to appear
in Google like YouTube does, but not anymore. Basically Google has restricted
this to two websites- it's just Vimeo and Youtube.
That seems dirty to me, but on mobile even more kinds of searches are limited to
particular networks. If you're on iOS, you've got to be in the
the Apple Store. If you want to be in Google android, you need to be in the
Google Play Store. You can’t rank in those boxes on your own network at
all. If you see this happening in your search
results and it almost certainly is, you should analyze it. You should go
and get a distribution like- ‘here's all the keywords I care
about’ and you tell me which ones show which features and how often. Because if I see
that lots of things are showing news or lots of things are showing site links or
images and I'm getting very few results with shopping, well great, now I know
which types of content I need to create and which types of markup I need to use- this
this is real opportunity. That will help me determine which verticals and Serps I
need to optimize for. If I see a surfer salt like this, it probably makes
a lot of sense, especially because these are terrible,
so I bet I could create an image that would rock the crap out of this and
I bet I could get a higher click-through rate than any of those other images.
I bet I could outrank all of these by getting that
higher engagement rate. One of the things that we have noticed with image search
results is the engagement rate. Google doesn't like
to say they use click through rate but they totally use it, in a
complex way where it's not direct. If you can get an image in here that
drives more clicks than any of these other ones through image results, you
will generally do a pretty good job of rank.
If there's certain searches that are impossible to target, my suggestion is
that you try and influence search suggest. This is challenging but
if your business life depends on it, like a lot of sightseeing tours and those
types- if your business life depends on it there are ways to
influence search suggest. They use both the text and anchor text that appears on
the web and they use search patterns. If you do advertising or branding
campaigns that can influence how people search, you can influence search suggest
as well. This is legit. You want to choose the right
platform for your search results, this is not a comprehensive list but if you have
content of these kinds you should appear in these places right. I'm local I
need to be in Google Maps, I need to be an Apple Maps, I need to be in Big Macs.
If I’m doing podcasts I better be in itunes, I better be lips-in, I
better be on Google Play, I better be in Soundcloud. If I'm doing e-commerce
google shopping is essential. All those things. We know too that Google is
getting masterful at understanding search intent. I did a search
‘what's that book about Jews living in Alaska?’, it is the Yiddish Policemen's
Union by Micheal Chabon. This doesn't mention any of those words, none of
those things appear in there, Google just knows what I'm looking for.
Here's another one ‘what's that famous circle of big rocks?’- oh yeah Stone Edge.
Very impressive, they know what I'm looking for. ‘What's the place with all
the food stalls in Seattle’?- Pike Place Market. ‘What about that big underground
science wheel in Switzerland?’- the Large Hadron Collider. They know what I
mean, that is meaning and intent not keyword matching. Keyword matching
is no longer a true competitive advantage. If you saw backlink, they did a
great research study using a number of folks including Moz and A aetra to put
together this new ranking factor study and they noticed that keyword
matching is very low on the scale of things that in general
correlate to higher rankings. That's not to say you shouldn't use keywords, you
absolutely should, you'd be crazy not to. If you know that lots of people are
searching for something, you should probably target that something and you
should talk about it in the same way the searching audience talks
about it. But, if you expect that to help you rank higher, you're out of your mind.
That just gives you a chance to rank. On page SEO is just not satisfied by
raw keyword use, one of the things that we have noticed that does strongly
correlate and that was in the backlink on ranking factors and in search metrics
new ranking factors is related terms and phrases. Those do
correlate with higher rankings. For example,
if I wanted to rank in here for the keyword growth hackers, I should probably
be using words like- marketers, startup growth, growth hacking,
Quora, Airbnb. Airbnb is a very frequent example used by folks who
talk about growth hacking. Just like if I wanted to rank for
Istanbul neighborhoods, I should probably use -here's my pronunciation challenge
‘Sisli?’. Don't make a page about Istanbul
neighborhoods that doesn't include the word Sisli.
It's still really smart to use keywords, you can see that if you want to be in
the results, you should be there. But matching your content to what the
searchers are seeking, that is critical, keyword matching is not
going to get you there. My on-page SEO list for 2016 and beyond are intelligent
keyword use in all the places we've done so in the past, use of smart related
topics, serving those keywords with matching intent together on one page,
meaning if someone searches for Istanbul neighborhoods and Istanbul city
districts, those should probably be on the same page because they have the same
searcher intent. Thorough answers, I need to make sure that no one clicks the
back button and chooses a different result because they didn't get their
query satisfied by what I was providing. And finally, I want to be able to provide
unique value beyond what everyone else in the search results is
currently providing. You can check this out in my Whiteboard Friday topic. We still need links,
links help us rank tremendously, we have talked a lot about the
decreasing value of keyword not that you can stop using them but just that
they won't help you rank any higher. It’s still kind of the wild
west of links. They are still hugely important and used by Google. If
you're going to be a great link builder in 2016 and beyond, you need a strategic
roadmap that looks like this, you define your goals, you take a strategic approach
to it, you define tactical initiatives, specific tactics you're going to pursue
based on that strategy and then you have metrics to measure your success.
I'll give you an example, this is book surf, they're based here in
Istanbul, I actually got to visit with their founder yesterday, Kerim. I'm going to like this website a lot I
love books, I love mustaches, what's not to like? Book Surf is this
Istanbul-based book sharing community, you lend books out
to folks so they can borrow them. If Book Surf wants to rank
better for all the book titles that they have in stock, so that when people search
from Istanbul in turkey for these different books- if they want to
rank better this might be a long-term link strategy. Their linked goal is
they want to rank for those individual book names, especially in Turkey itself.
Strategic approach, they're going to encourage their members to link to their
profiles and their favorite books. They're going to partner with retailers
and they're going to earn links from those retailer sites too. The
tactical initiative is that we're going to nudge people when they set up an account
to link to us from their sites, to share us on social, we're going to do direct
outreach to the authors and to the retailers as well. Our metrics are
going to be, links remembers, links and
mentions from the authors and retailers sites, ranking for key terms and search
visits. This is what that strategic process looks like
in practice and we're going to need a few other things, we need by in on
experimentation. If you want to try a new tactic you've got to have a
client or a team who's willing to try new things, you've got to have the right
expectations regarding the time that it's going to
take. Link building is a slow manual process, it is not something where
you have overnight success, very rarely will you have overnight success.
So you need that buy-in on time to ROI and you've got to have a balance too
between long-term investments and short-term hacks. Because there are hacks
that will help here. Long term investments, they have high upfront costs,
they're slow to show ROI, but they earn links while you sleep and there's a
virtually non-existent spam risk. They're naturally coming in
because of what you're doing. With short term you tend to pay in time
or in dollars to someone as you go. They can show ROI very quickly, effort in
means links out, so it doesn't scale with decreasing friction and it can have some
spam risk. What I see is great link builders focus on their flywheel.
Flywheel basically is this giant wheel concept- it takes a lot of
energy to get the flywheel started, a tremendous amount energy to get started.
But once it starts going momentum carries it and friction decreases every
revolution of that wheel. So Moz looks like this. For Moz’s flywheel
essentially we do keyword research, we get some information out the
industry, we publish content on that front, we push it to our email and RSS
subscribers, we promote via our social channels, we earn links and amplification
from that. Now we can grow those channels, so the next time we push out content
it's easier to reach that audience. We grow our domain authority, we earn search
traffic and now we can rank for more and more competitive keywords and the
process starts all over again. Take a journey back with me, if you visited Moz
in 2006, which was a couple years after I started the blog, I was lucky if any
piece of new content I pushed out led to one link. But in 2016
pretty much every single post we publish earns links, because that flywheel has
scaled and turned and now it's going so fast we barely
have to do anything except hit publish. It’s kind of cheating but that's
how a flywheel works. For Moz- our business is built on
content. For someone like Dollar Shave Club, they're built on press and media.
For a site like Dribble they're built on user-generated content and communities,
sort of like how I talked about Book Surf might be. Almost every flywheel that
you'll find out there will encounter friction, meaning that some part of that
equation doesn't work well and that is when hacks work brilliantly. Growth
hacks can help you accelerate that flywheel, they're not necessarily evil or
spammy or bad, they can be really useful when you pair them with the right
strategy. There's tons of hacks out there, I'll encourage all of you to go
find the link building hacks that work for you but I think this is the formula
that's going to work for years to come in link building. Last one, we talked a
little bit about Google using engagement and I’ll use engagement to refer to
everything that is around that, so that could mean clicks, it could mean return,
bounce backs, pogo sticking, returning to the search results. I think
Rank Brain’s the most obvious of Google's machine language based ranking
elements. There are many other elements in Google that use ML at
least as a training technique and possibly in the live algo itself or the
core algo. Rank Brain is helping Google know which results to return, based on
the interpretation of the query itself. All these queries probably
share the same search intent and therefore should return very similar
results- and google knows this. How are they doing this?
Well, here's Paul Haar from Google, one of the engineers, they are talking about it.
I hate bullet points in slide decks. I apologize, I’m
taking Paul's slide deck here. You can see that basically Paul is
saying ‘hey we're interpreting query success from how users engage and
interact with our search results’. That’s pretty important because that means all of us
need to do a phenomenal job of satisfying searcher intent and satisfying searcher
completion. Here's Jeff Dean, he's a fellow at Google,
one of their very senior engineers and he's telling us…
he did not actually say this, I just put these words in his mouth, so sorry Jeff
Dean. This is a good search
result because when people search ‘learn to code’, they tend to
click things in order. When they get to these sites they tend to
complete their query, they’re happy and satisfied, they don't come back, they
don't go to page two, they are happy searchers. This is a bad search result.
What happens when people search for this? They scroll really far
down, they go to page 2, page 3, they're not satisfied, they click a
result and then they click back because it didn't answer the query for them- bad
search result. If anybody thinks that pogo sticking is not being
measured, check it out- this was just launched a few weeks ago, when you
click a result and then you click back to Google they now show you this. They're
like ‘oh no we didn't satisfy you searcher, please try these other queries’.
At the very least we know they're measuring it and according to
Paul Haar, they're using it. Someday I think algorithms that are built by
machines will outperform the ones that are created by hand selection by the
engineers. We will take these potential ranking factors and
the training data and we won't even need the ranking factors to be fed
in by engineers, the machines will figure that out for themselves. And through a
machine learning process they will find the algorithm that works.
Google says they're using these machine learning elements without really knowing
how it works. I think as long as it
produces good results why do they care, why do we care? We might find that in
the future these ranking elements that we've always had with us are not
universally applied. We know that Google is going this direction, they've
been public about it, it's not going to be long. So what we have to do as
marketers is focus on that signal to noise ratio. We can't let bad pages drag
down a good site. So you can take a simple equation
like this- the quantity of pages that are earning visits from google, divided by the
engagement they get. Relative time on
site bouncing back to the search results, that gives you some type of
engagement reputation. Google just recently through a
hangout with John Mueller mentioned that google has these
site-wide signals that they apply to websites. It certainly seems to
me that engagement could be one of those potential site wide signals. If you
see that you have pages that are performing poorly in Google, you should
probably investigate them. I mean pages that are performing
poorly for Google's searchers. You want to find ways to beat your competitions
click-through rate, so that you're showing Google that you are better at this
than anybody else. You ranked number three but you have a higher
click-through rate than most of the other folks in position two and one-
Google is going to move you up. Over time they are going to figure out ways to
move you up. We've got to serve multiple search or intents here, not just
our own interests. If I’m selling cooking equipment, but I know that
people searching for sous-vide cooking are not buying, they are not ready to buy,
I better serve their intent, not my ecommerce purchase goals. None of
these Serps are trying to sell equipment
directly, they are all providing information. I couldn't find
anything until page three that tried to sell me anything. My last piece of advice
is if you want higher engagement, if you want people to stop clicking the back
button, make user experience a cornerstone of your SEO. And that means
authoritative comprehensive content that provides unique value versus what
anybody else out there does. It better be fast, because we know that
engagement is based on speed it. It needs to deliver that easy enjoyable experience
on every device at every connection speed. It has to compel the visitors who
see it to engage with it, to share it and to come back.
We want those people to be loyal to us and showing Google that they're loyal
will create a higher engagement score. Then finally we want to avoid anything
that's going to dissuade people or annoy them. You do this and you get two
remarkable 10x content that is insanely hard for your competitors to catch you
on. Once you start earning that Google and SEO, that's a place where
the rich get richer. So yes, it is true, there is massive change that's
happened in the last couple of years but I think with the right tactics and with
a smart approach to SEO we can take it on. Thank you very much. I think that we can take some questions. Hey Rand,
apparently I'm going try to milk questions from all of the speakers today.
You talked about link goals and defining a strategy, finding the
right tactics and measuring with the keep A matrix, but in the link goals you
mentioned that we need to get some links. Do you compare it to the
competitors, do you try to figure out how many links, what type of links, authority
and things like that? Also to estimate how long it will take before we
get those links and how big of a budget the client needs?
I'll be honest, generally I don't do that upfront work to try and estimate
that Delta. I think that the reason I don't is because the degree of
digging in and trying to figure out here's all the link types and here's the
different counts from every different type and here's the overlap
between places- they have links in places we don't. I actually feel like the broad
metrics of quantity and quality which you can get from
Moz or if you use a aa tref, you can get those in the count of linking
domains, the DA PA. Domain authority, page authority or for a HS that's called
Domain Rank. Then for majestic it's
trust flow and citation flow, those are generally good enough. You
could dig in, you could spend a bunch of time, but that doesn't help your client.
The only thing that helps you or your boss or your team or your manager- the only
thing it helps you do is estimate a little bit better what the work
involved is. If that's critical to your team or your client then go ahead, but I
don't love spending all that time trying to figure out all those metrics. I want to ask about
classic SEO, Youtube Search Optimization and App
Store optimization. They are very new concepts and our clients are asking us about
them, but right now there's nothing to do.
What do you think about that? I would disagree a little bit. I think that a few
of the very classic SEO practices still apply in YouTube and in App Search.
It turns out Youtube and apps are places where keyword matching is still
very powerful. So if you call your video something very precise, you'll
probably outrank someone who has higher engagement and more views on a video
that is not as precisely named as the keyword search. Google is a bit more
old school, very specific keyword matching on Youtube and they're even
more that way on app search, at least in most places. Not everything, but
they assume that every query you enter into app search and
Youtube has this very key word specific intent. Now the next piece of
those is that both of those rankings that we've seen are based on
engagement, but it's a different kind of engagement. So on YouTube a lot of stuff
is based on video views, how far people made it through the video and how
many of the people who view the video made it through that far. You'll also see
that embeds of YouTube videos and links to Youtube videos have some influence
there. I encourage you to check out a great blog post from Phil Nottingham,
who was with Distilled and is now with Wistia. He did a blog post all about how
YouTube tends to rank videos. For app search you can do the same thing, there's a post
on the Moz blog from the folks at Tune and I think his name's Ian Zephyrnick.
He did a great blog post about all the app ranking factors, but again engagement
is very high, so the number of installs, the ratings that people get, all that kind of
stuff. Do be awarel, when it comes to app search, people search for brands, they
search for the name of apps in the app world only very rarely do they actually
search for ‘I want a travel app that helps me find stores’ nobody
searches that. Rhey search for Rripadvisor or Yelp or Zomato.
I was about the same question because in 2014 you had a
question like this and I remember the answer was extremely similar.
You said that brand search is still dominant. And I was about to ask whether
you are still continuing to think in that way? Yes but with one exception.
The one exception is when you're on a mobile device and you
are searching Google.com rather than the App Store, the apps that Google will show
you tend to be much broader. So for example, if you search for ‘graphing
calculator’ on your mobile, Google will show you apps ahead of
the organic results, that have graphing calculator features. Or if you search for
‘travel app’ or if you search for ‘tower defense games’ they'll show you
apps that they think match that intent. So that's the one caveat to app search
being so brand focused. You typed circular
monolithic sculpture on Google lists, is it just in English or other
languages like Turkish as well? Unfortunately my Turkish searching is
not great, but I'm sure you could probably answer that question
better than I could. But we have seen them get much better at it in a lot of
Western European languages, so Italian, French, Spanish, German
and I think Dutch as well. I have heard that it's pretty poor in the
Scandinavian countries, so in Swedish and Norwegian it's not so good. I think
it's one of those things where they need a lot of data to learn and then they
need engineers to put time against it country by country, but how have you seen it in Turkish?
I don't have any exact experiments that I have done before but
I think it’s good. It’s working pretty well? Here's a question,
can you in Turkish do a search for me, the Turkish equivalent of
‘tall building where man tried to fly Istanbul’. If it's smart it should
bring back Galata Tower. So if not, then you know it has some
work to do. The results are not impressive, it's like the tallest
buildings and Wikipedia. Wikipedia – Hezarfen Ahmet Çelebi – how he flew over Galata tower. It’s not as good. Maybe we can train it.
Thank you guys, thank you so much for having me I really
appreciate it, cheers.

Leave a Reply

Your email address will not be published. Required fields are marked *