Ad Creative Testing for Mobile Marketers (Meta, TikTok, Unity, Applovin, Google)

So pretty much what works on Facebook not necessarily works on on on Applavin because for Facebook you need hook you need very strong um kind of gameplay or middle of the of the creative but then on applavin you need a very strong call to action at the end where on Facebook you don't really need that because 80% of the people don't really finish the video even if you have like one minute long video like 80% of the of the people Maybe maybe we'll watch like half of it and that's the best case scenario on Applavin is the other way around. Everybody needs to watch the whole minute and then afterwards even you go to playable and that playable also has like some best practices.
Hi everyone, welcome to another episode of ROI 101. I'm Roman from Tangent and I'm joined by Mate. Hi Mate.
Hello. Hello. Hello. I'm super glad to

be here again. As I said, I'm not in the driving seat, so I'm going to enjoy it way more. Thanks for thanks for inviting me.
Yeah, absolutely. And I would like this to be a sequel to a previous video from another member of Two and a Half Gamers from Yakob. And Yakob has shown us how to create thousands of ad creatives. And now I'll ask mate how to test those creatives because I don't think anyone has enough money to spend on those. But before that, this is just a teaser. I'll ask mate to talk a little bit about himself just in case if someone doesn't know who he is.
Yeah, thanks. Uh so very briefly because I don't like uh when guests are talking about themselves for 20 minutes. I've been in the gaming industry for 12 years. Um running a consultancy. I have 10 motion designers in my creative team as well. Have playable team and I also build a playable maker which is no code

playable tool and hosting the two and a half gamers. So uh that's pretty much it. Without further ado, we can we can dive into the topic.
Absolutely. Uh we'll leave all the links in the description.
Thank you. Yeah, appreciate it.
Just go ahead, man. also also funny because like how can game designer talk about how to produce thousands of creatives like
he's learning a lot he's learning a lot on the podcast which is great
and and Felix too like I'm seeing like him talking a lot about creative but yeah I guess that's a leverage that's a leverage
yeah because what we we have this new KPI on the podcast which is the like number of creatives produced by a game in the last 30 days
and it's becoming pretty crazy So some of the games like uh we always discuss with Mayong and Kingshot and Royal Major Royal Kingdom and all the big companies basically have th tens of thousands of creatives per per month I guess at least

in the last 30 days. So 23,000 creatives like that's a lot. That's a lot.
And I don't think like you can test them properly anyway because the thing is okay so I can I can actually share this screen so people can can visualize this a little bit because I'm pulled this from my creative bible which uh which is the ebook that I have obviously and look it looks really complicated but it it's actually not. And I'm talking in the in the ebook about different things and also like how to test creatives and with this amount of creatives you would need to spend ton of money right obviously. So I have multiple frameworks and I don't think there is like a silver bullet for for creative testing for each company because I have let's say I work with 10 10 companies and 10 companies have very different approach to creative testing and I'm not saying one or another is

better or wrong. It's just maybe suitable for different needs. So I developed this kind of pre-evaluation of concepts in like in house with my team which is kind of funny because then it's I call it boring test and so what test. So we're looking at I'm looking at the creatives that are produced by my team and looking at it and then like after a couple of seconds like ah I'm bored like okay so I mean like this is not really having any emotions or anything. It's like it's it's this doesn't really sends the message across or there is the the trigger like so what kind of uh pre-evaluation test where I'm looking at the creative and then I just I yeah I just ask myself like okay so so what so what now like what should I like what should be the message of the creative what should I do and this already like kind of pre- evaluates let's say 50% of the creatives so we can just go back and

uh and kind of redo do them. I haven't really done this before and I was really against it, but I've seen this work well because not every creative is, let's say, passing the the sniff test or it's quality enough. And I'm saying enough because I came to a conclusion that the creatives don't really need to be 100% polished, but it needs to be good enough. And good enough means, let's say it's 70% polished or it's it's done as a concept and if you look at it you're not bored sends across the emotion and you're not saying like okay well I don't understand like so what so that's like that's the first test with 23,000 creatives I'm not I'm not sure if you you could do it but with let's say couple creatives per per week you can you can definitely do that but then uh how to do it on like channels because then you have Facebook Google Tik Tok Unity applian

and how to do it on different channels. So, let's say uh people do this on
sorry
on Facebook. Yeah.
Quick question. Could you please zoom in on the part that you Yeah. Yeah. Yeah. Just because I think I'm following Yeah. Yeah. So, here should
So, yeah. Here we are. Yeah. Here we are. I was just talking about this and now we're moving into this kind of uh section because this is important. So I'm running testing on like all the channels and when I talk to companies they do it also on Facebook especially which is like makes sense. So so what are like the different different frameworks on on Facebook and you know what like I I can actually I will go back to the to to the diagram. I'm just going to use this because this is just like the winning setup for for Tik Tok t Facebook. This is what I use on on Facebook and it kind of goes well with different number of creatives. So for

example, the first methodology which is the IPM setup. So what do you do? you have um an creative testing campaign where you run a testing campaign in let's say India or Philippines or Brazil or wherever and then you pick winners based on the IPM so installs per thousand impressions and then you take these winners and then just go to business as usual. This is actually useful when you're producing maybe 10 creatives per week because that's okay. But usually the problem is the top creative or the winners are not picking up traffic in the business as usual campaign because it's different campaign optimization. So okay then in that case I developed this multi-step kind of framework where you just go and take those winners and then just test it in the app event optim optimize campaign

which can be let's say purchase which is closer to your business as usual campaigns because you're optimizing your campaigns for for veils or value basically for payers. This is good when you're producing let's say 30 plus creatives then I like ah but it's it takes long and more more in more kind of uh creatives you need more money so as we said like if we do thousands like h it's quite tough then I was like okay so what what should I what should I actually do instead so we I am starting to kind of go come back to this value optimized uh campaign where you can for example on Facebook you can have like 50 50 creatives per adset easy so you can just have this is pretty good like range of for any type of creative right so and it's simulating your business as

usual campaign optimization which is again you're using worldwide or tier three tier four countries as like a batch of these and then uh you can have loads of creatives in here. So let's say I have 50 creatives per adset and then you can have pretty much like a lot of different adset or ad groups in within the campaign. Again this is this wouldn't be really cheap exercise but when you are producing thousands of creatives per month or per week or whatever you are in the stage where money is not a problem. So this is a good problem to have and and this this creative um testing framework with the worldwide and value optimized campaigns it actually started producing pretty good results for different like games in recently. So then you have two options basically either just take the winners from this campaign and just move them to business as usual or actually scale this campaign

if it delivers strong results because this actually happened to me before. So this kind of this setup is is pretty good for loads of creatives. So I would say like if you're moving anything between hundreds and thousands of creatives like this is a good setup on Facebook especially. Then if you know like because there's so many different um gaming companies at a different scale of creative production obviously because not everybody can do like hundreds of creatives. If you do one, two, three per week, and that's what I do on some of like smaller clients or smaller accounts, every new creative that I that is produced, I'm just just putting it into the business as usual campaign because that's my main campaign. I don't have enough creatives for creative testing. I don't I don't have enough budget for creative testing. So, let's just ship it back. So, and that's that's pretty much like for like the small from small size to big size especially on

Facebook. But if you need like I said like if you need to have let's say um a creative testing campaign. So the example right here is the here the the go one that kind of simulates or target tier one geo which simulates the the optimization you usually have on the like business as usual setup because that's that's very important. So that's what people need to think about is to how to simulate the best the business as usual campaign but on the side on the side that's like very very important. So this is kind of like the the Facebook winning setup for different scales because again it's won't fits the all the all the like budget scales and uh and creative production from small to to me to big size. So 20,000 good problem to have very good problem to have
because then if we even go back to to

here
one
because that's pretty much
but sorry for business as usual this is basically our main campaign maybe you can elaborate
yes exactly that's the main campaign that you're running for for the ROI purposes obviously so you have let's say US which is the the main go and that's why we call it business as usual because that's the main campaign that you want. You you need this to be as best as possible obviously. So,
and all my winners basically go go there.
Yes, exactly. That's the end game. Exactly. That's the end game because creative fatigue hits quite quickly even for Facebook or other channels. And I found out last year basically if if I do refresh creatives let's say two or three times a week even if it's against like Facebook best practices it's actually beneficial for the performance. So it

helps again with high velocity of of creatives it it kind of creates a little bit of a headache because there is limit on the number of creatives you can have uh on each adset. But again, that's a good problem to have because then you can add and refresh creatives like couple of times a week and that's going to be beneficial for the for the optim like the performance because then as I have here one sec I'm just I can zoom it here like you can add two or three creatives or like let's say I will just actually change this add the new creatives weekly but two or three times per and uh because that's important that's important and then obviously promoting all the winners to the to the main business as usual campaign. So that's that's like the videos and we have we have here the the Google Tik Tok appin and unity. So if we go back pretty much

here it's kind of a different setup because for Unity there is a creative testing campaign. So you can create multiple of these. Then for Google, I'm not really having any testing campaign. I have only a business as usual ad groups. So every new creative, I'm batching them together for uh as a theme. So let's say I have freezing families. That's one. That's one ad group. I have a Kingshot ads. That's another ad group. And you can have 20 videos per each ad group. So I'm trying trying to do like by concept. So longer, shorter, fast-paced, anything still the same concept, iterations of the same concept in one ad group. And then then we come to the Tik Tok, which is also quite interesting because it's even more heavy than than Facebook. And this needs to happen even like more frequently because you can rush through the creatives on Tik Tok very quickly,

sometimes in the matter of days, which means like it's almost like every other day you need to add new creatives. And even Tik Tok is producing or like having a program where you you work with create creators and they're producing creatives for you. So you can have actually like 40 new videos per week. So you need to start loading them into the campaigns. And then here I kind of have this this uh setup where I add new ad group per campaign with like eight new videos and then just kind of rotate this. Again, automation here is the key because you if you do it manually you are I mean you you don't do anything else. Just upload creatives, add new ad groups and that's it pretty much. Thank you very much. I I need to go to sleep and then next day you do the same thing.
So it's depressing. It's very yeah it's very time time heavy but I mean then it's rewarding because if you do it the per performance go goes up.
We can see it clearly on all the big games they are doing this pretty much

like the loads of creatives and also like here the applian is saying the same thing which like if you add new creatives into the campaign the percentage of the budget goes straight to the creative testing. So if you do add new creatives, you are kind of getting like 5% of the let's say short-term boost where you can get traffic then algo kind of realizes if the creative is good or not and then goes and then just you have a new winners or you just keep testing pretty much. So this is super important. Oh that's why I mean obviously we're talking about it.
Yeah. And we started with Facebook, right? Yes.
Do we use the winning Facebook creatives on all those channels or we are testing every channel have it own?
Yeah. Yeah. Yeah. every channel as on its own because it's I mean honestly if you have a winner which is like a an

actual creative vinner then it's going to win across all the channels but you can't really uh rely anymore on the fact that what works on Facebook is going to work across different channels because those are very different channels and different behaviors and also different algorithms. So pretty much what works on Facebook not necessarily works on on on Applavin because for Facebook you need hook you need very strong um kind of gameplay or middle of the of the creative but then on applavin you need a very strong call to action at the end where on Facebook you don't really need that because 80% of the people don't really finish the video even if you have like one minute long video like 80% of the of the people maybe maybe will watch like half of it and that's the best case scenario on AppLavin is the other way around. Everybody needs to watch the whole minute and then afterwards even you go to playable and that playable also has like some best

practices especially on Applo. You need to have a playable where player can click at least 10 times and spend at least 60 seconds in there and then you have 1 minute on video, one minute long playable and then suddenly it's like hey where my my day went? I didn't even finish the uh the video and playable. So like that's that's very important. But
but then yeah
we still start like if I'm um I don't know small studio I still start with Facebook that would
yeah 100% yeah for sure for sure but super super helpful mate I I didn't know that it's so different
it is it is I mean it wasn't the case a couple years ago but now since you know things are changing obviously in the in the gaming industry then it this this changes as well and also like what I just mentioned uh within the the creative frameworks for testing it's going to be different in the six months most probably

and if it's going to be different in six months like that's the best case scenario because yeah I'm changing this I mean these are like four frameworks that I usually use like across the time but like with small changes here and there it's like like the worldwide campaign I mentioned like it's wasn't just only let's say five countries from tier two or tier three but now worldwide because I need more data. So again, it's a small change but can be very critical afterwards.
Gotcha.
And then I kind of touched the the playables and it's also like very important to actually test playables and I know what I'm going to share this which is way more intimidating.
But you know what? Let's see. Perfect. This is even better cuz this is basically showing the the whole kind of flow how to test create uh playables and this is something I'm usually doing on on Unity. So I have the best performing

videos and then you connect those videos with the playables and that's like the the connection or like the the format video plus playable end card that's like the best performing across the across the board. So So then like okay well it's pretty complex how to test it out. So you have basically the winning video, you combine that with different playables and test them against each other. So it's it's almost like the same thing as you have only videos, but now you have a playable end card. Then whatever wins, you know, it's kind of promoted here and goes to the to the new wave of testing. But you always use the winning playable as a control group or a control creative and then test the new combination against each other. It sounds very complicated but it's actually if you are a U manager and you go to to Unity or wherever it's actually it's not as complicated because you clearly see like what what this means.

So because it's like it sounds and sounds and looks pretty tough and complicated. It's not. Don't worry. Now even with um let's say the Apple creative sets which is here it even streamlines the whole process because what I just mentioned you have a video and a different playables you can have a different set. So here I have a video which is obviously called I don't know 500,3 and then the holes vong and those are different playables. So you already kind of added them into the campaign and let the algorithm do the the testing for you. But you can actually have like these couple of these different versions which again like it's simplifies the whole stuff. But this wasn't really the case six months ago because the creative sites on Applavin didn't exist back then,
right? So,
and
yeah,
what is a good ratio like for the video to playables? Because playables are

still like harder to produce. Jakob showed us how to produce videos. Uh, but playables.
I mean, look, that's why I built the play maker because
No, because honestly, uh, of course, it's very tough to build a playable and even even tougher to actually, um, test it out afterwards. I mean, if if we go back here, it's like the same thing. You need to test playables and then pick the best combo
and then and then even iterate the winners of the of this. So, so how do you iterate on on on the playables? So, it's again longer, shorter, whatever. and also like how to produce playables easily. It's that's why we have the templates on on playable maker because then I found out let's say with Andre on in August that let's say Evony hobby and like midcore and hardcore games and forex games they're using 3x3 um puzzle which is basically I mean doesn't

make any sense right? So they use this type of template. Okay. I have it actually here. So look.
Yeah. Yeah. Yeah.
So I mean it's just a static image and then it's a puzzle afterwards. I was and everybody was using this now when the jigsaw came came live. Now it's even match villains and royal match had the same kind of playable. So and this is a template. You can build this in 20 minutes. Seriously. You just upload you just upload a static image and then create the playable out of it and that's it pretty much and then you can you can start testing it. It's like a proper playable. So So this is how you can kind of do it. It's al of course it's not exactly the same kind of long form playable as you would need to build as as a programmer or your playable team. But this is this is going to do the trick for like let's say like filling the gap when you don't have playables and you don't have let's say

5,000 for for production for playable,
right? Okay. This is scalable, right? And
yeah,
I see. I see.
Pretty much cuz then you can create this and then there's like other like 100 other templates uh that we have. So from quiz to jigsaw there to just like void games and everything else and even like scratching like almost uh like scratch the the surface of the of the image. So it's kind of unveiling the the image like so many different things and and it's very simple but it actually works because what people do and I have this in in my kind of the prediction is like you have a highly scalable game which is now like the solitary associations or whoever or the jig solitary before and what you do instead of just kind of you cloning that game and starting to produce on your own you just take that game concept and you just use it in the

in the playable as a template or you just build it as a playable and then just put it on your creatives and then suddenly when people see this oh interesting yeah
it's good uh it actually works. So so yeah so we went from the creative testing to some of the trends and then all of the all of the different things. Oh, but that also an insight in terms of it doesn't necessarily needs to reflect your actual gameplay. And from what I've seen, it's just any
and even look, yeah, it's like in in this case in these trends, it's kind of like it doesn't reflect the actual game mechanics, but it's it's a game kind of asset basically. So, let's say best performing creative or best best performing static, you can kind of translate into the playable in kind of in in no time. Gotcha.
Yeah. So then yeah, what we have here? Yeah, this is basically after the whole kind of iteration test now all the testing and winners and then obviously

what you want to do is just iterate the winners, right? Because that's why you do the testing and then there's so many different things like adding UGC, AI, sound, voiceovers, whatever like hooks, anxiety, urgency, whatever. So yeah, it's like then you do everything again. So it's never ending loop. It's never ending loop.
Nice. C. Can you quickly zoom out so we could show what kind of So this is what we discussed. Yeah. Yeah. Yeah.
Yeah. This is what we discussed now. Yeah.
No. No.
And then it goes to all the different things in soft launch. But let's Yeah. Let's not go there too much because that that will be another half a day at least. At least you know. Yeah. But this Yeah. This is what we quickly discussed and
again there is so much more as always as always.
Yeah. And I wonder in terms of like even

on the first step the whole I guess I'll call it intuition right and I feel like you need to watch enough uh creatives of different kinds
to develop it.
Yeah, for sure. Because I've worked with a lot of games before and as we also do this on the podcast multiple times a week. So you you get that yeah you kind of develop the sense of like what works and what doesn't across different uh different genres and there are certain elements that always are kind of overlapping between each other and again I had one one game where we did two videos you know one video let's call it noob versus pro and the the company wanted to have in the first three logos like what some random headline or a logo I like guys like who who gives a about this in the first 30 seconds you need to it needs to grab the attention and we just lost like 70%

of people. So I was like okay let's let's test it out. So I put it on Facebook trim the first three seconds I was like okay let's get rid of this and obviously the the one without the logo and in front like it was performing way better because we need to grab the attention not just to poor people and if there's no supercell logo then nobody cares. If there's superog maybe maybe
like last really last question uh based on what you just said were there like any recent surprises for you like you thought that this creative would never work but it actually works somehow like weirdly. Yeah, I think well if I will I'm like joke I I saw a gameplay video which works really well. These days not really the case anymore.
No, but uh I were I was always uh talking about how the creatives and like

the creatives that look like always performs better. And I never really understood like why it's like oh my god. Yeah, it's like it's not polished enough whatever. And then I started seeing all these like ugly ads basically people call it. It's like you doing something at home or just filming yourself talking about the game and then I realized like okay it works because it doesn't really look like an ad
and it's showing basically the game play. You're talking about it. It's almost like you are an influencer whatever else. And it was really surprising for me to see because I wasn't really used to kind of produce these type of things, but I started doing it and it does actually work. So like it's called it's almost like low fidelity very Tik Tok kind of like raw which really works. So that was yeah that was kind of surprise for me.
Gotcha. Super helpful. Mate, uh, we'll ask our audience to leave their

questions in the comment section and we'll ask Mate to answer
or
I will
anything.
I love comments under the video, guys. Every time you ask something, it helps the the video and it helps us and in Roman as well. So, please ask as many questions as you you want. I will find time to answer all of these
right here. Appreciate it. And of course, likes. Give us likes. That helps the algorithms too. Um, yeah. Once again, thanks a lot. Until next time, mate.
Thank you very much. Thank you guys. Bye-bye.