00:00:00
you know the problem with shovelware
00:00:01
games as it turns out it's actually got
00:00:03
worse way way worse and today's story
00:00:06
basically started off because loads of
00:00:09
us me included maybe you went to steam
00:00:11
nextfest and early on in the most recent
00:00:14
nextfest we kind of saw it was a heap of
00:00:17
slop and that was really quite worrying
00:00:21
look for many of us there's the fear of
00:00:22
missing out that can come in bad ways
00:00:24
from Publishers but it can also be quite
00:00:25
genuine the idea that if there is a
00:00:28
torrent of slop games on Steam that you
00:00:30
will miss out on good games because
00:00:32
they'll just be buried it is becoming a
00:00:34
problem and the thing is steam actually
00:00:37
knows about it the sort of hidden angle
00:00:39
to today's whole story is what steam is
00:00:42
actually doing and how the publishing of
00:00:44
games actually works but a out the tone
00:00:47
almost 19,000 games released on Steam in
00:00:49
2024 that number keeps going up and by
00:00:52
the time of the most recent nextfest
00:00:54
things really began to feel different so
00:00:58
many slop games lots of just weird AI
00:01:01
stuff and aftermath posted this it's got
00:01:04
a good headline steam users worry good
00:01:06
nextfest games are being crowded out by
00:01:08
AI slop that's basically my experience
00:01:11
of the start of this nextfest but it's a
00:01:14
long headline and it ends with the real
00:01:16
problem is more complicated and that's
00:01:18
not just a tease to go and read the
00:01:20
article I mean maybe it is that but no
00:01:23
it's because it is actually more
00:01:25
complicated it's more complicated
00:01:26
because valve are doing something about
00:01:28
it and if next Fest felt awful to you
00:01:31
early on it's because you were kind of
00:01:34
part of an experiment I've also got a
00:01:37
truly novel experience to share it's
00:01:39
boot Dev who have sponsored today's
00:01:40
video you see there you can learn the
00:01:43
coding engineering and career skills to
00:01:45
break into backend web development it's
00:01:47
like an average salary of 100 Grand work
00:01:49
from home and a lot of really sick stuff
00:01:51
but with quite the twist because their
00:01:53
founder is a massive gamer and that's
00:01:55
why boot is different it's an RPG quests
00:01:57
levels XP boss battles real time
00:02:00
leaderboards and an active Discord
00:02:01
Community almost I guess like having
00:02:03
your programmer MMO Guild it's really
00:02:05
cool and when you're learning a new
00:02:07
skill that kind of thing matters and the
00:02:10
skills You' be learning are backend web
00:02:11
development using Python and go across a
00:02:14
wealth of content that cover the
00:02:15
fundamentals and even deeper learning
00:02:17
topics like really getting into
00:02:19
algorithms which by the way is real fun
00:02:21
you'll build projects for your portfolio
00:02:23
and you'll Dodge two massive problems
00:02:25
one tutorial hell you know where you
00:02:27
just Bounce from place to place randomly
00:02:28
in the internet without a cohesive path
00:02:31
and two get rich quickstyle boot camp
00:02:34
hell because Boot is actually realistic
00:02:37
it's got the sauce you can see that
00:02:39
through their UI their Vibes how they've
00:02:41
structured everything and they're just
00:02:43
playing better than the rest and you can
00:02:44
find that out for yourself with 25% off
00:02:46
your first purchase at boot. devb news
00:02:49
it's no risk you can browse their
00:02:51
content do demos and a lot more for free
00:02:53
and when you do buy it's a 30-day money
00:02:55
back guarantee to sort I what the hell
00:02:57
was going on afterm spoke to Simon
00:02:59
carless he's one of our go-to analysts
00:03:01
as well he runs a great newsletter now
00:03:03
we they and probably you notice this
00:03:06
heap of slop early on in nextfest awful
00:03:09
looking AI generated banner and capsu
00:03:11
art loads of shovelware games and just
00:03:14
Vibes that made steam feel real cheap I
00:03:17
mean if you love PC gaming and you do it
00:03:19
in Steam but you go to steam and you
00:03:21
just see lots of [ __ ] that's just not
00:03:23
going to feel good but before long it
00:03:26
actually stopped to feel like that we
00:03:28
began to see last of the slop and Simon
00:03:31
frames it as a classic experiment at the
00:03:33
start when you don't have data you
00:03:35
spread out Impressions more evenly and
00:03:38
by you I mean steam here right and then
00:03:41
as you get that data like click-through
00:03:43
rates download rates demo play time and
00:03:45
so on you then begin to Dole out those
00:03:48
Impressions based on performance right
00:03:51
so basically it's finding product Market
00:03:54
fit that's essentially what the steam
00:03:56
algorithm does by the way it doesn't
00:03:58
take long for steam users activity to
00:04:00
basically discover near nonexistent
00:04:03
product Market fit games because they're
00:04:05
slob they're [ __ ] nobody wants them so
00:04:07
they've got awful click-through rates
00:04:08
people don't download the demos and they
00:04:10
delete the demos that shows valve that
00:04:13
they should not show that game anymore
00:04:15
because valve wants you to download
00:04:17
games to play games to buy games so
00:04:19
showing you slop is against their
00:04:21
interests that's basically the point
00:04:24
valve's algorithm is there to sort the
00:04:26
weight from the chaf so to do that
00:04:28
sorting it needs to show you both now
00:04:31
their intent is to do that to show you
00:04:33
the wheat so that you're happy but what
00:04:36
happened at nextfest is that for so many
00:04:38
the chaff scared them off that's because
00:04:41
First Impressions count that's why
00:04:43
people were spooked hell steamdb went
00:04:45
and made an AI filter and steam users
00:04:48
are quite obviously wanting valve to
00:04:50
make an official AI filter I guess
00:04:52
people will do it with tags perhaps but
00:04:54
you get the point if you open Steam and
00:04:56
see loads of slot you will actually
00:04:58
worry for the future of P PC gaming
00:05:00
because steam basically is PC gaming
00:05:02
right now and you then think how is a
00:05:04
store raking in billions let torrent of
00:05:07
slop just be thrown at customers well if
00:05:10
you're a YouTube user which you probably
00:05:12
are because you're here you're probably
00:05:14
aware already and if you ask valve about
00:05:17
this they'll actually have a clear
00:05:19
answer we all know valve loves systems
00:05:21
and numbers so I'll start this segment
00:05:23
on their terms 1.7 million new customers
00:05:26
bought a top 2023 game they played 141
00:05:31
million hours of other games spent $73
00:05:33
million in premium games and $20 million
00:05:36
on
00:05:37
microtransactions that was a lot of
00:05:39
stats the source is valve and that data
00:05:41
is what we call a cohort analysis it's
00:05:44
when you grab a group of users based on
00:05:47
whatever your criteria are you know
00:05:49
usually a time period and then you see
00:05:52
what those users went and did and valve
00:05:54
highlighted this cohort for a very clear
00:05:56
reason the thrust of their 2024 year in
00:05:59
review was that steam is not just a
00:06:02
store that steam attracts and retain
00:06:05
customers who will then go on to
00:06:07
purchase more games that's why the
00:06:09
selection criteria for that cohort is
00:06:12
new customers who bought a top game it's
00:06:16
essentially valve saying that for their
00:06:19
economy the whole trickle down thing
00:06:21
really does work and works at scale and
00:06:23
it works because their algorithm makes
00:06:26
it happen and that's also how they
00:06:28
justify their 30% which makes sense yes
00:06:32
you can put your game on the epic game
00:06:34
store and you can only have Tim Sweeney
00:06:36
take
00:06:37
12% but you're going to get no sales
00:06:40
basically steam does in fact drive so
00:06:42
many customers that epic may as well not
00:06:45
exist at least in terms of it being a
00:06:47
traffic source and to Ram that home
00:06:49
valves say the new game revenue is up
00:06:51
10x over 24 and that it's up massively
00:06:55
over 2023 now they then get a bit more
00:06:57
granular in their numbers and that's
00:06:59
where today's video will really kick
00:07:01
into gear they say over 500 games made
00:07:04
over 250 Grand in their first 30 days
00:07:07
and more than 200 hit a million within
00:07:10
that same time span and that those
00:07:12
numbers people hitting those two
00:07:13
Milestones of 250k and a million are up
00:07:16
by 27 and 15% over 2023 which does look
00:07:21
like bloody good news more games are
00:07:24
succeeding than ever but there's another
00:07:26
way to read that and that is that well
00:07:29
even per those stats only 3% of games
00:07:32
hit 250k within 30 days which to many is
00:07:36
[ __ ] depressing but in a way it's
00:07:38
also not true so now it's time for
00:07:40
something that honestly blew my mind
00:07:43
when I first learned about it and I have
00:07:44
to start with a confession I kind of
00:07:46
knowingly pulled a fast one on you so
00:07:48
the 19,000 number that I opened with is
00:07:51
a true number but it's uh it's
00:07:53
misleading that's basically the point so
00:07:55
get this 79% of games released in Steam
00:07:59
may as well not count toward that number
00:08:01
this is where it gets real fascinating
00:08:03
because that 19,000 is eyeg grabbing
00:08:06
it's a really good headline but what
00:08:09
about the 79% it sounds wild but it's
00:08:11
not and actually credit to kitaku for
00:08:13
pointing this out in January as it turns
00:08:15
out 79% of games are what's called
00:08:19
limited this means that they don't have
00:08:21
trading cards Badges and can't have
00:08:24
achievements displayed on player
00:08:25
profiles and they won't until steam says
00:08:28
they achieve quote broad player
00:08:30
engagement and some Commercial Success
00:08:33
now these games can still be sold to
00:08:35
people they can be recommended to people
00:08:37
but steam basically does not acknowledge
00:08:39
them otherwise here's where it gets
00:08:41
interesting we've got some analysis from
00:08:43
Bill Young who's the head of games at
00:08:45
twitch so he found that of the
00:08:49
18,99 to games that launched in 2024
00:08:54
14,95 are still limited on Steam and
00:08:59
we're at actually seeing that as the
00:09:00
number of game launches increases the
00:09:03
portion of limited games is also
00:09:05
increasing that 79% is up from 2019 66%
00:09:10
number now I know that I have just
00:09:14
absolutely bared you with stats and I'm
00:09:16
sorry I know it it can be hard to follow
00:09:19
and kind of annoying but they are all
00:09:21
going towards somewhere and our first
00:09:23
order lesson of those numbers is
00:09:25
basically this our feeling of worry
00:09:27
disgust and fomo J nextfest is actually
00:09:31
Fair because it is a real problem that
00:09:34
we are
00:09:34
seeing but valve already thought about
00:09:37
it because the algorithm already deals
00:09:40
with it nextfest in fact had so much new
00:09:43
that for a very brief period we had a
00:09:46
glimpse into the true hell of raw
00:09:49
unfiltered steam but before long the
00:09:52
store algorithm learned and the sloth
00:09:55
was filtered out and to understand this
00:09:58
let's hit some of those numbers again
00:10:01
valve count new release revenue is
00:10:03
anything within 30 days of release so
00:10:05
what we can do is we can take a look at
00:10:07
the week starting February 18th okay
00:10:10
steamdb says the 385 games were released
00:10:14
that week 385 is obviously a lot of
00:10:16
games that comes out to 55 games a day
00:10:18
some are slop some are adult some are
00:10:20
clearly student projects but most fell
00:10:23
into steam's limited category now can
00:10:26
you imagine falling into that category
00:10:28
right um I I mean that slop is filtered
00:10:30
out by that category is really good what
00:10:32
about false positives what if you fall
00:10:34
into that well if the steam algorthm
00:10:37
feeds us a tailored personal reality
00:10:41
then what are we missing actually this
00:10:43
gets to some things that are [ __ ]
00:10:45
scary so to work it all out dug a bit
00:10:48
deeper as customers we've got a fear of
00:10:50
missing out in a great game and as devs
00:10:52
well we're afraid of being wrongly
00:10:55
limited so let's work the problem by
00:10:57
breaking our sample week down a bit more
00:10:59
more of the 385 games 20 were premium
00:11:02
games with over 1,000 followers which we
00:11:05
know because we highlighted them in our
00:11:06
members newsletter which will be linked
00:11:08
below anyway there's a mix of double a
00:11:10
AAA Indies a whole bunch and accepting
00:11:13
King of Fighters getting completely
00:11:15
wrecked over networking supp pretty
00:11:17
typical for a fighting game all of those
00:11:19
games have at least a 75% user rating in
00:11:22
fact half are above 90 yes half are
00:11:25
above 90 that seems absolutely insane
00:11:28
and amazing
00:11:30
until you think about it steam's job is
00:11:32
to show you things that you will like so
00:11:35
assuming it's got enough games to work
00:11:37
with and enough people to match those
00:11:38
games too steam will identify winners
00:11:41
and grow
00:11:42
them is because there's a little more to
00:11:45
this diving into those games only six
00:11:49
have got over 1,000 reviews three of
00:11:51
those are AAA games most have only got
00:11:54
300 to 600 reviews and that is a problem
00:11:58
the reason why it's fairly simple the
00:12:00
general belief is that the revenue that
00:12:02
corresponds to about 4 to 500 reviews is
00:12:05
what will typically pay a full year wage
00:12:07
for an American solo developer making
00:12:09
that in a month is absolutely awesome by
00:12:11
the way but we are talking about the
00:12:13
most successful games within our sample
00:12:16
and remember our sample is a selection
00:12:18
of the most successful games that's kind
00:12:21
of grim and when you look at most of the
00:12:23
other games even within our very
00:12:25
successful sample you see that most have
00:12:28
not increased their review count much at
00:12:30
all over the last few days and in a way
00:12:34
that's actually by Design now I could
00:12:35
give you some anecdotes based from well
00:12:37
our time releasing the pil Beyond and
00:12:39
steam but Chris zakowski has a lot more
00:12:41
experience than I do and a lot more
00:12:43
perspective aftermath talk to him and
00:12:45
here's how he explained what's going on
00:12:47
here steam's release process begins by
00:12:49
giving a relatively equal number of
00:12:51
Impressions to just about any game that
00:12:53
does not fall into the J AI slop bucket
00:12:56
this happens for the first two or so
00:12:58
days and in this period many steam users
00:13:00
can see a game if steam thinks the game
00:13:03
is particularly well targeted to you
00:13:05
it'll actually show it to you somewhere
00:13:07
and to get even more exposure and even
00:13:10
more data for the algorithm steam will
00:13:12
feed that game through the discovery Que
00:13:14
Now what then happens next is what makes
00:13:18
or breaks Studios if your game sells
00:13:21
well from those initial Impressions
00:13:24
steam will give you more visibility if
00:13:26
you do not sell well you will stay
00:13:29
firmly in your Niche now you will get
00:13:32
the odd algorithmic sale here or there
00:13:34
but by and large it will be on you to
00:13:36
sell your game and if this happens to
00:13:39
you the steam store will feel a lot more
00:13:41
like the epic game store because you're
00:13:43
just kind of in sad algorithm land
00:13:46
because as far as Steam knows when it
00:13:48
shows people your game they don't decide
00:13:50
to buy it and to be clear in this period
00:13:52
the thing that matters the most is
00:13:54
actually Revenue steam tracks loads of
00:13:56
Statistics but revenue is King
00:13:59
now Chris is not talking about Bad Games
00:14:02
staying in their Niche though that
00:14:03
absolutely is the case that does happen
00:14:05
with bad games what he's talking about
00:14:07
here though is gross revenue and he has
00:14:09
a number for him the number is
00:14:12
250k right 250k seems to be one of the
00:14:16
real thresholds where if you cross it
00:14:19
you will start to be shown far more
00:14:21
broadly and funny enough that
00:14:23
250k actually matches valve's use of
00:14:26
250k that we talked about earlier from
00:14:29
their 2024 year in review so most games
00:14:32
need to break 250k Revenue in Steam to
00:14:35
see major long-term success long-term
00:14:38
success generally is possible but not
00:14:40
major long-term success so with that in
00:14:42
mind we'll go back to our cohort of 20
00:14:44
games of them six hit the 250k Mark 14
00:14:48
did not and that's even though these 20
00:14:50
had ratings over 75% and half of them
00:14:53
had ratings over 90% most of our 14
00:14:56
again that's the games in this cohort
00:14:58
who did did not make 250 or more well
00:15:02
they are making between 10K and
00:15:05
35k which is not a lot of money if you
00:15:08
are supporting jobs and remember that
00:15:11
this 20 is not a representative sample
00:15:13
of the 385 games that launched in that
00:15:16
week it's the best 20 so uh yeah this is
00:15:20
even with selection bias cutting in our
00:15:23
favor now those cases that I talked
00:15:25
about are net revenue not gross but you
00:15:27
will often hear gross Revenue numbers be
00:15:29
talked about for games and when you do
00:15:31
there's something really important you
00:15:32
need to know a very common rule of thumb
00:15:35
is that you will see about 50% of the
00:15:38
gross revenue that is reported on your
00:15:40
steam dashboard basically 30% goes to
00:15:43
valve then you've got refunds taxes
00:15:45
Payment Processing just other stuff and
00:15:47
if you had a publisher well depending on
00:15:50
your publishing terms you may see
00:15:52
30% of that 55% until you've paid back
00:15:56
recoup some devs actually see no money
00:16:00
at all until they've hit recoup or maybe
00:16:02
even hit double their recoup because
00:16:04
some publishing deals in this industry
00:16:06
are
00:16:07
absolutely vicious anyway so far this
00:16:10
does look rather Doomer doesn't it but
00:16:12
in a way it's not in a way it's
00:16:13
competition so what I'm now going to do
00:16:15
is flip the script to make this hit
00:16:18
Let's Pretend We're devs and we're
00:16:19
talking shop so is the solution to say
00:16:21
that steam is bad and you should not
00:16:22
release in Steam obviously no it's not
00:16:25
yes there is slop and that slop is
00:16:27
worrying but what I've covered in this
00:16:29
video shows how steam systemically
00:16:31
quashes slot as soon as it is identified
00:16:35
if there's a problem then I think the
00:16:37
problem is basically this it is launch
00:16:39
day importance you live or you die by
00:16:41
your launch window and in some ways that
00:16:44
can feel bad in a sense it is valve's
00:16:46
fault but fault is maybe too loaded a
00:16:49
term launch is the time that a game
00:16:52
should be judged that's fair that makes
00:16:54
sense and steam's got a bit of a problem
00:16:57
it can only get data on a game
00:16:59
performance when the game is out there
00:17:01
generating data I.E when it's launched
00:17:04
and it's less that you are destined for
00:17:06
Success until steam murders you and it's
00:17:09
more that if you perform well on your
00:17:10
own merits valve will then expand your
00:17:13
reach until New cohorts of users begin
00:17:15
to stop buying as much and that's pretty
00:17:18
much as far as the algorithm will push
00:17:20
you it's on you to make a fun game that
00:17:23
is visually appealing with a broad
00:17:25
enough audience and a well selected
00:17:27
price point that releases at a good
00:17:29
level of quality for the best shot what
00:17:30
you have to do is nail that have a
00:17:32
quality launch and cluster all your
00:17:34
marketing efforts onto day one with
00:17:36
priority number one being 50 reviews
00:17:39
ASAP that tends to be the general advice
00:17:44
that does mean that it h we'll fix it
00:17:45
later attitude is obviously not
00:17:47
acceptable because that will hurt you
00:17:48
and in a way you could say This Is Us
00:17:50
jumping through valves Hoops but let's
00:17:52
again frame shift cuz it would be lovely
00:17:55
if steam just fed you impressions for
00:17:57
longer but think about it from From
00:17:59
steam's perspective there's a torrent of
00:18:01
new games absolutely all the bloody time
00:18:04
and that's not even counting the store
00:18:06
space taken up by events by sales DLC
00:18:08
and everything else so what's it going
00:18:11
to do how is it going to Dole up its
00:18:13
Impressions there's always going to be a
00:18:16
balance there so there will naturally
00:18:17
have to be some sort of limit around
00:18:19
your launch now what I've just said is
00:18:21
essentially how we should act when we
00:18:23
think about steam but that's not
00:18:26
necessarily because everything's perfect
00:18:29
that's how we should act to get good
00:18:30
results there is a deeper concern and
00:18:32
it's what actually Spooks me steam can
00:18:34
only act based on your past behavior
00:18:37
that means that by default it will
00:18:38
pigeon hole you and that means it can
00:18:41
actually make it hard to find truly new
00:18:44
games because on Steam on YouTube on
00:18:47
everything you live in a personalized
00:18:49
reality a developer may have a totally
00:18:51
new game that you had [ __ ] love that
00:18:53
you would 100% bu have given the chance
00:18:56
but steam just literally cannot show you
00:18:59
because it's only able to follow your
00:19:02
past actions in a way it's kind of not
00:19:05
steam's fault it's a natural problem
00:19:08
that emerges from all of this being
00:19:09
algorithmically driven yet it being
00:19:12
algorithmically driven is what saves us
00:19:16
from the Torrance of slob it's what made
00:19:20
the later days of steam nextfest
00:19:22
actually feel really good versus the
00:19:24
earlier days when the algorithm did not
00:19:26
have data and it is how valve justif if
00:19:28
I rightly in many ways their 30% cut
00:19:31
because their algorithm and their
00:19:33
traffic does generate sales it makes
00:19:36
successes it is absolutely awesome to
00:19:38
watch valve problem solve and come up
00:19:41
with ways to just deal with the slot
00:19:43
problem but doing it all with systems
00:19:45
and algorithms comes at the deeper cost
00:19:48
of personalized reality that makes
00:19:52
change hard that can make Innovation
00:19:55
hard it can mean that if you want to
00:19:57
make a game you need to make that game
00:19:59
based on what you know does well on
00:20:02
Steam because steam operates everything
00:20:04
based on the data that it has and if we
00:20:07
only follow lagging indicators how can
00:20:10
we be truly new and creative how can we
00:20:13
discover truly new things these are the
00:20:15
problems of our modern world you see
00:20:17
them in YouTube you see them in news and
00:20:19
here we see them in Steam both in the
00:20:22
good that they drive The Innovation that
00:20:24
they do actually help the games that
00:20:27
only exist and get funded because steam
00:20:30
can basically get you shitloads of sales
00:20:32
without a marketing department but it is
00:20:34
at a cost a cost you will feel and
00:20:37
reckon with because as this is happening
00:20:39
we've also got the crumbling of physical
00:20:41
games and Rising Hardware prices those
00:20:44
are going to precipitate quite a lot of
00:20:46
change in our industry and you can learn
00:20:48
about that in our next report
