welcome everybody to another YouTube
hangout super excited to be here with a
new friend someone where we rescheduled
to get back on because he’s got some
amazing content ideas to share with us
mr. Matt Ellen he is an expert on
YouTube and figuring out the algorithm
and we’re going to talk for the next
half hour about how to dissect it and
help your YouTube channels grow so Matt
welcome aboard man thanks for having me
man if you want to give a little context
little background for the audience’s so
they know what you’re all about
sure my name is Matt Gillan and I’m
formerly the VP of programming and
audience development at a place called
Frederator you may have seen some of our
stuff before we made Adventure Time in
Fairly OddParents and a whole host of
other cartoons that were extremely
popular on TV I got to work on the the
digital side focusing on building the
audience for cartoon hangover which had
two cartoons bravest warriors and being
puppycat a whole host of really awesome
shorts on there and then I oversaw the
programming an audience development for
a channel frederator 22 we did really
well there in about a year and a half we
built that channel up to about 1.1 1.2
million subscribers doing anywhere
between you know I around 8 and 17
million monthly views and then um I also
helped build the channel Frederator
Network love it so you put out this
detailed report I forget where it was
published it was in one of the blogs
talking about the YouTube algorithm and
I think I reached out to you on Twitter
afterwards and we had a little bit of
discussion back and forth about it
what are what are some of the things
that you discover the people haven’t
read the article and we should we should
link that up so people can dive deeper
if they want but what are some of the
things that you discovered that people
aren’t doing that that could really help
the youtube channels grow um things that
people aren’t doing that would help
their YouTube channels grow I would say
is continue look at your content and
recognizing if there is content that’s
actually hurting the viewership on your
channel um you know we found out as
evidence to suggest I don’t think it
necessarily
Sara we proved it outright I don’t
necessarily forget outright but it
suggested that uploading a video that
got less than 3% of your subscribers to
watch in the first 48 hours could
actually cause viewership on your entire
channel to drop including your library
the videos and we see this most commonly
like in channels that kind of go into
like a bit of a death spiral where
either you know people aren’t watching
the newest videos or maybe their
contents gotten a little stale and one
is up happening is that their library of
videos begins losing traffic from like
suggested videos and search and they
start dropping out and so this suggested
to us that there’s there’s kind of like
a meta score or meta algorithm that’s
basically like if you’re kidding these
combination of factors we’re going to
rank all of your videos and our
discovery algorithms higher and so
that’s like things like you know browse
and suggested and recommended and
related you know in all the various
places those show up most importantly on
the home page and on watch pages and you
know if you keep uploading those you
know quote-unquote negative uploads it
can contain to your channel is at what
we saw and conversely if you’re
regularly doing you know 10 12 15 % of
your subscribers in the first 40 hours
you’ll see your channel really really
begin to grow to a certain extent so
those are some of the higher-level
findings of the of the article
we also found really cool stuff about
like duration you know literally videos
that were just longer right so whether
that video is 15 or 20 minutes or 25
minutes made a huge difference in terms
of viewership and then kind of as a
subcategory to that the audience
retention on those videos also had an
impact on how think that’s news to
anyone I think this was the first time
anyone could’ve put out any sort of like
data behind that that showed you know
when we had a video that we get a 12 or
13 percent average view duration we
would get like 2x are subscribed
years in terms of overall viewership so
you know we had about a million
subscribers when we did this research
you know videos that would get two
million views we’re doing about a
hundred and fifty a thousand views from
our subscribers in the first 48 hours
and the reason I use subscribers is a
number I don’t actually think the
algorithm really cares whether
subscribers or not I think it has to do
with a concept called view velocity
which is that they see this video
accruing so many views in the first four
hours YouTube says oh this video is
really important for the topics and
keywords with which it falls under and
the interest with which it falls under
so let’s serve this video to people with
that same interest and you’ll get a lot
of click-through and if that retention
the audience retention and click-through
stays the same as it expands it will
continue to expand it over time now I
think there’s probably stages we’re
going beyond the article now mainly
because this stuff fascinates me and I
can’t help but continue to think about
it um if you think about a video’s
lifespan and kind of like certain stages
right like you have your the first 48
hours window where YouTube is really
kind of analyzing your video and seeing
what it’s all about right its first and
foremost it’s taking in kind of the
metadata that you put in right and this
is everything from average video length
title description tags closed captions
and then it starts take into account
things like comments and who’s clicking
on it and what they’ve watched before
and so on and so forth right it starts
building that into it and there’s kind
of like this 48 72-hour window where
YouTube is basically saying you know
what is this video all about and how
good is it and by good it means really
like how good at is it at creating a lot
of watch time on our site right has
nothing to do with quality or anything
else it’s how good is it at Google on
our site now you can pull from that that
you know the better quote-unquote
quality video is going to generate that
but that’s a whole another conversation
for a whole other time and then outside
of that right there’s kind of like the
first week two weeks that you know the
initial two or three days kind of leads
to then after that is kind of a long
tail
the video and I think there’s different
kind of weightings at different times I
have to do a lot more research on it but
kind of what I’ve been seeing from like
a top-level I’m going oh that’s
something really interesting I want to
do the data on that
that’s what’s popping out at me cool and
so what’s interesting there too is
you’re saying that if if a new video
comes up doesn’t go well then that kind
if you can all your old videos – oh yeah
absolutely big time and it would we
should also take note that there’s
there’s always going to be an element of
like if one video that’s doing really
well there’s going to be a certain
percentage you know whether that’s you
know a fraction of a percent or 50 or 60
percent of the people that land on that
video to watch that are going to go and
watch your other videos right and you
know that obviously is going to have an
impact on viewership but what we saw
kind of more in the details is it’s more
like a circular effect where or like a
wave effect where the waves kind of
start building on to each other and
start overlapping right so if you start
having all these real successful videos
right your your graph starts to go I
think this is the right way on the
screen it’s a little confusing okay yeah
so like your graph starts going like
this with the ups and downs versus just
kind of like this right along with like
the uploads and it kind of build on each
other but that being said if we think
about kind of the meta algorithm or the
high level algorithm as like kind of a
channel score and say well if your
videos are regularly gaining more
audience than your subscribers that
means that your channel should have more
subscribers right because you’re
reaching beyond your audience from a
viewership perspective and I think
YouTube takes that into account and
similarly with watch time
now it could boil down to something as
simple as like well your channel is
generating a ton of stickiness for our
site so we want to send more people to
your channel so that they you know stay
on the site longer and so you know I
think I think it’s a combination of
factors but certainly keeping people on
the YouTube platform
you in to outperform yourself is going
to generate more audience interesting
let me ask you this I don’t know if you
have data on it or not uh for
subscribers who are watching your video
if you had to have one call the action
at the end of the video would it be to
leave a thumbs up leave a comment share
it
what would both what the what’s the
priority of engagement afterwards or do
you know it do have that figured out yet
and so you were saying if you had
already get like 15 20 25 percent of
your subscribers to watch what would you
ask them to do at the end of the video
yeah or even if not like even even you
only gain 3 percent or whatever it is
like what’s the highest impact thing
that a subscriber can do for you where
if your creator you have a lot of
creators who have like 20 calls actions
the end of a video as a result they’re
not going to anything right so you get
to pick one like hey a question of the
day leave it down below the comment or
please guys add this to a playlist or
give it a thumbs up does it make a
difference on the engagement to the
algorithm um I heard a saying recently
that I really loved which was tyranny of
choice where if you give if if you give
them you were too many choices they’re
gonna choose to do nothing uh and go
away right so I think a pretty good rule
of thumb is like no more than 3 calls to
actions so that they have an option
right it’s a you know pretty solid rule
of three right like you see that you
know throughout so many things in life
and so giving people three options I
think is probably the you know best
place to be I have not done any testing
on I don’t have data to back it up
I can say from a non scientific
standpoint that essentially when we
reduced our calls to action from I think
seven or eight to three we saw a better
performance we’re also making better
content but uh essentially in order of
importance routes say is watch more
first and foremost mainly because watch
time and keeping on people on the
platform longer is so incredibly
important and this kind of speaks the
programming strategy right like if
you’re making a video about LeBron James
you don’t want to then tell people to go
watch a video
know about you know horseback riding
maybe the people that are watching
LeBron James video probably don’t care
about horseback riding sure there’s a
small percentage but you know I don’t
think it’s a huge percentage right you
want to tell them to go watch another
video about LeBron James or if you don’t
have that tell them to go watch a video
about Kobe or you know Steph Curry or
whoever it is right you go from most
relevant to least relevant in your video
inventory from what video you ask them
to watch next the next thing I would say
is subscribe mainly because from from
that windowing perspective we talked
about earlier in a videos life cycle
those first 48 hours are so important
and that makes subscribers unbelievable
or right and so when you upload a new
video the first people that are likely
to see that video and click on that
video giving that video like a watch
time score giving that thumbnail like a
click-through rate are going to be your
subscribers right and they’re going to
be the best performers and so from that
perspective you know the more
subscribers you have right the greater
reach you have the more watch time the
more session duration time you’re able
to create the better is going to be for
your video and for your channel so
number two would be subscribe and then
number three would probably be some sort
of engagement I’ve seen no indication
that engagement has any impact on the
algorithm whatsoever
you knows if there was a in this day and
age historically yet it’s incredibly
important which makes me believe makes
me think that there might be still some
weighting going on in the algorithm and
then if you also think about it from a
more theoretical standpoint
videos that are engaging are likely to
accrue more watch time simply by the
fact that while someone is going to do
that action they’re still watching the
video right say hey comment down below
right well someone then has to think
about your question they have to think
about what to write right that’s
actually type it out alright chances are
well maybe not on YouTube but another
place is people will go back and like
check it for grammar and spelling you
know they’re going to want to add their
emojis not your audience I’m sure your
audience is good on that I know there’s
lots of them in the live chat doesn’t
mean lots of emojis happening
that takes time right it’s so like yeah
maybe videos with like a
disproportionate number of comments
right are racking up this incredible
watch time because people are literally
having to sit there while the video
plays and write out the comment in that
scenario it’s like well you know which
is it is the watch time or the comment
like well the watch doesn’t happen
without the comma so you know it’s some
sort of engagement action would be good
at the very least to show that you
actually care about what you’re your
subscribers and fans think and keep them
engaged by the show right so comments to
what happened someone who has a comment
somebody else responds then coming back
to the video if they come back to the
video and just respond to the comment
but without watching again because
they’ve already seen the video they just
want to be part of that thread could
actually check it out I should be a
negative signal um no I don’t think so I
think it’s it’s just engagement
generally like I don’t I don’t think I’m
sorry are you asking more about like if
a repeat view happens yes he’s like
hopefully think so on this video we’re
gonna get comments afterwards right
maybe I respond to one and say hey you
know Tamara thank you so much for
watching video she comes back to see my
comment and just to leave another
comment to say hey it was awesome but
she’s not gonna watch this whole
half-hour thing again because she’s seen
it if she ends her session there does
that negatively impact the video now
um I mean theoretically it could again I
don’t have any any data just yes one way
or the other
what I would say is that I know
historically YouTube has given algorithm
waiting to videos that generate repeat
viewing I think there’s probably a lot
of evidence in the kids space to suggest
that like oh yeah that can be really
powerful if they do repeat viewing so
from that perspective I would say you
know it’s probably toss of the coin
either way you know generally speaking
though I think if someone’s coming back
to stay engage in the conversation
that’s got to be a good thing right I’m
thinking now at the end of a lot of my
videos I’ll ask a question
the question today is whatever leave it
below but I’m never really doing hey
here are some related videos that you
guys might enjoy if you’re saying that’s
like the primary call to action should
be watch more than you know should i has
some selfish consulting here you know
should I change my strategy to like put
three links in the description say hey
guys here are three videos I think II
need to watch next go check them out in
the description below and not do a
question of the day well I think you
still do a question of the day but I
think yeah you should absolutely tell
people you know what they might enjoy
nests cuz like they think about it from
two perspectives right you ever think
about it from your audience who has
watched you know a lot of your videos or
many of your videos you have to think
about it from a new audience right
because a lot of people who watch that
video are not going to be people who are
a subscriber who may have ever seen your
video before right so it’s a way of
introducing them to your content letting
them know that you have a ton of other
videos up that they might really like
and then think about like well the
reason why people are on YouTube is to
be entertained and to discover things
and if you know you can help them do
that you know I think that’s really
valuable to the audience and yeah it’s
really helpful to you as well so you
know from that perspective absolutely
yeah I think so cool well the people in
the live chat are saying don’t stop the
question of the day please so keep that
for you guys okay don’t worry uh
otherwise you’re gonna get tons of hate
and tweet Matt come on man uh-huh so
okay so for an ounce of creators who are
looking to launch a new series on their
channel maybe them in one video week you
want to go to two videos a week uh the
metric you would look at the judge the
success of a new series would be
subscribers watching the video that’s a
tough one and and that’s a tough one
because almost no matter how much
promotion and marketing you do for the
new series right it’s not going to get
as many views as like your current
popular series mainly because in my view
it shows up in a subscribers feed like I
don’t
recognize this I don’t understand this I
am NOT going to click on this right this
is not something I usually watch your 10
thumbnails or something I usually watch
I would like to click on one of those
now there’s a whole bunch of strategy to
to get around that but to answer your
question about the success of a new
series I would focus on more about what
the average view duration is and if
you’re getting really good average view
duration like let’s say you’re 60 to 80
percent range which i think is pretty
damn good for longer videos if you’re in
that range great if you’re a little bit
below that range you might want to look
at like the structure of the video are
there ways to keep more people engage as
longers their way to get into the video
quicker that sort of thing but generally
speaking I think average view duration
is going to be the most like apples to
apples that you can do to your current
series in a way to kind of measure the
long-term viability of it
in addition um you know I wouldn’t I
wouldn’t cut a series until you’re at
least like five six seven episodes into
it if it’s something you care about and
are passionate about mainly because one
that’s how long it’s going to take the
build you know any sort of audience for
it out of out of your current subscriber
base but also it gives a good amount of
content for people to go back to and
check out and one go right like that
word YouTube video Genma five seven
minutes that’s not a huge duration of
entertainment time right where that is
if you have five or six episodes of that
well not you’re talking about like yeah
you know I could sit down eat a hoagie
and watch three or four these videos and
determine if I like it or not right like
do you see the Netflix study that came
out that was like you know Netflix knows
exactly how many episodes of a show you
need to watch to fall in love with that
show entry right yeah and like for
certain series it was like need to watch
three and you’re going to watch every
season for other shows it was like watch
seven or eight and so you know giving
yourself a good enough kind of data set
of headspace did a set of eight videos
to get a better feel for it I think is
kind of a good benchmark somewhere in
the 5-8 range so one of the things that
we’ve been looking at as we add a new
series in would
so subscribers gain per video maybe
that’s not a great metric but you know
our goal is for every video series we
put out we want every video to be in the
double digits of new subscribers joining
you know if it’s if it’s negative and
that’s a really bad series obviously but
if it’s like single digits five yep your
current audience right it doesn’t
necessarily mean it’s a bad series right
which is the prop right like you it
might be you you two doesn’t give a hoot
about any individual video all that
lunch there there are a lot of
exceptions to that rule but for the most
part any individual video is not all
that important to YouTube in the sense
of you might have the great give you an
example from a movie and TV channel we
had a Fredrik called cinematic a’
cinematic a’ we built about a hundred
thousand subscribers after like three or
four months or something we do have
three or four million monthly views and
we built that audience mainly by doing
like listicles and theory shows and this
kind of crap about like The Walking Dead
Game of Thrones Star Wars and Marvel
right and we wanted to be a more
well-rounded audience we wanted to bring
more females in so we did a listicle on
orange is the new black there’s a
perfectly fine video perfectly good
video nothing revolutionary nothing
crazy like it shouldn’t have done ten
million dues but it should have done way
more than like the four thousand views
that it did because the average video on
that time did a couple hundred thousand
views for us and so that was the first
indication I had that was like okay
didn’t really give a crap about how good
of an orange is the new black video that
is what big of a crap about is whether
the people that they served it to which
happen to be your subscribers right
because that’s who gets to see it first
clicked on it right if those people
didn’t click on it right they’re not
going to go and say well this is a great
orange is the new black video right and
they’re not going to give it a chance to
be a great orange is the new black video
maybe over time yeah you know five six
seven eight months it gets picked up
something happens to it but out of the
gate no right so that that’s part of
what led to this idea of like you have
to focus on this very specific niche and
so with that in mind that’s why I think
like a negative subscriber number is a
really hard way to judge a new series
because it’s it’s only about that series
in your audience right or I’m sorry it’s
only about how the very specific
audience that you currently have access
to interprets that video right and
whether or not they like it or it makes
them unsubscribe or whatever it is so
from from that perspective I would
really encourage you to to not not look
at that as a primary metric when when a
new series first goes up like give it
some chance let me fight for your
children okay I love it um I know I know
you know you are obsessed with data and
trends and frequent stuff out and
hacking your way what what’s the next
thing you’re looking at what are the
some of the things that are on your
radar that you’re exploring the next
blog post is going to come out and drop
some new knowledge bombs on um well I’ve
gotten really excited about a couple of
things so like one of the biggest
questions I see like when I’m trolling
on like you know our partnered YouTube
and our YouTube is like you know how do
I launch a channel or how do I get
audience and um you know it feels like
there’s no way to build a YouTube
channel just in these day and age and
all kind of stuff and like no that
definitely hurts me because I think
there is now I also understand that I’ve
had the benefit of working with existing
audiences right so like yeah we built
cartoon hangover more or less from
scratch but we’re kind of riding on the
coattails of the creator of Adventure
Time right there was a bacon audience
there for bravest warriors which led us
to our first you know a couple hundred
thousand subs and then you know when we
started launching Channel Frederator we
leaned on cartoon hangover and we’re
able to send audience over but that said
you know I’ve helped channels launch and
over the last like a year or two and
I’ve seen them become extremely
successful and I think I’m gonna start
my next big one would probably be that
you know how to launch a channel in 2017
right and drive audience the the second
one that I’m been thinking about since
VidCon which I’m really excited about is
actually the audience side of the
algorithm right because um I took a
picture of a whiteboard drawing I’m
gonna look up well while trying to do
this but I was I gave my whole you know
reverse engineering algorithm
presentation at at VidCon and then
afterwards I was talking to a YouTube
engineer and you know he came up to me
and he was like he’s limited in what he
could say he was basically like you
weren’t wrong you weren’t necessarily
right but one of the things you should
think about is the audience right and I
was like oh man that’s a really good
point because my focus is always on
creators and content creation and that
sort of thing but what that got me
thinking about was like you know I’ve
never seen anyone do a really really
interesting analysis on what YouTube
serves audiences based on what they
think about them and part of what
spurred this was I went into my Google
account and I saw that they had me
labeled as like you know having ten
different interests and like of course I
could first ones like YouTube youtubers
I was like well shocking but essentially
you know Google labels you and what
you’re interested in and it serves you
content and search results based on that
and so my thinking was there has to be a
like very moderately and I get that it’s
like okay apparently I didn’t take a
picture of it there’s um there has to be
a moderately scientific way to try and
see what YouTube would serve to someone
that they believe is interested in XYZ
right whether that’s like cars beauty
and makeup whatever it is so like I want
to create like 10 demographic profiles
within Google check that against what
they say that I’m interested in and then
go to YouTube on like a fresh cookie
free browser and see what they serve
right and watch some videos and see what
happens I just record the whole thing
right to be like oh
well you know YouTube served us 50
thumbnails when we’ve ended on the home
page and of those thumbnails 80% of them
were 10 minutes or longer or just trying
to see what it is now I got a feeling
I’m gonna get a ton of Jimmy Kimmel and
Jimmy Fallon right no matter what
demographic because they need help
better what kind of demographic
information I put in but I’m curious I’m
gonna see what happens so that’s that’s
gonna be the next one cool and so for
people to follow along on your
experiments uh you know I discover you
through the blog post that you put on I
think it was tubular labs as a tubular
lab filter to filter house to filter it
was it um are you gonna are you maybe I
make theory are you are you posting this
stuff do you have your own platform is
it are you putting up on Twitter your
own blog your on YouTube channel where
you’re sharing this or is it just no 100
and I follow Matt around and not just
hope my findings are because I want to
make sure you know my audience is the
next goal that comes out you can
certainly follow me on Twitter at Matt
Gillan that’s ma TT g IE l en i post a
lot of work to filter and i posted in
other places
I am potentially looking to do some
other other pretty cool things here
assume that can’t talk about just yet
but uh I’ll be certain to let you know
as soon as that’s up in life all right
okay okay little mystery there little
ninja stuff happening on in the
background um well yeah I mean I
followed along most of the guys who are
doing they like YouTube advice and you
know how to how to explode your channel
all that stuff and you definitely have a
different perspective which I think
people really you know if you wanted to
start a YouTube channel talking about
this stuff or blog or whatever I think I
think you bring a lot of a lot of value
and build an audience pretty quickly um
and again I’m not going to help out with
that too man I want to know the next
thing that launches you got to hit me up
will do another one of these and drop
some knowledge bombs on the audience um
cool will amazing quickly half an hour
flies and I got to get back to the
client you’re working with there
in New York is there any final message
that you want to leave the audience I
know we talked about following you on
Twitter is there anything else you’re
you’re trying to promote or spread
awareness of no nothing that I’m trying
to promote or spread awareness of just
yet you know I was on um
Tim’s podcast and I have a tendency to
look very coldly at the numbers and stop
thinking about it in terms of audience
or creators that I must be gaining that
there was Moses from YouTube but uh you
know one of the things that he brought
it back to and I was like that is a
really good point and I should probably
try to keep that in the forefront of our
mind which is you know at the end of the
day you know make content that you’re
passionate about
um you care about right I think that’s
actually one thing that anyone can learn
from looking at what you do even because
you’re very clearly passionate about
this you really care about helping
people I think that’s really really
awesome I love it well thank you for the
time thank you all you guys for watching
live go check them out on Twitter
continue to believe or whatever your one
word is and we’ll see you soon
hangout super excited to be here with a
new friend someone where we rescheduled
to get back on because he’s got some
amazing content ideas to share with us
mr. Matt Ellen he is an expert on
YouTube and figuring out the algorithm
and we’re going to talk for the next
half hour about how to dissect it and
help your YouTube channels grow so Matt
welcome aboard man thanks for having me
man if you want to give a little context
little background for the audience’s so
they know what you’re all about
sure my name is Matt Gillan and I’m
formerly the VP of programming and
audience development at a place called
Frederator you may have seen some of our
stuff before we made Adventure Time in
Fairly OddParents and a whole host of
other cartoons that were extremely
popular on TV I got to work on the the
digital side focusing on building the
audience for cartoon hangover which had
two cartoons bravest warriors and being
puppycat a whole host of really awesome
shorts on there and then I oversaw the
programming an audience development for
a channel frederator 22 we did really
well there in about a year and a half we
built that channel up to about 1.1 1.2
million subscribers doing anywhere
between you know I around 8 and 17
million monthly views and then um I also
helped build the channel Frederator
Network love it so you put out this
detailed report I forget where it was
published it was in one of the blogs
talking about the YouTube algorithm and
I think I reached out to you on Twitter
afterwards and we had a little bit of
discussion back and forth about it
what are what are some of the things
that you discover the people haven’t
read the article and we should we should
link that up so people can dive deeper
if they want but what are some of the
things that you discovered that people
aren’t doing that that could really help
the youtube channels grow um things that
people aren’t doing that would help
their YouTube channels grow I would say
is continue look at your content and
recognizing if there is content that’s
actually hurting the viewership on your
channel um you know we found out as
evidence to suggest I don’t think it
necessarily
Sara we proved it outright I don’t
necessarily forget outright but it
suggested that uploading a video that
got less than 3% of your subscribers to
watch in the first 48 hours could
actually cause viewership on your entire
channel to drop including your library
the videos and we see this most commonly
like in channels that kind of go into
like a bit of a death spiral where
either you know people aren’t watching
the newest videos or maybe their
contents gotten a little stale and one
is up happening is that their library of
videos begins losing traffic from like
suggested videos and search and they
start dropping out and so this suggested
to us that there’s there’s kind of like
a meta score or meta algorithm that’s
basically like if you’re kidding these
combination of factors we’re going to
rank all of your videos and our
discovery algorithms higher and so
that’s like things like you know browse
and suggested and recommended and
related you know in all the various
places those show up most importantly on
the home page and on watch pages and you
know if you keep uploading those you
know quote-unquote negative uploads it
can contain to your channel is at what
we saw and conversely if you’re
regularly doing you know 10 12 15 % of
your subscribers in the first 40 hours
you’ll see your channel really really
begin to grow to a certain extent so
those are some of the higher-level
findings of the of the article
we also found really cool stuff about
like duration you know literally videos
that were just longer right so whether
that video is 15 or 20 minutes or 25
minutes made a huge difference in terms
of viewership and then kind of as a
subcategory to that the audience
retention on those videos also had an
impact on how think that’s news to
anyone I think this was the first time
anyone could’ve put out any sort of like
data behind that that showed you know
when we had a video that we get a 12 or
13 percent average view duration we
would get like 2x are subscribed
years in terms of overall viewership so
you know we had about a million
subscribers when we did this research
you know videos that would get two
million views we’re doing about a
hundred and fifty a thousand views from
our subscribers in the first 48 hours
and the reason I use subscribers is a
number I don’t actually think the
algorithm really cares whether
subscribers or not I think it has to do
with a concept called view velocity
which is that they see this video
accruing so many views in the first four
hours YouTube says oh this video is
really important for the topics and
keywords with which it falls under and
the interest with which it falls under
so let’s serve this video to people with
that same interest and you’ll get a lot
of click-through and if that retention
the audience retention and click-through
stays the same as it expands it will
continue to expand it over time now I
think there’s probably stages we’re
going beyond the article now mainly
because this stuff fascinates me and I
can’t help but continue to think about
it um if you think about a video’s
lifespan and kind of like certain stages
right like you have your the first 48
hours window where YouTube is really
kind of analyzing your video and seeing
what it’s all about right its first and
foremost it’s taking in kind of the
metadata that you put in right and this
is everything from average video length
title description tags closed captions
and then it starts take into account
things like comments and who’s clicking
on it and what they’ve watched before
and so on and so forth right it starts
building that into it and there’s kind
of like this 48 72-hour window where
YouTube is basically saying you know
what is this video all about and how
good is it and by good it means really
like how good at is it at creating a lot
of watch time on our site right has
nothing to do with quality or anything
else it’s how good is it at Google on
our site now you can pull from that that
you know the better quote-unquote
quality video is going to generate that
but that’s a whole another conversation
for a whole other time and then outside
of that right there’s kind of like the
first week two weeks that you know the
initial two or three days kind of leads
to then after that is kind of a long
tail
the video and I think there’s different
kind of weightings at different times I
have to do a lot more research on it but
kind of what I’ve been seeing from like
a top-level I’m going oh that’s
something really interesting I want to
do the data on that
that’s what’s popping out at me cool and
so what’s interesting there too is
you’re saying that if if a new video
comes up doesn’t go well then that kind
if you can all your old videos – oh yeah
absolutely big time and it would we
should also take note that there’s
there’s always going to be an element of
like if one video that’s doing really
well there’s going to be a certain
percentage you know whether that’s you
know a fraction of a percent or 50 or 60
percent of the people that land on that
video to watch that are going to go and
watch your other videos right and you
know that obviously is going to have an
impact on viewership but what we saw
kind of more in the details is it’s more
like a circular effect where or like a
wave effect where the waves kind of
start building on to each other and
start overlapping right so if you start
having all these real successful videos
right your your graph starts to go I
think this is the right way on the
screen it’s a little confusing okay yeah
so like your graph starts going like
this with the ups and downs versus just
kind of like this right along with like
the uploads and it kind of build on each
other but that being said if we think
about kind of the meta algorithm or the
high level algorithm as like kind of a
channel score and say well if your
videos are regularly gaining more
audience than your subscribers that
means that your channel should have more
subscribers right because you’re
reaching beyond your audience from a
viewership perspective and I think
YouTube takes that into account and
similarly with watch time
now it could boil down to something as
simple as like well your channel is
generating a ton of stickiness for our
site so we want to send more people to
your channel so that they you know stay
on the site longer and so you know I
think I think it’s a combination of
factors but certainly keeping people on
the YouTube platform
you in to outperform yourself is going
to generate more audience interesting
let me ask you this I don’t know if you
have data on it or not uh for
subscribers who are watching your video
if you had to have one call the action
at the end of the video would it be to
leave a thumbs up leave a comment share
it
what would both what the what’s the
priority of engagement afterwards or do
you know it do have that figured out yet
and so you were saying if you had
already get like 15 20 25 percent of
your subscribers to watch what would you
ask them to do at the end of the video
yeah or even if not like even even you
only gain 3 percent or whatever it is
like what’s the highest impact thing
that a subscriber can do for you where
if your creator you have a lot of
creators who have like 20 calls actions
the end of a video as a result they’re
not going to anything right so you get
to pick one like hey a question of the
day leave it down below the comment or
please guys add this to a playlist or
give it a thumbs up does it make a
difference on the engagement to the
algorithm um I heard a saying recently
that I really loved which was tyranny of
choice where if you give if if you give
them you were too many choices they’re
gonna choose to do nothing uh and go
away right so I think a pretty good rule
of thumb is like no more than 3 calls to
actions so that they have an option
right it’s a you know pretty solid rule
of three right like you see that you
know throughout so many things in life
and so giving people three options I
think is probably the you know best
place to be I have not done any testing
on I don’t have data to back it up
I can say from a non scientific
standpoint that essentially when we
reduced our calls to action from I think
seven or eight to three we saw a better
performance we’re also making better
content but uh essentially in order of
importance routes say is watch more
first and foremost mainly because watch
time and keeping on people on the
platform longer is so incredibly
important and this kind of speaks the
programming strategy right like if
you’re making a video about LeBron James
you don’t want to then tell people to go
watch a video
know about you know horseback riding
maybe the people that are watching
LeBron James video probably don’t care
about horseback riding sure there’s a
small percentage but you know I don’t
think it’s a huge percentage right you
want to tell them to go watch another
video about LeBron James or if you don’t
have that tell them to go watch a video
about Kobe or you know Steph Curry or
whoever it is right you go from most
relevant to least relevant in your video
inventory from what video you ask them
to watch next the next thing I would say
is subscribe mainly because from from
that windowing perspective we talked
about earlier in a videos life cycle
those first 48 hours are so important
and that makes subscribers unbelievable
or right and so when you upload a new
video the first people that are likely
to see that video and click on that
video giving that video like a watch
time score giving that thumbnail like a
click-through rate are going to be your
subscribers right and they’re going to
be the best performers and so from that
perspective you know the more
subscribers you have right the greater
reach you have the more watch time the
more session duration time you’re able
to create the better is going to be for
your video and for your channel so
number two would be subscribe and then
number three would probably be some sort
of engagement I’ve seen no indication
that engagement has any impact on the
algorithm whatsoever
you knows if there was a in this day and
age historically yet it’s incredibly
important which makes me believe makes
me think that there might be still some
weighting going on in the algorithm and
then if you also think about it from a
more theoretical standpoint
videos that are engaging are likely to
accrue more watch time simply by the
fact that while someone is going to do
that action they’re still watching the
video right say hey comment down below
right well someone then has to think
about your question they have to think
about what to write right that’s
actually type it out alright chances are
well maybe not on YouTube but another
place is people will go back and like
check it for grammar and spelling you
know they’re going to want to add their
emojis not your audience I’m sure your
audience is good on that I know there’s
lots of them in the live chat doesn’t
mean lots of emojis happening
that takes time right it’s so like yeah
maybe videos with like a
disproportionate number of comments
right are racking up this incredible
watch time because people are literally
having to sit there while the video
plays and write out the comment in that
scenario it’s like well you know which
is it is the watch time or the comment
like well the watch doesn’t happen
without the comma so you know it’s some
sort of engagement action would be good
at the very least to show that you
actually care about what you’re your
subscribers and fans think and keep them
engaged by the show right so comments to
what happened someone who has a comment
somebody else responds then coming back
to the video if they come back to the
video and just respond to the comment
but without watching again because
they’ve already seen the video they just
want to be part of that thread could
actually check it out I should be a
negative signal um no I don’t think so I
think it’s it’s just engagement
generally like I don’t I don’t think I’m
sorry are you asking more about like if
a repeat view happens yes he’s like
hopefully think so on this video we’re
gonna get comments afterwards right
maybe I respond to one and say hey you
know Tamara thank you so much for
watching video she comes back to see my
comment and just to leave another
comment to say hey it was awesome but
she’s not gonna watch this whole
half-hour thing again because she’s seen
it if she ends her session there does
that negatively impact the video now
um I mean theoretically it could again I
don’t have any any data just yes one way
or the other
what I would say is that I know
historically YouTube has given algorithm
waiting to videos that generate repeat
viewing I think there’s probably a lot
of evidence in the kids space to suggest
that like oh yeah that can be really
powerful if they do repeat viewing so
from that perspective I would say you
know it’s probably toss of the coin
either way you know generally speaking
though I think if someone’s coming back
to stay engage in the conversation
that’s got to be a good thing right I’m
thinking now at the end of a lot of my
videos I’ll ask a question
the question today is whatever leave it
below but I’m never really doing hey
here are some related videos that you
guys might enjoy if you’re saying that’s
like the primary call to action should
be watch more than you know should i has
some selfish consulting here you know
should I change my strategy to like put
three links in the description say hey
guys here are three videos I think II
need to watch next go check them out in
the description below and not do a
question of the day well I think you
still do a question of the day but I
think yeah you should absolutely tell
people you know what they might enjoy
nests cuz like they think about it from
two perspectives right you ever think
about it from your audience who has
watched you know a lot of your videos or
many of your videos you have to think
about it from a new audience right
because a lot of people who watch that
video are not going to be people who are
a subscriber who may have ever seen your
video before right so it’s a way of
introducing them to your content letting
them know that you have a ton of other
videos up that they might really like
and then think about like well the
reason why people are on YouTube is to
be entertained and to discover things
and if you know you can help them do
that you know I think that’s really
valuable to the audience and yeah it’s
really helpful to you as well so you
know from that perspective absolutely
yeah I think so cool well the people in
the live chat are saying don’t stop the
question of the day please so keep that
for you guys okay don’t worry uh
otherwise you’re gonna get tons of hate
and tweet Matt come on man uh-huh so
okay so for an ounce of creators who are
looking to launch a new series on their
channel maybe them in one video week you
want to go to two videos a week uh the
metric you would look at the judge the
success of a new series would be
subscribers watching the video that’s a
tough one and and that’s a tough one
because almost no matter how much
promotion and marketing you do for the
new series right it’s not going to get
as many views as like your current
popular series mainly because in my view
it shows up in a subscribers feed like I
don’t
recognize this I don’t understand this I
am NOT going to click on this right this
is not something I usually watch your 10
thumbnails or something I usually watch
I would like to click on one of those
now there’s a whole bunch of strategy to
to get around that but to answer your
question about the success of a new
series I would focus on more about what
the average view duration is and if
you’re getting really good average view
duration like let’s say you’re 60 to 80
percent range which i think is pretty
damn good for longer videos if you’re in
that range great if you’re a little bit
below that range you might want to look
at like the structure of the video are
there ways to keep more people engage as
longers their way to get into the video
quicker that sort of thing but generally
speaking I think average view duration
is going to be the most like apples to
apples that you can do to your current
series in a way to kind of measure the
long-term viability of it
in addition um you know I wouldn’t I
wouldn’t cut a series until you’re at
least like five six seven episodes into
it if it’s something you care about and
are passionate about mainly because one
that’s how long it’s going to take the
build you know any sort of audience for
it out of out of your current subscriber
base but also it gives a good amount of
content for people to go back to and
check out and one go right like that
word YouTube video Genma five seven
minutes that’s not a huge duration of
entertainment time right where that is
if you have five or six episodes of that
well not you’re talking about like yeah
you know I could sit down eat a hoagie
and watch three or four these videos and
determine if I like it or not right like
do you see the Netflix study that came
out that was like you know Netflix knows
exactly how many episodes of a show you
need to watch to fall in love with that
show entry right yeah and like for
certain series it was like need to watch
three and you’re going to watch every
season for other shows it was like watch
seven or eight and so you know giving
yourself a good enough kind of data set
of headspace did a set of eight videos
to get a better feel for it I think is
kind of a good benchmark somewhere in
the 5-8 range so one of the things that
we’ve been looking at as we add a new
series in would
so subscribers gain per video maybe
that’s not a great metric but you know
our goal is for every video series we
put out we want every video to be in the
double digits of new subscribers joining
you know if it’s if it’s negative and
that’s a really bad series obviously but
if it’s like single digits five yep your
current audience right it doesn’t
necessarily mean it’s a bad series right
which is the prop right like you it
might be you you two doesn’t give a hoot
about any individual video all that
lunch there there are a lot of
exceptions to that rule but for the most
part any individual video is not all
that important to YouTube in the sense
of you might have the great give you an
example from a movie and TV channel we
had a Fredrik called cinematic a’
cinematic a’ we built about a hundred
thousand subscribers after like three or
four months or something we do have
three or four million monthly views and
we built that audience mainly by doing
like listicles and theory shows and this
kind of crap about like The Walking Dead
Game of Thrones Star Wars and Marvel
right and we wanted to be a more
well-rounded audience we wanted to bring
more females in so we did a listicle on
orange is the new black there’s a
perfectly fine video perfectly good
video nothing revolutionary nothing
crazy like it shouldn’t have done ten
million dues but it should have done way
more than like the four thousand views
that it did because the average video on
that time did a couple hundred thousand
views for us and so that was the first
indication I had that was like okay
didn’t really give a crap about how good
of an orange is the new black video that
is what big of a crap about is whether
the people that they served it to which
happen to be your subscribers right
because that’s who gets to see it first
clicked on it right if those people
didn’t click on it right they’re not
going to go and say well this is a great
orange is the new black video right and
they’re not going to give it a chance to
be a great orange is the new black video
maybe over time yeah you know five six
seven eight months it gets picked up
something happens to it but out of the
gate no right so that that’s part of
what led to this idea of like you have
to focus on this very specific niche and
so with that in mind that’s why I think
like a negative subscriber number is a
really hard way to judge a new series
because it’s it’s only about that series
in your audience right or I’m sorry it’s
only about how the very specific
audience that you currently have access
to interprets that video right and
whether or not they like it or it makes
them unsubscribe or whatever it is so
from from that perspective I would
really encourage you to to not not look
at that as a primary metric when when a
new series first goes up like give it
some chance let me fight for your
children okay I love it um I know I know
you know you are obsessed with data and
trends and frequent stuff out and
hacking your way what what’s the next
thing you’re looking at what are the
some of the things that are on your
radar that you’re exploring the next
blog post is going to come out and drop
some new knowledge bombs on um well I’ve
gotten really excited about a couple of
things so like one of the biggest
questions I see like when I’m trolling
on like you know our partnered YouTube
and our YouTube is like you know how do
I launch a channel or how do I get
audience and um you know it feels like
there’s no way to build a YouTube
channel just in these day and age and
all kind of stuff and like no that
definitely hurts me because I think
there is now I also understand that I’ve
had the benefit of working with existing
audiences right so like yeah we built
cartoon hangover more or less from
scratch but we’re kind of riding on the
coattails of the creator of Adventure
Time right there was a bacon audience
there for bravest warriors which led us
to our first you know a couple hundred
thousand subs and then you know when we
started launching Channel Frederator we
leaned on cartoon hangover and we’re
able to send audience over but that said
you know I’ve helped channels launch and
over the last like a year or two and
I’ve seen them become extremely
successful and I think I’m gonna start
my next big one would probably be that
you know how to launch a channel in 2017
right and drive audience the the second
one that I’m been thinking about since
VidCon which I’m really excited about is
actually the audience side of the
algorithm right because um I took a
picture of a whiteboard drawing I’m
gonna look up well while trying to do
this but I was I gave my whole you know
reverse engineering algorithm
presentation at at VidCon and then
afterwards I was talking to a YouTube
engineer and you know he came up to me
and he was like he’s limited in what he
could say he was basically like you
weren’t wrong you weren’t necessarily
right but one of the things you should
think about is the audience right and I
was like oh man that’s a really good
point because my focus is always on
creators and content creation and that
sort of thing but what that got me
thinking about was like you know I’ve
never seen anyone do a really really
interesting analysis on what YouTube
serves audiences based on what they
think about them and part of what
spurred this was I went into my Google
account and I saw that they had me
labeled as like you know having ten
different interests and like of course I
could first ones like YouTube youtubers
I was like well shocking but essentially
you know Google labels you and what
you’re interested in and it serves you
content and search results based on that
and so my thinking was there has to be a
like very moderately and I get that it’s
like okay apparently I didn’t take a
picture of it there’s um there has to be
a moderately scientific way to try and
see what YouTube would serve to someone
that they believe is interested in XYZ
right whether that’s like cars beauty
and makeup whatever it is so like I want
to create like 10 demographic profiles
within Google check that against what
they say that I’m interested in and then
go to YouTube on like a fresh cookie
free browser and see what they serve
right and watch some videos and see what
happens I just record the whole thing
right to be like oh
well you know YouTube served us 50
thumbnails when we’ve ended on the home
page and of those thumbnails 80% of them
were 10 minutes or longer or just trying
to see what it is now I got a feeling
I’m gonna get a ton of Jimmy Kimmel and
Jimmy Fallon right no matter what
demographic because they need help
better what kind of demographic
information I put in but I’m curious I’m
gonna see what happens so that’s that’s
gonna be the next one cool and so for
people to follow along on your
experiments uh you know I discover you
through the blog post that you put on I
think it was tubular labs as a tubular
lab filter to filter house to filter it
was it um are you gonna are you maybe I
make theory are you are you posting this
stuff do you have your own platform is
it are you putting up on Twitter your
own blog your on YouTube channel where
you’re sharing this or is it just no 100
and I follow Matt around and not just
hope my findings are because I want to
make sure you know my audience is the
next goal that comes out you can
certainly follow me on Twitter at Matt
Gillan that’s ma TT g IE l en i post a
lot of work to filter and i posted in
other places
I am potentially looking to do some
other other pretty cool things here
assume that can’t talk about just yet
but uh I’ll be certain to let you know
as soon as that’s up in life all right
okay okay little mystery there little
ninja stuff happening on in the
background um well yeah I mean I
followed along most of the guys who are
doing they like YouTube advice and you
know how to how to explode your channel
all that stuff and you definitely have a
different perspective which I think
people really you know if you wanted to
start a YouTube channel talking about
this stuff or blog or whatever I think I
think you bring a lot of a lot of value
and build an audience pretty quickly um
and again I’m not going to help out with
that too man I want to know the next
thing that launches you got to hit me up
will do another one of these and drop
some knowledge bombs on the audience um
cool will amazing quickly half an hour
flies and I got to get back to the
client you’re working with there
in New York is there any final message
that you want to leave the audience I
know we talked about following you on
Twitter is there anything else you’re
you’re trying to promote or spread
awareness of no nothing that I’m trying
to promote or spread awareness of just
yet you know I was on um
Tim’s podcast and I have a tendency to
look very coldly at the numbers and stop
thinking about it in terms of audience
or creators that I must be gaining that
there was Moses from YouTube but uh you
know one of the things that he brought
it back to and I was like that is a
really good point and I should probably
try to keep that in the forefront of our
mind which is you know at the end of the
day you know make content that you’re
passionate about
um you care about right I think that’s
actually one thing that anyone can learn
from looking at what you do even because
you’re very clearly passionate about
this you really care about helping
people I think that’s really really
awesome I love it well thank you for the
time thank you all you guys for watching
live go check them out on Twitter
continue to believe or whatever your one
word is and we’ll see you soon
Please follow and like us: