Loading the document
1 free read left. Create a free account for unlimited reading + AI research.
Create free accountDistilling the key insights…
You control the fail rate much more aggressively. Oh yeah. Because you want to like force people to watch. Forcing people to watch Ness is much softer than forcing them to pay which all the three games do. Therefore, we can retain this extremely high retention and squeeze and squeeze and squeeze and squeeze them.
It's diceen [music] up the game. Midnight crew talking UA adverts [music] and game design too. Mate feeling shak bringing the insight. We're rocking those vibes till the early daylight. [music] Master eyes on the prize tracking data through the cyerspace skies. Feeling stacks colors like a wizard in disguise. [music] Jack and craft realms lift us to the highest. Two and a half gamers talking smack. Slow hockey sick. Got your back. Ads are [music] beautiful. They light the way. Click it fast. Don't delay. Uh-huh. Uh-huh. [music] Uh-huh. [music]
Uh-huh. [music]
No, I'm just hosting.
Welcome to Two and a Half Gamers. Uh, we haven't recorded an episode in quite a while, so it's quite fun to see you all here today. Today, we're talking about a very, very special game because we're doing a very, very special episode. [ __ ]
like it always is. Today we're going to talk about Come on. like just
it is very special. Every episode with you guys is special cuz you're so special to me.
Okay, the [ __ ] Help everybody. This is
gamers. Yes, [laughter] this is enough gamers [gasps]
and I'm not sure this is going to make it into the like the final episode, but anyway, my name is Matthew Lancharic.
I'm Felix Broberg.
And I'm Yakubier.
And we are your hosts and we're talking about Blogbust again, but this time from very different, very, very special point of view. as well. Right, Felix?
You say you say special.
[laughter] It's very nor I have to say like this video was sent to me by a lot of different people actually.
Yeah,
that what we talk about.
Yeah.
Yeah. So, so bit back.
Yeah, me neither. But
so for those of you that don't know or that do know, Block Blast has been one of the largest games on mobile by DAU.
They supportive users for some time.
Maybe they implemented Kinoa into the live system. Woohoo. Because Kinoa is also sponsor of our podcast. Thank you very much Kinoa. Gina means you know AI operating system for mobile game operations live segments flows live ops game ops all you can imagine no death tickets so go to kinoa.ai integrating hours and turn your players into the money m
just into more revenues kinoa.ai AI. There you go. There you go.
Me and Yak. Me and Yakob have gone back and forth quite a lot about this. What is it? Is it Genchin Impact you want me to put here? Or
what? You don't even [ __ ] remember?
Garina Free Fire.
Free fire. [ __ ]
So made a prediction.
So,
oh man,
look at this. So small. So small. I can't even see it. Yeah, I won. Anyway, so Click the wrong game.
You clicked the wrong game, man.
I know. I know. I'm not going to go there. Anyway, so Block Plus is one of the largest games on mobile and it's the admon revenue monster.
Did anyone say web revenue? Oh yeah. This episode is brought to you by Fastpring, the no drama D2C partner that's been scaling with mobile publishers for over 20 years. They're fast growing, profitable, and seriously
reliable, which is more than you can say about most vendors in this space. Anyway, whether you're launching a new web store, replacing your current D system, or adding a backup plan, Fast Spring is here for you. Visit fastpring.gg to set up your web store today. Fastpring.gg. Hello, Fast Spring. Thank you very much, guys, for sponsoring this episode.
As you can see here, it's basically right now about 50 50 or 45 million DAU. It peaked around 60 million. Essentially, this game is made by a Chinese studio called Hunger Studio. When we were in China, they said to us that they had like 150, was it ML people working in the team or something like that? And
all together if I
Yeah, 200. And all they care about is retention. So the revenue mix is 100% ad revenue and essentially what the game is
is that you it's like a Tetris version or a Tetris evolving game and the long term
long iteration from Tetris just yeah
but it's basically their long-term like retention is insane and the kind of discussion was always like what is actually happening in this game what's actually beneath the surface because a lot of people had taken apart the game and essentially to see what's happening on it. But most of the things that were being decided was actually been decided server side. So they were doing decisions on the server pushing it to the game and that that was having a lot of impacts in the game. So what happened was that a couple of weeks ago, I think it's about two or three weeks ago now, a guy called Auella, am I saying that righta?
I don't honestly have no idea.
I don't know. But you're saying Garina Free is 100 million DAU compared to your 45 million [ __ ]
We're talking about last night.
Yeah. Like like why why you're so bad
still? Come on. Like move on. Move on. Nobody
I won. I I move on already.
Anyway, to this piece, right? So he made a video, he posted it on Instagram, Twitter, a bunch of stuff or X and it got like a lot of traction because he broke down what was actually happening server side on Blockb Blast and then posted it publicly. This website was also taken down briefly for a while.
So I guess yeah, Hungry Studios weren't too happy about it, but I'm assuming he used some some kind of AI to basically see what's going on on the server.
Break it down. AI break it down. Break it down. Very good done.
But the big thing is he actually was actually able to see what's going on server side and people watch
which is actually very interesting because usually
AI is not
very capable wasn't available to do this [ __ ] Now it is available now we can look at this and now we're talking about this.
Yeah. Exactly.
And the the first real interesting thing here and we're going to go through a bunch of stuff that actually what block blast is. And I think we joke quite a lot on this show about what a like ad simulators are. This is the ultimate.
Yeah, this is it.
This is it. Like we talked about like some games that show you a lot of ads. No, this is the real world OG ad simulator. So the first interesting thing that he talks about is how Blockblast the game is not the same between you and the person sitting next to you playing Block Blast because most of what's actually in the game is decided by a bunch of things that are downloaded to your iPhone or Android device after you download the actual game from the App Store. This is quite clever because basically what this is allows them to do is change difficulty and basically align with traits. There's over 700 traits that are in modules that essentially get downloaded to your phone depending on a bunch of circumstances.
But here you can see a list of all the traits. So it's everything from level config tray to is zooi Shanghai trait
is warm h warm hands trait. But essentially these are the ones that our hero from this got like assigned. So couple of things here.
So the guy wait a second. So the guy says that there's 700 traits and his install already was scripted to download only 254 of them.
Yeah.
100 were download after the install.
Yeah. Exactly. So they downloaded after install. So the big insight here is right, this is the perfect way to do stuff to make sure that Google or Apple never actually see what's in your game, right? Because if you download it from server side afterwards, that's kind of like the Trojan horse for circumventing the review process, right?
So if you download a bunch of stuff that's happening in the game after, then yeah, the reviewers won't essentially see it.
Oh, they can still go on the store and
see it there. No,
this is all server side. So this is none of these traits are in the game when you're downloading it. So basically it's only when you download the game, right? So it's not going to show up in any other
review and I look it up on the store. It's get downloaded to my phone anyway.
Yeah, but they don't do that. They review the build. You're doing it.
Exactly.
Oh yeah, I forgot that. People don't actually check.
Yeah, exactly.
People don't actually check.
That's the first thing that happens, right? So these traits are designed to basically keep you playing and keep you challenged. But also what it starts by doing is that it starts to price you. And what does that mean? Right? So uh it runs a small model on your phone that essentially decides what type of estimated eCPM that you will generate. So it looks like it looks like a bunch of things. So it looks at 27 inputs. to predict your eCPM. Part of it is country, recent ad prices seen on other
users, your RAM, the install source, and how good you are at playing the first game. So, basically what your high score was from the first game. What IO actually did was he actually changed the location from US to Brazil on his iPhone to basically see what would happen. So his iPhone first was estimated to be a $25 eCPM and when he moved it to Brazil it was $8pm. So yeah, so after that what it then starts to do is that let me find it here. No, the lethal puzzle is not yet. But uh basically after that, after it estimated what kind of eCPM it thinks it can get from you, it basically starts to try to decide what ad format you're worth. And what does that mean is basically they're deciding if this person will eat and watch a lot of interstitial ads or if it's more money to be earned by having rewarded ads for that user. So, if you
remember back to when we reviewed block blast, the main ad types is interstitial ads when you lose a level, rewarded ads where basically you get extra revives after you lose a level. And that's pretty much it. There was one more rewarded placement, Jakob, if you can remember what it was, but I those are the main ones, right?
Yeah.
Revive is the main one.
Yeah, but there was one also on the main menu, I remember. But yeah,
main menu.
Mhm. Yeah, there was, wasn't there?
Anyway, not important for this discussion,
but basically what it decides is like is this a person who's going to watch a lot of rewarded ads, interstitial ads, and if you are bas basically a high value player, you see more interstitial ads. And if you have a low prediction score, you actually don't see ads because it's a server costs to run these predictive models. So, they turn off the ads.
40 million.
Yeah.
45 million DU right now, right?
Yeah. So the next thing it decides on the rewarded ads side is basically
that it decides how good you are at watching rewarded ads. And how they do this is after you watch your first three rewarded ads, what they do is they go, okay, if you completed all these three rewarded ads, they put you into a bucket saying, hey, this guy likes rewarded ads because he's completed watching three of them. So we're going to show difficulty.
Exactly. Yeah. It increases difficulty to basically watch more for revives, right?
Also, what it does in the background, which is crazy, is it waits for expensive ads. So, basically, they might load an ad every 30 seconds to 3 minutes. And basically, what they're doing is they keep checking what kind of cached eCPM you can get at different points in time of your game play. And then essentially if they get a high value ad cash cached that's then when they also start increasing the difficulty. So either for interstitial ads in real time based on the real time.
Yeah. Exactly. Yeah.
Wow.
So they basically wait for an expensive ad. Right.
So they're fishing for ads not for users.
Yeah. The other way around.
It's basically like an army of like bot army waiting to watch ads and just wait for them to
Yeah. I want to make more money from this one.
Yeah. Here's the the code on the the finish the last three rewarded ads. And then basically if the ad loader right now is worth more than your last one, that's when they make it harder. But basically going down here to the lethal puzzle, this is what I think was the point that most people were getting the most angry about
because the rule is in block blast and I'm sure there's some kind of psychology here we have to ask Dr. Remiar about, but
basically it's right after you beat your high score, that's when they find out that you want to revive the most. So, the highest propensity to watch a rewarded ad. So, a rule
beat it, but you don't beat it as for
you don't finish your run. You just get more points in the current run than your high score. So, if you can spike because you won't beat it.
Yeah, exactly. So, the rule kicks in. Once you're 10% past your record, from then on, every new piece that you get has a 20% likelihood of being impossible to place. Wait, wait, wait, wait, wait, wait. Today's episode is brought to you by PvX Partners. Developers now have a new way to fund their marketing. It's called PVX Card. If you're scaling your gaming business and need more capital for UA, here's the deal. Your credit limit grows with your marketing performance. The higher your ro is, the more credit you will have available. It's dynamic, flexible, and scales as fast as you do. Just check out pvxpartners.com/cards and scale your UI with PVX card. again, pvxpartners.com/cards
and get your PVX cards to scale your game. Thanks PVX for sponsoring this episode.
Wow.
To basically push you to get you to the revive place,
that revive option. Yeah.
Yeah. Because you you slowly start to accumulate these impossible pieces on the bottom and then you're pretty much
[ __ ] there. I mean, if it's 20% chance, right, that means within five pieces of your high score, you're always going to lose, right?
So, they basically force force force.
Yeah.
So, you watch the revive option.
Yeah. Force to revive.
Yeah. So, also in the level mode, right, they also have a rule here that the model before you even start playing picks what part of the level you should lose in.
Mhm. So basically, it's designed to make you lose when you're most likely to watch a rewarded ad. So if you want to boost your revive so you get a higher
Yeah. You get a higher score after it.
Sorry. Is there a priority if it's like we want you to lose now because we think so or we want you to lose now or we wait for you until we catch a big eCPM C.
Uh they're both tied, right? So if if it's also tied if you have a high ecm, right? They know that also you're more likely to watch it after you beat your high score, right? So, it doesn't say here, but I'm assuming they're clever enough them to work out they can make the game easier to get to the high score.
They have the data 45 million, so they exactly know which one works better.
Yeah. Yeah. Also, there's some more interesting things below. This is how he shows a normal hand, but we we'll link definitely link this beautiful piece in the show notes cuz it's it's perfect, right? Plus, it was taken down, right? But the next thing
I'm guessing we we're also going to hear from hungry guys. We have to
Yeah. He's like, "Hello."
Yeah. We won't be living on the hungry picnic uh after this.
Living on the internet anyway. So, come
on guys. Like this is nothing new.
Exactly. Yeah. Anyway, uh so picks where you lose. We talked about this already. But then the interesting part here uh the pricing model is also that block blast measures how hard it can push you by setting the difficulty curve. All right. So one experiment tests six different timings for when the impossible hand can show up from never to 30 minutes into a session. And basically it also tracks if a player has started x amount of games within the last 15 minutes. So they make the game easier if you've had two bad games in a row where you didn't get a lot of easy pieces. And then basically there's also a predictor that says if you predicted that tomorrow you're not going to play block blast then basically the next two revives are ad
free. So basically can revive without a reward or not.
This is basic basic match three DDA logic.
We make the game easier if you see you're failing at it. [clears throat]
Yeah.
So a lot of people were super angry about this because the the comparison that
Yeah. The compare it's basically like the old it's very easy.
Yeah. Right. Yeah. Yeah.
But people are sending this this to me it's like oh my god like have you seen this? Like guys, like I work in the gaming industry for 12 [ __ ] years.
Yeah. Like I've done worse.
Like I've done worse. But yeah, like this is genius, right?
I started, you know what I started saying like what you said in in Istanbul, Jacob, like if you start working in gaming industry, you leave the moral door
and then you enter. It's like [laughter]
it's very easy. It's still not that bad like some companies did before like having gachas where you actually cannot roll something that's being advertised at being there. So yeah, we've seen
worse, but this is also quite quite like sophisticated I would say.
I wonder when this logic actually got in in full force because this is definitely not when they started for sure.
Like this this this was a process of those 2008
most probably. Yeah. Yeah. Yeah. Yeah. I mean like and of course you can't you don't you don't know this from the start. You kind of test it out
as you go and grow.
So according to sensor towers ad revenue feature which I don't think is really what I got here as well but basically April $26 million December $30 million. I basically think these guys are earning usually somewhere around a million a day. Uh probably bit more as well on certain days. So weekends
so it matches though
like Friday They're a bit below, I would say. But yeah, but
the interesting thing is here, right? Like Nebula is Block Blast. So I think they've already kind of realized that they can hit the peak on this game. So
probably they're focusing just on growing the Nebula Studio, right?
Yeah. They're focusing so much that those two games are advertised at the main menu in Blog Blast.
Yeah. Yeah.
Like Rotate Rings, it still sits right there under like main menu. Not even more games. You don't even need to click it. It's like right there
cuz now there's [ __ ] ton of games in the more games column.
That's not in the ad monetization, but the active users. Let's see how big it is compared.
It's growing. Yeah, there's the the small block, man.
Everything's minister compared to blog blast.
Yeah, except Karina Free Fire.
So triggered. Anyway,
I don't know like what the takeaways from this are, but I think what's interesting here is that
what do you mean? What are the takeaways from this? Every fight and basically you're going to start using this.
Exactly.
Yeah. It's genius. It's so genius, right? But like
I will
the takeaways I think is just like don't like dude if you're playing mobile games
this is what you get, right? Like
no I mean like this is super advanced stuff honestly.
Super advanced stuff. The super advanced stuff is you [ __ ] control the difficulty now because of eCPM. That's priority number two. You [ __ ] control the fail rate much more aggressively because you want to like force more people to forcing people to watch this is much softer than forcing them to pay which all the three games do. Therefore, we can retain this extremely high retention and [ __ ] squeeze and squeeze and squeeze and squeeze them.
So, I told my boss I made a playable [music] web with Playable Maker and he said to me, "What language did you code it in?" And I said, "I don't know. With Playable Maker, all you [music] do is drag, drop, click, and it does it all for you." So he again, he said, "What language did you make it in?" And I went, "I don't know. English." At the end of the day, [music] as long as a playable maker does it all for you, why do you care? It's just like magic.
So that's going to happen.
How How would you adapt this to other genres than casual games?
That's the problem that this is this is
this is literally like completely custom made for adbased game. and with its core because you can't do this [ __ ] with other cores. You cannot just, you know, can't do this with match three. You can't.
No, you can't. No, no.
You said there was one match three components with a free revive. You said
no, no, no, no. The match three component was like lowering difficulty based on your favorites. Exactly. You know, with match three, you are pretty much tied to the level. The level starts, you can control it midlevel. This is the whole point of this. They can control it mid game at any point at any second because we just cash in this big eCPM. You can't do this [ __ ] with all these other games. Not even sword games can do this. Not even hybrid casual games can do this. Only block can do this.
But why why not hybrid casual? Because if you have like a lot of ads in here it's you need to have a core that is is flexible enough that you can control it mid game play. Mid gameplay not before start and end. Anybody can have that. And it also needs to be fun enough and compelling enough that people like keep
playing it
like you know you can you can easily you can easily spot a game that can do this if you see that pieces or like parts of the core are not visible from the beginning.
Yes. If we take let's say they are visible from there's the same picture there same order of the pigs blah blah blah they can do this [ __ ]
candy crab or whatever gardens they also cannot do this [ __ ] like they can do degree
but not not this way
yeah how about
hexasord can do this
yeah how about big bang survivor
big bank survivor you don't see the waves
yeah but it's like too obvious. I mean, too obvious for who? I mean, it wasn't too obvious for 45 million DAUs,
but yeah, [laughter]
but those are people like you see some of those blocks and it's like so obvious that this block was generated literally for this opening here. Like it's like is this that block basically that looks
like some kind of alien because it's completely fits this completely asymmetric hole somewhere. So, it's obvious that it's there. You know what I think is really funny though? Like how majority of the mobile games that are IP first, right? They're basically designed to capture a large audience but basically convert one to 5% of people to paying users, right? And that's when the game is the best for those users. This is the opposite. Like if you're a low value ECPM users, you get to play the game without seeing any ads because you're too expensive.
It's like the biggest It's the biggest Robin Hood thing I've ever seen. But it's great because these people just you know they keep them spreading basically. So that's okay for them. But again it's because of the specific block blast core only like again maybe there are some other course but in this custom setup block blast core can only do this like you
how much do you want to bet that they started with that right they had these ideas and they're like how can we design a game where we can do this. Yes.
Yes for sure. It's Chinese mentality.
You start from monetization. Don't start from the, you know, game play.
Yeah,
game play is the monetization in this case. Literally. Yeah, literally.
I think uh we can wrap it here, but I think this was one of the better articles I've read in the last 12 months. This was super interesting. So, thank you, IO. I hope you didn't get
What do you mean? This is a must do for every [ __ ] single adbased game from now on. At least some degree.
Yeah. And if you don't do it, then you're losing money this.
Yeah. you're already behind because you were already behind like 3 years ago. Everybody knew about this,
man. I've already done this a couple of these takes on some clients. It's like, yeah, it's a lot of fun. Yeah,
of course. For us, it's a lot of fun for players, less fun until they realize.
You never know, man. Like, you don't know. Why would you care? You don't know. Like, it's an illusion. Illusion of choice anyway.
That's very true.
With that, thank you very much for coming. I hope we didn't destroy your dreams and the game plays, but you know,
somebody needs to pay the the salaries and keep the lights
tokens. The tokens. Exactly. Thank you very much, guys. See you later in the lecture or in the favorite comment section under the video because I hope this is going to this will be eyebrows. Cheers. Bye-bye.
Bye-bye.
Block Blast, a puzzle game with roughly 45 to 50 million daily active users and nearly all revenue from advertising, is presented as a case study in how casual mobile games may price players before they ever play. The central thesis is that the game's real decision logic runs server-side rather than in the app itself. A researcher's teardown reportedly found hundreds of traits downloaded after installation, outside the app store review process, and an on-device model that draws on 27 inputs, including country, ad prices, RAM, install source, and high score, to estimate each player's eCPM, or expected ad revenue. A striking example cited was the same phone being valued at roughly 25 eCPM in the United States but only 8 in Brazil.
The speakers explain how these predictions are used to decide when to show interstitial and rewarded ads. Tactics described include holding back until high-value ads are cached, making a nearly impossible piece more likely to appear once a player beats their personal best, and selecting the point in a level where a player is most likely to fail, which prompts revive offers. The hosts argue that this approach is practical mainly for ad-supported casual games whose core loop can be altered mid-game without the player noticing. They estimate the game's ad revenue at roughly one million dollars per day.
The closing portion contains no substantive findings and mainly concerns the creators' need to fund their work.
Mobile Dev Memo · 2026
Comments