2.5 gamers x Live AM(F)A #8 - Ask Me Anything

So eCPM is 100% decided by a group of factors. So algorithms for networks are all decided by various factors.
Google uses 94 factors to decide eCPM. Rumor has it that one of the factors is the weather.
So basically they link the weather with whatever you're doing. Applovin has 71 different factors.
And apparently what their algorithm is weighted towards is device quality.

So how expensive your device is.

IronSource has 61, I was told by someone.

But basically there's like seven or eight main factors that decide what your ACPM is.

And everything is tied back. Bring in the insight We're rocking those vibes Till the early daylight Mate, UA Master eyes on the prize

Tracking data through the cyberspace skies

Felix, tax, dollars like a wizard in disguise

Jackups, craft the realms

Lift us to the highs

Two and a half gamers talking smack

Slow hockey stick

Got your back

Ads are beautiful

They light the way

Click it fast

Don't delay

Ah-ha

Ah-ha

Ah-ha

Ah-ha

Ah-ha

Thank you. Oh, hello, hello, hello.
Welcome. Let me actually ping the Slack channel.
Hello, everybody. Thanks for coming for another live AMA.
That is true. Yeah, finally.
I mean, we lost one person. Yeah, to fatherhood.
To fatherhood, yeah. yeah so if you have any questions about who's this no the mustache man hey how's it going guys hello hey welcome the cheerful fin that was pretty good intro it was really good thank you yeah still cheerful despite the the horrible luck that we've had in hockey for the past few days but uh oh man i have no idea what you're talking about yeah why don't you uh why don't you move to vietnam and launch game? Yeah, I guess I have to.

Oh, my God.

How's it going?

Good.

Nice to have you, man.

Yeah, it's good.

It's good.

You came to visit us in our live AMA, which is actually 8-1.

We lost the guy because Jakob is not coming because he just had the child the child oh i don't know if that's a free pass or not but uh kind of i guess kind of is yeah i will see if we have uh if we have actually any questions because it was kind of last but yeah yeah do you want to do an intro maybe you see and then tell us what you're working on and maybe i'll ask you some questions and then we'll get someone brave enough to ask questions. Sure.
That's good.

Yeah. Not a hockey player, even though I kicked off with hockey.
So, yeah. We've been running Kigla for a bit now.
I'm one of the co-founders. Johnny was actually with you guys a while back as well, and he's his one of us as well.

Yeah, we've done ASO concept validation for about five years through Kiglab.

And just for the past sort of two years, we've been developing like iOS attribution and analytics tool that we've now been like really, really focused on for the past half a year or so.

Like actually doing things on that.

But yeah, that's pretty much it.

Look at the feelings like, whoa, wait a second.

I know, I have a question, and I just want to ask straight away off the bat.

ASO takes a special personality to do.

How much is going to get replaced with AI?

Like, it just seems to me that this is an AI task.

Right. Yeah, and I think the entire term ASO in itself as well, it's quite broad.
Like people think about keyword optimization. People think about like this generally, like more, more about search engine optimization.
The point is, it takes a special person to do. So maybe to also maybe clarify a bit.
So we've more been on the paid user acquisition side still. So focused more on just making sure that whomever ends up on your App Store page actually will be converted to an install.
And maybe even be educated on that platform as well. So that when they come from a misleading ad, you'll be slowly transitioned into understanding

what the game is all about.

Okay. But can you talk a little bit more about the actual iOS framework that you are now fully focusing on? Yeah.
Because that's interesting. That's very interesting.
Sure. Thanks.
Yeah. Well, very, very short.
kind of what we noticed was that all these frameworks that we've now put in place instead of like having access to IDF, AJS, SCAN, like they're still like frameworks that focus on kind of user level information. And when it comes to making creative decisions, when it comes to actually running campaigns you're not necessarily you don't have to dive into user level information so that's kind of our entire mentality so what we did we built up a new framework a creative level framework that allows us to we have an SDK that you put into your game or a server-to-server solution as well, which does all the kind of sensitive calculations on device and then shares only information that we can attribute on a creative level by creative on iOS.
And because we're not using scan, we're doing everything in there. We're privacy compliant.
And at the same time, everything works in real time, which is quite cool. And then for attribution, we utilize any kind of more things from e-commerce, I guess.
So conversion API on meta and things like this on the self-attributing networks to then allow them to do their probabilistic matching and so on to attribute correctly. But yeah, that's kind of the high- picture of what we did what does apple usually say about circumventing scan yeah like but we're not we're not touching any of the things that scan is kind of exactly so that's that's kind of the the thing we're we're we don't care about like what you guys specifically for breakfast.
We care about what do people that eat cereals as a whole have in common, right? So that's kind of the mentality in there. Yeah, because you don't really need to scan or just bypass the scan at all.
You know, like Facebook is doing it. TikTok is doing it.
So it's kind of easy. But yeah, we already have a question.
We do. Yeah, when doing a soft launch for a new casual or hyper-casual game, is it a bad idea to put the game app in all major countries, app stores, and then focus UA efforts in the countries you are testing? Honestly, it's not a problem or a bad idea.
There were times in the past where this would lead to not getting featuring. Maybe now it's also leading to that point where you can't get featuring because the game is out there already.
But what if instead of doing this, you just go to a beta testing on google and you can do basically the same thing just go to all the countries but then you are still in the beta so you have that tag which is called you know beta testing and you don't really get the reviews or ratings and still out there and you can test it out. So I would do that.
Your mic is scratching your sweater a bit. Oh, sorry.
I thought you were writing something. You should write.
I try it like this. I don't know.
It's fine. Better? Yeah, yeah, yeah.
That's perfect.

Yeah, that's good.

Oh, I can think again.

I'm sorry.

Yeah, in regards to that, hopefully people can still hear what you were saying.

This is perfect.

No, this is good.

Yeah, this is good.

Awesome.

But also, like, around PTC London, there was this Google for Games event.

And they were, like, actually going through a ton of, like, pre-launch strategies and pre-launch things that they have. I can try to pull up.
I have horrible memories I can't remember by heart. But they were going on about, as an example, how you can actually do different types of...
Because they have the internal testing, closed beta, open beta, all this. And you can do those separately in separate geos as well.
So you don't have to do it one for everything, but you can kind of define what your goals for a specific geo is and go from there. It's good.
I think we discussed this also on the Think Games in Istanbul, where I was with Nimrod on the stage, and he was talking about exactly this, like going from the internal testing until you actually global launch. And I use it all the time.
So that should be easy. That should be easy.
Okay, so we have then... So summarize it? Yeah.
yes. A summary is like you can launch worldwide,

but you don't need to

because you can be in different internal stage testing,

alpha testing and beta testing stages for different geos

and you're basically doing the same thing.

But you can then have actually a chance

to get featured afterwards. Perfect.
Sunbite for LinkedIn. Boom.
And I have to quickly tap onto that. Like, yeah, you might not be able to get featured anymore and I might shoot like all the ASO, like people in the leg here.
But like, honestly, what's the actual, like it used to be featuring, be kind of make it or break it five, six, seven years back. But, like, is it really anymore, right? Like, it's not the most important thing of your, like, global launch.
Even from an ads perspective, being featured is like, because the ECPM is terrible. Right.
Yeah, but you know why it's terrible because you get shitty countries.

That's why it's terrible.

If you get

the majority

of the

countries

it's tier one

then it makes sense.

But now

it's

kind of tough.

Well,

we have another question.

Yes,

so I heard

that special characters

like what's like

I guess it's

Dot

in iOS

ASO titles

could

count as double.

Count as double.

Thank you. special characters like what's like i guess it's dot in ios aso titles could count as double count as double so special characters you mean like having a dot or a coma or something uh is it the bug still is this bug still a thing i got recommendation to avoid using all 30 characters if special characters are included we haven't we haven't least, like, it's not in my radar that that would be an issue.
Yeah, this is the first time hearing it, actually. Yeah.
Interesting. Happy to chat more on the Slack channel, though, and hear more if that's a thing.
Mati, I have a UA question that I've been wanting to ask. So this scan update that Search Ads did or Apple did yesterday that was all over LinkedIn, what's that all about? The view through? Can you explain that to me? You mean the Apple? Okay, so, well, they're adding view through attribution.
So basically, if you saw an ad, then you can be attributed to that ad. But it's like these days, Facebook is almost claiming all the installs because you check your phone and you see the Facebook app icon and then suddenly you're counted as an install.
Jokes aside, jokes aside. So it's basically they would be counting now almost like all the impressions in the world.
It's an install and it's like basically driven by Apple search ads, which is obviously not true. So basically what you're telling me is that if I saw, if you run a campaign and I saw an ad for that in this Apple search ads, then basically if I download that game two weeks later, they're going to be like, yeah, that was actually us.
Could be. Yeah.
So we are experiencing some issue, but it's solved. It was a problem.
Isn't that sneaky? Yeah. Right.
I understood that for some extent it was a't that sneaky? yeah I understood that for some

extent it was available

but now they were

kind of pushing through the ads attribution

API which means that

yeah

they're part of the attribution

game now as well so not only the self attributing

networks and app lovins

but literally every network

is now like

we have another question from Sylax. Have you ever experienced CPI campaigns outperforming ROAS campaigns? Should they ever run at the same time? Yeah, 2014 when there was no ROAS campaigns.
Yeah. I guess.
Okay, good answer. Maybe scale would be the only, like if you have like very, very, very, very minimal scale, so then it could be that, you know, you could achieve CP, like better results with CPI campaigns as the attributions, like you would get more conversion events back to the networks.
Yeah, that's the myth I'm hearing always. You should run CPI campaigns to feed the VO or the ROAS campaigns.
I'm not sure how that kind of works. Then the part of the question is, should they run at the same time? Well, look, I run all optimizations, CPIs, up event roas campaigns doesn't matter and till they're kind of meeting the goals right so if they're producing results yeah why not rarely they do usually yeah usually the cpi campaigns are quite bad in terms of us cool all right so uh.
So Honor Onse has a question. Really cool name, by the way.
I like it. Hi there.
I run campaigns, install campaigns on Facebook for my mobile game. We sent impressions to Facebook, but I don't use a third-party solution such as Adjust or MMP.
I sent these events from Facebook SDK, but ROAS is still locked, verified account. What do you suggest? Well, what do you want to do? I mean, do you want to run the V value optimized campaign? If so, you need to get data.
So you need to have for value, I think it was like 100 purchases. So that is unlocked.
It doesn't matter if you have adjust or whoever, singular. You can still do this tracking with the Facebook SDK, which is fine.
So what you need to do is try to run up even optimized campaigns for purchase. And then that's going to unlock you or help you get unlocked the value optimized campaigns afterwards.
If this is around address campaigns, same thing. Only this case is that you're actually going to get the campaign unlocked quite quickly.
Because you don't really need 100 purchases for that. You just need some ad revenue coming in from the campaigns in the last seven days.
So, Omer, if you can specify if you're looking for VO, that would be awesome. We don't know.
I actually wanted to ask you a question as well marty because like uh i remember when we first started this podcast three years ago we did an episode on like soft launching and everything like or like testing for ideas of games like the really early stages and basically i think your advice back then is just run campaigns on facebook to landing pages is that still the recommended thing to do when you're ideating a game? Well, it's a good thing to be a PS over here because it was back then, because you still can do multiple things. You can have the landing page, but then you won't have the additional data, right? Afterwards, after someone clicks on that landing page because there is no tracking.
Well, I work very closely with Geek Club and we worked before because you can create that landing page or fake store page in Geek Club as well and then just use that as a link. Oh, my camera is shaking.
And then what is really interesting and important to do is to run actually a conversion-based campaigns because you want to kind of simulate the CPI campaign as much as you can. And with clicks, you are not doing that.
Not even close. Not even close.
Echo echoing 100 like with clicks you get like um

the conversion rates like on the actual landing pages uh for instance with nothing like you would

in the real world you get completely different audiences um so that loop is important and then

what we often do it's still a thing uh in in short like uh but also depends a bit on the stage as well so um i guess now there's a lot of like more opportunities with google as an example so if you have a like someone of a beta out so you can actually then go for google for these things where previously was only we were kind of the only option but in the very early stage you should still should do it and then after the landing page you can have like a survey where you can you can ask like motivations and actually start to pay like build up your player profiles and understand like what type of users like are actually coming through or do they match emails yes exactly so what would you be your advice if you're starting you don't you have a couple of ideas for making a game. What's your best advice? How do you start? Just test, honestly.
And now with AI as well. The thing is...
When you say test, you mean like test running it on Facebook or test setting it up or what do you mean? Yeah, exactly. Just test running it on Facebook.
Then if you don't have the resources, so even just doing it on Facebook

is better than not doing it at all.

And if you have the resources,

so use us, use Splitmetrics.

They provide more or less similar services.

So just do that so that you'll get like the good picture.

And well, you guys know this even better.

Like with AI now,

like you're able to actually get to that testing stage

a lot faster.

Yeah.

And the thing I would say is that

Thank you. you guys know this even better like with ai now like you're able to actually get to that testing stage a lot faster yeah and the thing i would say is that if everything is equally unpolished so then the test is valid so you don't have to like kind of make the most polished final art whatever if it's unpolished enough like equally it's it's a valid test for the case.
All right.

So Sylax is back with another question.

What up, what up?

How are companies scaling tier three regions?

From our experience, the MMP feels like T3 region is not profitable.

Is there something we're missing?

Well, of course, it's not profitable when you pay five cents for CPI or for installs and then seven cents for attribution. Yeah.
I remember last game, back when I was working at Rovio, I was part of team launching was Dreamblast. Back then, we were trying to do stuff on India.
and uh that ended up costing us more like than like actually get any any like essentially ad revenue from there uh i remember you guys have been talking about this too earlier where i think felix you mentioned something about having great success where the timing was right and then there was like a it's every four years and it's during the cricket World Cup, which is right after the Kabad-D World Cup. And then for like for two weeks, basically Indian CPMs become, I want to say Romanian, but not really Romanian.
Let's just go with maybe Egyptian. Okay.
Yeah. But like back to the actual, the tier three regions, non-profitable.
Yes, of course. That adds a lot of burden on the CPI.
If this is the case for attribution, what we did or were doing with different gaming companies, it's just turning off the attribution for tier-free countries. You don't care, right? Exactly, I don't care.
You can just use the SDK of Facebook or Google Firebase and just turn on the MMP. And then use the solution that provides the UA channel because that really helps with the CPI and LTV equation.
All right. Makes sense.
Oh, sorry. Yeah.
Facebook for us campaigns via ads. We don't have an option to send ads impressions on Facebook.
Well, it's like only half of the problem because we don't have the full picture. What option are you talking about?

If you just need to send the event,

which is called ad impression event,

and you send it either via Facebook SDK

or you send it via MMP, period.

There are different methods

or just guidelines how to do it

on Upsly or Singular and Adjust, But as soon as you have that event sending, which is, for simplification, let's call it ad impression event in the MMP SDK, you need to pair that with Facebook ad impression event and send value as well. You do this, you're fine.
I just did it with three different games in the last two weeks, and it works. You just need to set it up correctly.
If you want, hit me up on our Slack channel, and we can discuss. All right.
Good answer. So Martin from M&M Games.
Nice. Everybody seems to be discussing UA and monetization segmentation in hybrid games these days, but nobody gives any insights how it's done.

Any idea advouses how and where to start.

Things like having two monetizing segments in your game, the ad monetized segment for non-paying users versus IAP monetized.

One, for example, how do big casual and mid-core games usually do that? Ooh. That's a very good, very good question.
All right. Take a sip, and then you can talk for 30 minutes.
Yeah, I'm going to talk. I'm going to talk.
Okay, so basically the key to actually start doing segmentation, right, is that you just need to have IAP revenue and ad revenue. So like if you have 100% ad revenue, it's better to segment on things.
We had a really good podcast you can check from Anatoly from Bernie Games, where essentially he kind of opened up the kimono a little bit, not too much. But basically, the hint that we got from that conversation is that they

segment users based on eCPM which was a bit counter intuitive because essentially they

actually decrease the amount of interstitial impressions if they have a high eCPM because they want to have a user for longer so basically what they had built in was logic basically that checks every time an interstitial impression is served and then if it's above a certain level then they decrease the frequency so it's not after every level but every two levels so basically that's logic you need to put into your code on your ad serving and when you put in kind of the the the serving logic when you have an interstitial in your game you can put that on some code. So it's like an if and then statement where you pull the data from your mediation platform.
So if ECPM is higher than $40, don't show interstitial every level, for example. Then kind of the gold standard, which I actually don't know how to do technically.
Maybe any one of you guys would know. But basically that is if you have a little bit of ad revenue and a little bit of IAP revenue together, so it's hybrid casual, basically that you segment the users based on the type of user acquisition campaign you're doing.
So basically if Mati brings a user using an IAP campaign, that means that that user is a premium user, right? So then you usually don't show ads for the first three days, let's say. And then if a, you know, organic traffic, which we said has bad ECPM, or if a ad row as campaign brings a user, if you see that basically the string that the campaign that you're getting from the MMP comes from that user, then essentially you don't, you show ads as quickly as possible.
Well, that's the technical side, Felix. You have the source and you have the attribution, the campaign as well.
And based on the campaign, you can show different ad experience. How do you guys do that on iOS? Ooh, that's a great question.
Same as plugin here. Nice.
Yeah, you can try our attribution and analytics solution didn't see that coming but it's amazing but it's amazing because yeah how would you do that if you don't you can't use it on you can't use it on user level but you can do it on campaign level. Exactly.
And even creative, right? Yeah. So each creative can have a different experience if you want.
So, okay. Nice.
Yeah. There you go.
Easy. Easy.
I mean, obviously, I think you were discussing this campaign ad segmentation multiple times.

I feel like that it makes so much sense.

Yeah.

I mean, like there's a company in London also that's doing some interesting things

where basically they're using AI to determine how many ads

you should show to a specific user based on in-game metrics.

Who doesn't use AI these days?

I mean, I am artificial and I am intelligent. Yeah, because I'm Botox? I don't know.
Or is it because I'm clever? Do you want to see something funny, actually? I do. Of course you do.
Of course you want. Let me bring something up quickly.
Where do I have the...

Here.

Anton from Sane Games was indeed the first one to talk about that, Martin.

I'm glad you took something away from that podcast.

Glad you enjoyed it.

Can you hear the sound of the guys?

Again, the previous UA Killer articles were super successful,

so I started writing a UA Killer tip segment. I will keep sharing these tips.
Don't worry. Keep reading the tips.
That's not you. Of course, it's not me.
English is way too good. Yeah, the English is way too good.
That's the problem. Yeah, that's the problem.
But this is my AI avatar, basically. So I created myself and the lip syncing is insane.
Of course, the English is too good to be true, but I'm working on that. I'm working on that.
They didn't clone my voice that much. No bullshit and brutally honest.
Yeah, it's like, obviously, it's not me, but still. I will definitely nail this because, as you might have noticed, I'm talking how this year the ua is more about creatives and how people need to be able to write script to a little bit of story here and there and obviously make a video so i want to be you know i know i i i really shoot a lot of videos when I was on vacation.
And even now for the playable trends and for some games, like for a game which is Modern Combat Mayhem, I recorded myself playing with toy soldiers and stuff like that. Man, it's killing it.
That creative is really working well so so i mean i'm doing all this crazy stuff because well this is what this is what you need to do just just say yeah oh sorry go go go no sorry i was just quickly saying the yeah and kind of the pattern that we've noticed both work and then when we look at like what's happening in the market is that like you have that like uh normally horrifying ctr ad like kind of be in

the core of it but then you add something in the beginning which is like a very solid hook

it could be even like a meme if you don't care about copyrights uh but uh like that's the

interesting thing for me it's like as soon as like these these video models and we've been trying

this out it's not quite there yet but soon it will be so you'll be able to create like very actual meme like usage like that looks something yeah completely out of like the the funny fail army videos or whatever that you have seen and then you can put that as the hook you have all the copyrights you can scale that up you don't have to care about your like uh legal department getting tons of messages it's crazy yeah yeah it is yeah i wanted to ask more about the question we talked about martin asked about the segmentation on ios and stuff like are you allowed to without naming any client names or even if you can like what they're doing and how they're doing it because i'm just very curious i was i was that was a bit of a shameless plug there because we actually haven't done that too too well yet but it's a technical capability yeah so so yeah so essentially just to paint you a big very high level technical picture uh it's an old bullshit like podcast so i can't bullshit you guys yeah We've done something we haven't yet. Exactly.
Followed up with a question that came clean.

But, but uh but but yeah technically again what we do is that we're able to like we send back to the sdk a creative token uh that like essentially tells from which creative specifically the user came from and then that start locally and then you're doing all the stuff for that creative token, right?

So you can do whatever you want with that.

We're able to share that.

We have the ways of sharing that directly with you.

So if you want to build logic within your game to actually have different things happening

based on what type of a creative token it is,

that's something that already technically we allow you to do

because of the entire data plot that we have.

Thank you. based on what type of a creative token it is, that's something that already technically we allow you to do because of the entire data plot that we have.

But no one's done.

YOLO with a heavy question.

Wait, did we talk about the previous, like the future hurdles

or issues we see coming for the IDFAs and cookies

like are removed to track users and devices? Like is there trouble on the horizon for tracking? I mean, for the IDFAs and cookies, like are removed to track users and devices.

Like, is there trouble on the horizon for tracking?

I mean, duh, obviously.

Upflyer's valuation is looking very big.

Well, look, there is maybe the Google sandbox privacy,

which was supposed to kind of come,

but I'm not sure when it's going to even happen there's always something cooking under the under the hood for for privacy yeah but i mean at the same time they they just like re like define their policy in terms of ip address being used in in terms of thing and printing as well and and it's funny because you have like all these regulations saying one thing on the other side and And typically they're saying no to everything, right? But then truly what kind of defines where the industry can go is what the platforms say, what actually Google and Apple say, honestly. And now Apple's been kind of vague when it comes to all things probabilistic, right? And they've said, no, no, no, you can't do this.
But then on the required reasons API, it still has stuff that's very... Well, IP isn't there.
And now Google just said, okay, this is actually fine. This is sort of the limit what we're going to put to fingerprinting, I guess.
So hard to tell. Can I put on my tinfoil hat and share my theory about iOS here? Yes.
So basically, Apple was getting worried that Facebook was taking over and the iPhone would basically meta run because they were doing all the UA on iOS, right? So that's why they kneecapped by basically taking away the IDFA, right? So at the same time, Apple search ads scaled up, right? A lot. And in a perfect Apple world, you'd be paying all your UA money to Apple to get users to download your Apple ads, right? Because 12% of the Apple income, like the revenue comes from the app store in terms of IAPs, right? So they're dependent on people finding new apps, downloading and buying stuff, right? The trouble was that Apple search ads, you can't scale campaigns there, correct, Matej? To a certain degree, because there's not enough ad placements, right? It's not enough.
It's not about ad placements. It's because it's a different channel.
It's intent-based. If people don't really search for certain keywords, you don't have the impressions.
That's the thing. Yeah.
So my theory is that they still allow fingerprinting to occur on iOS because otherwise the IAP revenue on the Apple platform would go down by so much that it would hurt Apple's bottom line.

So Apple, the logical move here would be to build a DSP and open an audience network.

So Apple search ads would be inside of an ad stack because that would take away the intent problem from Apple search ads and basically start competing on the open market. Then since they allow basically tracking for Apple search ads, that's when you take out probabilistic stuff and that stuff in its knees, meaning that everything would just go to Apple search ads because they have an audience network all right let's see what you hear because yeah all right great theory yeah i mean like a ton of it makes sense but uh um well at the same time uh in meta like they have three billion people using their platform like one of their services a month that's such it's such a vast like audience that are like completely pushing that outside of like profitable scaling and marketing would like it's hard to think that that would actually apple could do that and secondly for them to to kind of be able to completely cut down probabilistic uh like technically that's not they can't like there's like well practically because there will always be a possibility for the developer to do some server server side of things together with with uh the networks and and there always be these options um the only way would be to basically loop all the traffic coming in and out of the games through their private relay, which essentially would cost them probably more that they could profit out of owning the entire ad space, right? So that's my take on it.
But yeah. Yeah, look.

The audience network and the DSP,

I think you were talking about it a long time ago.

2022.

Yeah.

It's leaked.

I know.

Yeah, yeah, yeah.

You know how long it takes to build a DSP?

Two years.

Four years.

Four years.

Four years. Four years.

That's what I've been told.

From scratch, it takes four years to build a DSP. It's coming.
Okay. Then we have one more year to go.
Exactly. Then we can discuss this.
I don't think it's going to happen. What do you mean? I don't think they're actually building it.
Is this based on the truth? Based on the hired 30 people, all demand side platform related in 2022. I don't know if they're still there, but I haven't seen anything about it yet.
Exactly. So we don't know.
It's a good theory though, admit it. Of course, yeah.
It sounds and feels good on paper. And honestly, I think that can happen apart from the fact that they would be able to actually pull the plug on all things probabilistic.
But definitely them becoming a DSP essentially. Would we care then? Because then we could just buy all our ads with tracking from Apple.
But still, again, 3 billion people on Meta's like network network it's a lot of people yeah

I don't think they're gonna have

even close to the

same amount of

people like

Aplavin has on max

180,000 games

which definitely helps with Algo

right so

even if they build it in 4 years, it doesn't really mean that we are going to be able to use it to a certain scale and certain quality because then you would still need to have a purchase optimization and even optimization, ROAS optimization, all of this fun stuff. which some of the bigger ad networks just added

recently and they're on the market for a long time.

So it's not that easy.

If it's happening, it's going to be 2030, 2028, let's call it.

So 4AQHA2, he said it's true.

He's building it.

So it's confirmed.

Hi.

Hero. I'm here.
You heard it here first. Yeah, you said it's true.
He's building it. So it's confirmed.
Hi. Hero.

You heard it here first.

Yeah, you heard it here.

Yeah, yeah, yeah.

Speaking of, by the way, what do you guys think about alternative stores?

Like now with specifically on EU, like regulations pushing it there.

Now alternative.

Aptoid, I think, just actually opened up publicly yesterday i think or so yeah well i mean more stores more um traction on the on the ua side of things i can uh and also pretty good deal for the developers so i don't honestly hopefully for everyone, but I don't see that as easy as it sounds again. Huawei tried to do this with a lot of money and they failed.
Some network effects are really big, right? So let's see. Exactly.
On Huawei, it was already even pre-installed in all the devices as well. And yet they still couldn't get big games on board, even with the rewards and incentives and everything.
They were paying people crazy amounts. There was rumors that Playrix got offered millions to rebuild the game on Huawei stores and stuff like that.
And it's still like that meme, try to stop making a thing. It's not going to happen.
Yeah, but still, I mean, I can see a world where this gets some traction, but if it's going to compete with App Store or Google Play, most probably, no. The same thing with Huawei or Samsung.
It had and still has some DAU and some games but if you want to scale there it's not possible, it's just not going to work Right A fun thing regarding those I just remembered actually now that Felix mentioned Huawei but back when they were really pushing this it was 3 or or something. So, of course, they don't have any A-B testing or anything available on Huawei App Gallery yet.
So they reached out to us and we actually ended up building kind of for specifically the European and the Western market, like this sort of official A-B testing platform for them. So you wasted your time for six months.
100%. I won't tell any revenue figures here, but you're going to love this.
Guess how many tests did we actually end up running through Huawei App Store? Four. Thales got it right.
100 points to Gryffindor.

Like zero.

Oh my god.

Okay.

Coming a mile away.

Oh man.

It's crazy.

Wait, wait, wait.

We have questions actually.

Yeah, YOLO Games.

This is a painful one to read.

Is there anything to be done

when your App Store account is terminated for getting rejected for the same reason many times and the issue is still present? Then the follow-up question, is there any marketing thing you can do for a Steam game? Yeah, but it's different. I'm really sorry to hear that, by the way.
That sucks. One, it's very hard to get, like, unless you're at a certain stage, it's very hard to get, like, actually on the line.
And you have to be kind of a supercell to be able to actually get on the phone directly. And also the rules are not fair, to be honest.
Like, there are certain guidelines on the app stores that you have to follow. But no one follows those.
And, of course, like, we've seen them over and over again like if you have like a forex game and and you're showing actual gameplay which is essentially maps with a ui that has a lot of like different red notification icons in it it's not going to convert as well as something that actually portrays your characters nicely uh so and and and that the second one is something that is not allowed by Apple's guidelines. It should just be screenshots from the actual game, right? And we've noticed that like during, like when they get like a lot of, I don't know why it is, but like could be, now this is my tinfoil, okay? But like before the summer, we see a lot of rejections happening as well.
so it could be that they have like some eager like trainees coming in or whatever that actually go through the list like item by item yeah and uh yeah like we've seen things get rejected because a corner of a app review video that had a play button like was missing of the play button whereas at the same time you have have cinematic trailers that are good. Yeah, all the AI bullshit stuff that we see on the UASA and then it's in the store as well.
How is that even connected to the game? Also, the freezing families in the video and then on the store and the first screenshots, no rejections. I mean, come on.
That's how it works. It's just, you know what you need to do? Just resubmit, no change, and you're good.
Does that work a lot of times? Yeah, it's so many times. Really? No, it's like, it's unfortunately, but it's, I guess, it's just the rejections like these, it's just random and someone really eager to make it and is like kind of doing this and that's it.
That's pretty much it. Yeah.
But the Steam games, the marketing for Steam games. Do you have any shameless plug, mister? Because what I heard when I worked on a few Steam games, like, you have a solution how to run some interesting stuff on the Steam side.
Is there anything you don't solve? Huawei A-B testing? We've tried a lot of stuff. And I'm happy to say that on Steam, the amount of tests is not zero.
Yeah, it's not zero, yeah. Because I ran at least one, so it's not zero.
Yeah. And we have some testimonials as well.
So there must be more. That's good.
No, but yeah, like honestly, like for the question itself, when you think about it,

players are players, viewers are viewers.

So kind of the same methodologies like applied to whatever in certain extent

happens in mobile marketing

can be applied in Steam as well.

The differences are then in terms of the technologies

and the stack that you use.

What we've done is that we have like a testing, like you can test the actual store layout through our like the same landing page formula that we talked about earlier here. So essentially driving from ads directly to landing page, we use add to wish list rather than like actually the actual value to get more conversion events to better understand like kind of how the page works.
But that way you'll be able to test out what page actually converts the message in a good way. But in terms of the whole marketing for Steam, so again, the campaign logic is the same.
It's just different optimization so i use facebook i use google

i even ticked i even use tiktok there's their reddit kind of ads that i i tried before but i need to i need to work with reddit ads a little bit more because i was treating it almost like a facebook where you just go broad and then let algo to do the magic well reddit is not there yet So you actually need to do a lot of targeting on that side.

But what I'm doing usually is either a landing page where I track the buttons and basically, like you said, add to wish list and send it back to Facebook, Google, and TikTok as an event and then optimize for that in the campaigns. So it's kind of,

it's almost like an installer up even optimized

campaigns on mobile.

So nothing

that different. You just need to know

that you run

the landing page instead of

Steam because there's

basically no Steam,

very limited Steam attribution.

You can use the UTM tags to just identify the campaign. But the event optimization on the landing page brings you better quality of the users rather than just using clicks and sending people to Steam page.
So you add one step in between the actual Steam page and your ad, but it's worth it. It's absolutely worth it because the quality increases dramatically.
Yeah. And then you need to do a landing page conversion optimization and landing page testing.
And, you know, it gets a little bit more complex,

but still, I remember those days.

It's basically like you're running a landing page

for any other web-based games.

It's, you know, you run short landing page,

long landing page, different buttons, layouts,

visual designs, and all of the fun stuff.

So, but yeah, the campaign logic, basically the same. All right.
Felix. Martin was wondering if you heard about any games doing these segmentations and the numbers of different segments present in the game, because it's also quite a big amount of work to do for a studio.
Look, like if you have a small studio and you don't have that much, like the easiest thing is just that you basically change serving logic by country. So like I was operating a game at one point that was making 50 to 80K a day.
And basically it was a very lean team. and basically all we did was just increase in geos by 25 the interstitial show rate so the timer and basically that's segmentation by super simple and then we just looked at the retention right so like if if you don't want to make it complicated you can just change that like for example go to indonesia and increase the amount of interstitials and see if the retention is the same.
Like, that's still segmentation, right? So, like, most games that are ad-driven that I work on, they do have some sort of form of segmentation. But it doesn't have to be hard, right? Like, it can literally, that is segmentation.
In Indonesia, just show more ads. Like, have a lower cooldown period.
In India, in wherever, like, that's okay, too. That's segmentation.
But the thing is, once you start, it's usually easier to keep going with the harder stuff because you know it adds value. So just start with something simple.
Like, yeah, Indonesia. Like all these ones that work, they don't care as much about ads ads and see how it goes.
Yeah.

Oh, there was a message deleted.

I hope you will repost it.

Okay.

We are small.

Oh, fuck.

Now it's just jumping.

Okay.

Now it's... Okay, retrack it.

Anyway, so we have that

two apps being relatively equal,

audience, gameplay, retention, etc.

Why could one game have only half ECPM of the other? Would it be because of placements? So basically, ECPM is 100% decided by a group of factors. So algorithms for networks are all decided by various factors google uses 94 factors to decide ecpm rumor has it that one of the factors is the weather so basically they link the weather with whatever you're doing uh applovin has 60 uh 71 different factors and apparently what their algorithm is weighted towards is device quality.
So how expensive your device is. IronSource has 61, I was told by someone.
But basically there's like seven or eight main factors that decide what your ACPM is. And everything is tied back to the performance that networks see of the ads that they show in your game on a seven-day rolling basis.
So if the performance from the ads that they're showing is not converting, giving them money, they will pay less for your users. That's all it is.
Simple as that, right? It's not a conspiracy. It's just capitalism at work, right? So what are the seven or eight main factors that contribute in the most weighting? And those are the ones that usually every network has access to.
So the first one is geo. Very simple.
Are you in the US or are you in Angola? That will be different ECPM that you're going to be paid out for those users because of the purchasing power, right? The other thing that is also given a lot of weight is, are you running a lot of UA? And did the network that sees the UA or sees the user in the mediation, did they also bring that user? So if Mathieu buys a user on Mintegrel and then basically it shows up in my app and then inside of the mediation platform, Mintegrel sees that user, they have data that tags that GID or IDFA with a bunch of data in their platform. So that becomes meaning that they can bid more accurately.
So higher or lower for that user. So basically, are you doing a lot of UA or not? Then also, are your users converting a lot? So basically, do you have 80% IAP? Are you showing those purchasers also? Because that's valuable to, let's say, a Forex game that will bid a higher ACPM.
Then you get into the other things, right? That also has an impact. One of the biggest things is also the recency effect.
And what does that mean? That essentially means that networks assume that we're all stupid because we basically lose interest over time.

So basically that means that the first impression that a user sees per day is called first look. And that is worth more because essentially users have more attention.
so basically the value of each eCPM per impression on an impression level decreases

as you show them more throughout a session or day so an idle game which shows 18 impressions per day the 17th impression is going to be worth pretty much like so much less than the first one some things you can do to kind of spoof that that uplovingving tries to fight. But basically, those are the main factors that decide what it is.
So in summary, what's happening in your two games, one converts better than the others for user acquisition campaign than the other one. And that's where the delta in the ECPM comes from.
Or maybe the weather. Maybe the weather.
Nice. So we have two more questions, but I will start with this one, which was retracted, but now it's finally up there.
So we have a daily budget of $1,500. Daily budget.
It's not bad, actually. Does it make sense to test different UA channels or stick to one? We are using Unity ads currently and mediation is level play.

We create puzzle brain games.

Felix, well, what did I say?

Felix, as a second best manager or UA manager on the podcast,

what do you think?

Should they use only one channel or more?

You need more channels and you should maybe switch

your

radiation, yeah.

Yes, go and I will

I was just about to say that there was a rumor

going around that said

according to

it was something that leaked out of something in meta

that showed that if you

spend 1,500 bucks a day

which again, it's a good number

but it's of course

that's up to half a million a year. No, well, sorry, too late.
Doesn't matter. Yeah, anyways.
But according to that, you would have been in like a segment that Meta called like the top 5%, which that was actually going around in a lot of these lag groups, not sure if that's the case that you would actually be part of the top 5 spenders in Meta with 1500 a day ok, interesting but the thing is with 1500 a day running Unity only good luck Obviously, as you have the level play, which makes sense if you run Unity. I would run Iron Source rather than Unity in this case, but okay.
You should definitely try other channels. I mean, there is the UA diversification topic that I'm talking about all the time.
I mean, what if Unity dies? You are screwed. So I would take 20% of that budget and try Google, try Facebook.
Uploading is not going to work, but then other channels as well, MintyGrel. And I mean, slowly, step by step, diversify this.
So you actually have that 1,500. What's going to happen, I believe this is going to increase because you will get better results on the other channels.
So you will have more money on your bank account and you can spend more so definitely makes sense to test others if you can say what kind of monetization you have that would be also helpful because that can allow you to pick uh the right channels for you yeah is it uh is it like puzzle brain games that are like impulse where it's 70 30 or? Or is it like puzzle games where it's like 100% add games? How is puzzle brain games like impulse? That is a brain training game. I don't know if it's a brain training game, but it is a game.
It's an app, which is a game. Yeah, yeah, yeah, yeah.
I know, I know, I know. Jesse, you be the arbiter of truth here.
Who do you side with? Me, who's right, or Mati, who's wrong? I'm going to be a very good Finnish politician here. On the App Store, I think the category is undereducational.
But in LinkedIn, those are gold games.

Yeah.

I thought you were going to say, like, in a Finnish way,

like, it doesn't matter as long as Russia doesn't invade.

Right.

It's all ads driven.

All ads driven, yeah.

In that case, Facebook AdRoss campaigns can work.

Also, Google AdRoss campaigns can work. It doesn't matter what kind of optimization you have, mediation you have.
So I would definitely go there as a next step, at least. Then we have another question.
I think that's going to be one of the last ones. I've heard that Upload is much better at Adroas campaigns compared to others for an idle game.
Is this true?

If you're on max mediation, then yes.

If you're not, then you can't run other as campaigns.

Well, for idle games, I think that was the case for a long time.

There was a very large portion of idle games running on Uplavin

could be quite a big success.

So I would say, I would incline to say, it's kind of true. Because yeah, I would also run other channels for AdROAS.
Because Facebook has it, Facebook will soon implement Day7 ROAS, from what I've heard on ad ROAS so that can improve some numbers. Can I confirm or deny because I'm never on a UI campaign? But this has been a lovely episode.
And that keeps giving me missing object notifications. Okay, so last comment.
Meta keeps giving me the missing object notification for meta ads how do I get past this log out, log in, first thing second thing, create the campaign from scratch, don't duplicate anything and also might be just a bug that disappears tomorrow because that's what usually happens to me all the time yeah so yeah and there was so i i kept this for for the end because there is a person hi guys i got here late can i know what services geek lab offers so oh love it yeah there you go well this is this? So how much do we have? 10 minutes, okay? No, I'm just going to keep it very short. Thanks, though.
But yeah, so first, if you're about to launch a game, we help you with concept testing, validation, understanding who your audience is. And then when it gets to actually acquiring users, specifically on iOS, where you have issues with attribution and analytics, we'll help you solve all those without scan, without actually even MMP as well so it's completely outside of that.
Nice. Awesome.
Yeah, thanks for coming. Yes, it was great.
We will... It's a lot of fun.
Yeah, we would love to have you again at some point. Yeah, I actually learned shit ton from Felix and you, as always.

So thanks a ton for having me as well.

Awesome.

Also, dear listeners, thank you very much for coming.

We will do this again next month, actually,

and I will let you know a little bit more in advance because I will be at GDC,

but then after I come back from GDC we should do it so

it can't wait like end of the month

see you there, thank you very much

ciao ciao

okay

I ended the stream on YouTube

now I'm

stopping this here Thank you.