Session: Demystifying China UA: Reverse-Engineering AI Advice (Fact vs. Fiction) - Yolanda Huang.

Heat. Heat. Um
uh very good afternoon everybody and uh welcome to the session. So today we're going to talk a little bit about UA in China. Uh so for our friends some of you who have been listening to our talks you know over the years and over different events um we've been trying to update

the market about uh the new informations from China or in general from Asia all the times but the difference this year as we can see it's more AI related. So on one hand we are seeing that of course uh more and more people we are relying on more on AI nowadays and of course at the same time uh we we hear feedbacks from the market we hear from our clients or potential clients that when they've been trying to uh do the market research inhouse based on AI tools um and then of course after collecting all the information by AIS they uh add another layer of uh human validation to assert certain extent and based on all those facts they make decisions on whether it's the right time right now to try a little bit about China or a little bit little bit about Asia. Yeah. But when doing that also thanks to those clients who are sharing those in-house uh uh procedures with us we realize that the

information points that actually gathered by these large language models despite they should be we expect them to be really good with languages but the information points we're getting those AIs are gathering for for us they are either uh they could be outdated they could be incomplete they could be partially true so they could be missing points as well. So it's not 100% facts out there. They might be giving only partially the picture. And so when connecting the dots with a certain type of logic, we're seeing that not everything is true and it's actually the important information being missed to assist better decision making. So that is what we are trying to do today. Uh we are going to review some facts and fictions especially. Uh so what we did is that we carefully designed a questionnaire uh that we brought this questionnaire towards uh the uh the different AI tools and see what kind of information they will be giving us.

Yeah. And then we are going to uh you know go through the results with all of you here. So this is the methodology. We brought a question to a list of mainstream AI tools listed on the left side of this this table. uh I said okay I'm a developer looking at marketing in China please recommend the five uh best AI tools for me and this is the results of basically the reasonings given as well. Yeah. So the idea is that our selection of AI tools is not going to impact or skew the results too much. So we decided to go with all the AI tools listed here recommended by among themselves and ask them the same questions. I'm a developer looking at marketing in China. uh give me the list of five questions that you think is the has the most priority so we can go through together with our team and this is the results we got. Yeah. So putting together the answers from all

the AI tools uh we made a word cloud out of it and this is what they were talking about. Of course you know the biggest word appear to be China Chinese game. Uh it makes sense. Uh there's uh there are words about different kind of platforms, different kind of strategies and uh yeah so if we look at the words circled in blue these are more the legal and compliance side of the things which we do think should be coming as the very first question when you're tapping into a new market. We'll go through that and there's a little trap we left in the question as well. Uh so look at the word circled in the yellow color. Yeah, that's the type of games or the type of genres at all. Uh whether this is a PC game, whether this is Android, what kind of stores or not to mention, I didn't really specify that I'm a game developer. So if there's any app developers out here, if you ask the question, AI is not really identifying

it will right away give you all the relevant answers from the gaming side, right? So of course uh in this presentation later you're going to see we're going to explain as well you know there are different approaches whether your game is a PC title or it's Android it's iOS or apps but the approaches would still be different from each uh type of this products. So um the first question as mentioned would be around regulatory and compliance. Yeah. So did eight six out of eight AI tools uh think so as well. But the interesting part is only one out of the six tools they recommended to ask what restrictions are out there. Yeah. All the other tools told me okay the first question you should go for is to ask to get a stepbystep guide to obtain a ISBN. Is it really necessary? So what is an ISBN? That's the license that helps you

you know launch your game properly in China for certain type of games. Right. So here comes the common misunderstanding number one that we see in the market. I'm receiving this question at least a few times a week. Uh so uh it seems to be a common understanding that uh games need ISBN for China. ISBN is hard to get so they don't consider it with the current setup. Is it true? I mean um for certain it's it's not necessarily for actually all the titles. Yeah. But we when we deep dive into this question, if you look at the lower part of this slide, these are all the noises that we get from all AI tools. And I can tell you that none of this statement is true. There all seems to uh you know uh mix a little bit of uh facts with with one another. Some of them might be partially true. The others um they are even self-contradicting. For example, the one on the I'm not sure how to point, but

the one on the in the middle on the right. Basically, it says advertising is fine, but listed getting your game listed is a problem. And basically, what does it mean? What what are you advertising about if your game is not listed at all? Right? So, this these are all these kind of information out there. And the answer to it is that um there's a yes there is a regulatory uh uh requirement. This is a concept called software copyright license. Some of you might be be familiar with this term already. Sorry. Some of you might be uh familiar with this term already. So this is the much cheaper and much uh faster version of the license that you could get. This is also a prerequisite for the ISBN which is the full license. What this license does is that it enables you to do a marketing test already. Yeah. Both on the paid media side and on the influencer side. So with IA titles with

this license you are set to run a test. With AP titles of course you have to wait. But Steam titles is a different story. In certain cases, your test could be initiated even without this license. Why do we want to do this? It helps you to validate the market potential before any decision making. What kind of decisions? Right? For example, if you want to further improve the monetization of the title uh to in order to reach your row as goals, if you see the potentials in it, if you want to spend more resources negotiating with the publisher and go through the AP route to get an actually a full license. Yeah. So those are all the decisions uh that could be helped uh with you know running an initial test. So that's the answer to regulatory and compliance. Yes, your game can get onto it. What else? What's the next? All of them

recommended channel mix and strategies to be asked about. We agree with it. We need to know what kind of channels we can get on for this test. Five out of the eight AIS we checked with, they also recommended to check for K strategies as part of the UA. Yeah. And one problem we identified here is that everybody assume that this question come from a China entity? Can foreign entities not pro promote games in China? Yeah. Or in another words, can I you know uh not keep all my user data and learning data under ad account that's actually owned by me. This is the answer given by different type of AIS. But what we are seeing here that the whether yes or no is not really important in this graph, right? But the key information is that there's no single AI tool that is giving the most

right. They're getting everything wrong. And the second one is that there's no single question that the most AIs are getting it right. So answers are basically all over the place. You come there's not even a uh almost source of truth. So that's the that's the that's what we are getting from the from the AI tools and the correct answer that we can provide you of course is that yes everything is possible. That's why we exist right that's why I'm standing here. Some of the AIS I really like them. They say okay it's possible via agency like us. That's really good answer. Uh but what I'm trying to say is that um uh you know uh foreign entities yeah if you want to own all the campaign data all the user data all the learnings by yourself it's possible you are having actually you're observing very similar rights and platform guidelines as compared to a

local entity. The only difference is is sometimes logistic. For example, you have to go through an agency who has the right like from the legal side of the things to deal with foreign entities and that's it. Yeah. So that's the common understanding. Yes, foreign entities can do it. And uh in terms of what exactly are the channel mixes uh gun they're giving all different answers. Yeah. So for this I can't I couldn't even actually put all the different statement here because basically everything was just noise. But if we do cut all the noises, these are a few success cases. Well, rail cases that we run for our clients for different genres, you would say, okay, the channel mixes might be different, especially if you look at the second largest channels. Some of them being Apple search ads, others being buou and tensson ads, others being quite show for example. So they could all be

different, but there's one thing very similar. The most scaled channel of most of the titles that we deal with all of them are ocean engine. Ocean engine is the ad platform powered by bounce. So this is the traffic uh getting you access to the China part of the Tik Tok. Yeah. So that basically means if today I'm very mature, I already know how to play with China. Yes, there are a lot more solutions out there to be tested out. But if I want to launch a test, which platform should we actually go for? Yeah. To prove the business concept at the very first place. So that's the answer to the channel makes and strategies. And now it comes to the next question. Now I know that whether my game or whether my app can advertise in China at all. I know that which platform I should be going on to. What should I expect in terms of testing budget and benchmarks? what should I spend and what can I

expect to achieve? We asked this question with very clear prompts. Yeah. Specifying what we expect to get, specifying the type of deliverables they should be giving us. But unfortunately, only four out of eight AI was giving any number uh advices at all and the other half of them are just talking around. So out of this four AIS, another interesting thing is that the recommended test budget range from 1 to 40,000 and the estimated CPI range from 70 cents to $480 for exactly the same title. Yeah, they vary a lot and that actually aligns with the questions we usually receive as an agency. So uh we do provide you know information to the market about China. We help our clients with Japan as well with Korea with Southeast Asia. But China is among the countries uh that we

see extremely opposite perceptions from the market. We do have rules of sams for it. What to do there? The first thing of course don't start with Android. Yes. Android is like having 70% of the market share there. Yeah. Uh but uh uh what we could tell you is that to achieve very sim to achieve similar rows uh Android campaign can scale as much as to onethird of that of iOS. Not to mention that in order to cope with the Android ecosystem there, you would actually need to deal with 20 plus uh Android stores with different bills, different trackings. There will be a lot of resources being involved. Of course, eventually you're getting to a point that you think, okay, Android is definitely necessary now in talking about incrementality, but don't start

with Android. Number one. Number two, use your tier one benchmarks. That's a very good indication unless unless the say the art style and the general is really not fitting into that market right but we all know there are less than a handful of options that it might might not work. Number three is this is your recommended daily budget. So we can basically calculate yeah based on our regular approach how much it would cost for a oneweek testing campaign for a two week testing campaign how much I would need to put down to understand okay if there are rooms that we could actually optimize to actually reach to my KPI or not or maybe the CPI is like $60 there's no point of in continuing yeah so the testing budget is really not as you know crazy as what the market is thinking Uh so that was the answer to uh benchmarks and test budget and another interesting point is that

every single AI that we ask they all agree that KL's or UGC's or influencers they are mandatory right now everywhere but they are more mandatory for China because of the audience behavior. Everybody could give benchmarks of cost per video. Yeah that's the easy question. one to five million uh followers, how much uh you know you usually typically spend to engage with an influencer. However, none of them actually mentioned that influencers could be performance-based. Yeah. So, this is the latest success case that we see. This is a performance-based influencer campaign. Uh so just to give you a background of us, we are the first and exclusive uh agency for the creator marketplace from bite dunks. Uh so we are also the very first one to make this project work at all for foreign entities and this is the

results. After one month of campaign, we collected over a 100 UGC's from the players. The viral rate was doubling what the what we see from the regular UA creators. Yeah. CBR increased by four times and CPA improved by 16%. So the UGC's created by influencers joining this program became very fast after three or four weeks the top spenders in the account. So this would be a very interesting program uh available for games for apps uh and basically for the other verticles as well. So yeah uh I I hope the information was helpful. Uh this is uh about us. We are native a we are cross regional digital marketing agency. So as mentioned besides China we also you know help our our clients with the other part of Asia or the other part of the world. Uh under the gaming sector we have mature solutions for PC titles

for iOS and of course we serve uh app clients as well. This is a list of media that we could help you get on. Uh on the lower part of this page basically are the things you're seeing the local things from from Asia. Yeah. From Japan, Korea and China. And uh if you have any questions, I mean Joan and I, we are here. Uh feel free to add us on LinkedIn and then uh let us know if there's any questions. Yeah.
So do we have any questions given that? By the way, I thought that was fantastically interesting. It's over the last maybe 10 years there's been so many different changes that have been happening in Western teams going to China and Chinese teams coming to the you know European markets or or American markets and it's it's kind of it's still unclear and the level of misinformation that you're talking about seems incredibly well scary how much information there is.
Does anyone have any quick questions?

On that note I am going to say thank you very much.
Thank you so much. I mean, I think, you know, the the more we realize we're in a global market, the better. I think that's always incredibly important. And I think what I found really fascinating about that in particular is the last bit when they're talking about the important roles of the streaming you know the the content creators as part of that the the use of content creators in mobile in particular in um Asian markets as opposed to in the European markets where I don't think we've seen that level of mobilebased uh content creation despite the fact that the device we are using to enjoy most of that media happens to be the one in our hand anyway. Heat. Heat.