Affected by the tightening of domestic game version issuance, as well as the peak of local market dividends and stagnant growth, since the second half of last year, the urgency of Chinese game companies to go to sea has risen sharply.
In 2021, the sales revenue of China's self-developed games in overseas markets reached US$18.013 billion, an increase of 16.59% year-on-year, and the growth rate was ten percentage points higher than that of the domestic market. It is not an exaggeration to say that the sea is the number one priority of many Chinese game companies this year.
However, in the unfamiliar overseas market, how should Chinese game companies advertise, buy, attract new players, and do a good job of user retention and monetization? You know, overseas is not a whole big market, but a concept composed of many national markets. The Market in the Americas is vastly different from the Market in East Asia, and success stories in a national market are actually difficult to replicate even in markets that are very close to it geographically.
Moloco is a company that leverages machine learning to provide global marketing solutions for different industries. In the past few years, they have played the first overseas advertisements for many Chinese game companies in many industries such as games, e-commerce, finance, pan-entertainment and other industries that have gone overseas for the first time.
"Moloco is a machine learning company that helps companies target users with massive amounts of data." Morden Chen, Moloco's director of sales for Greater China and Southeast Asia, said. "Moloco helps customers reach more than 10 billion devices and more than 2.6 million apps per month in more than 240 countries around the world. Customers don't need to go to the placements one by one, they can do it quickly with Moloco's system. ”
Over the past year, Moloco has helped customers make more than 35 billion ad impressions per month, making 4 million predictions per second about the ads themselves, speculating on whether users will install the ad, whether they will pay after the install, and so on. Moloco has a lot of observations and insights on how Chinese game companies buy, how to retain, how to monetize and so on after going to sea.
36Kr discussed with Morden Chen the current situation of Chinese game companies going to sea in his eyes, and hoped that by summarizing the rules, the latecomers would take fewer detours in buying and marketing.
Murder Chen
The following is a transcript of 36Kr's interview with Morden Chen:
"Novice villages" may not be easy to fight, and culture has a long way to go to sea
36Kr: According to your observation, is there a clear regional preference for Chinese games to go to sea? Can we call the Japanese, Korean, and Vietnamese markets in the Confucian cultural circle the "novice villages" where Chinese games go global?
Morden Chen: That's a particularly interesting question. Japan, South Korea, and Vietnam that you just mentioned have really been called Confucian cultural circles for some time. If we go back to 2014 and 2015, we will call it "Novice Village", where we practice playing "little monsters", and then fight "big monsters" in other markets after we have experience.
But in retrospect, Japan has not really become a "novice village." At first, everyone thought that Japan was a "novice village", but after a battle, it was too difficult to find that it took a long time to gradually occupy the market share of Japan. The Hong Kong, Taiwan and Southeast Asia regions did exist as "novice villages" during the first wave of sea trips in 2014-2016. Because it is similar to Chinese mainland in terms of language and culture, some themes are more acceptable, such as game products such as three kingdoms, xianxia, and fantasy.
But now I don't think it's an easy "novice village" to fight. The reason is simple – there is a proliferation of products of the same type. People will have aesthetic fatigue, and it is impossible to stare at dozens of Three Kingdoms games every day, each of which is played. Players will also want to play something else when they are tired. Now these regions, in fact, there are already games from Japan, South Korea, the United States and even Russia, Israel and other countries landed, and now the competition in Southeast Asia, Japan and South Korea, Hong Kong and Taiwan and other markets is actually very fierce.
36Kr: As far as I know, Japan is a relatively closed and mature market, and the competition is very fierce. How to reduce costs and increase efficiency in domestic games going to sea in Japan?
Morden Chen: In the past, When China went to Japan, advertisers who did well — we defined "good" as products in the top ten in a particular category and continued to be in the top ten for at least a year or more — almost all of them did a lot of localization work.
Localization can be divided into product level, operational level and market level.
The product level refers to the localization of the product itself, including language, gameplay and even image. For example, some strategy games in Japan will be specially customized Japanese versions of heroes, introducing characters wearing sailor uniforms, student uniforms, or heroes who look like Japanese stars. At the operational level, it will reflect the cultural characteristics of Japan, such as celebrating local festivals, launching festive gift packages and events, and so on. The promotion level will be due to many of the usual routines in the Japanese game circle, such as pre-registration, voice actor endorsement, and influencer endorsement.
Courtesy of Moloco
36Kr: After such a heavy localization operation, is the CPI (cost per installation, representing the price of buying each new user) in the Japanese market very high?
Morden Chen: It depends on the genre of the game. From our current data, the CPI of moderate and heavy games in Japan is indeed significantly the highest. The cost of buying ultra-casual games in the Japanese market is also relatively high. Ultra-casual games are generally very light and do not have too many cultural attributes and regional attributes, suitable for global distribution, but even so, Japanese players still have a strong demand for localization, more than the Korean market. We have customers who specialize in making kawaii-style games for Japan, from art to gameplay are very Japanese, which is very in line with the Japanese aesthetic at first glance. In this competitive environment, if you don't localize enough, it's easy to compromise in the Japanese market.
Courtesy of Moloco
36Kr: At present, do users in different markets have obvious game preferences?
Morden Chen: Yes. We generally call the United States the most comprehensive market, and the United States basically has all categories and types of games. Taking social games as an example, this type of game has a large market in the United States and English-speaking cultural circles, because they have a tradition of playing slots, Texas Hold'em, solitaire and other betting games. But none of this is in the Chinese cultural system, so you rarely see such games appear in China or in the Japanese and Korean markets.
In Japan and South Korea, the proportion of three-country strategy and RPG games is particularly large. Japan and South Korea, Taiwan, Hong Kong, Vietnam, these places of the Three Kingdoms game can almost be singled out as a category. The Three Kingdoms are an inexhaustible source of free IP, and it can become a large genre on its own, very popular in East Asia.
Europe and the United States do not have a phenomenon-level fairy-level product with oriental elements for the time being. However, in Japan, South Korea, Hong Kong and Taiwan, and Vietnam, some fairy tales have a certain market share. There are many advertisers and developers in the Guangzhou-Shenzhen area who do this type of product. Xian Xia is a genre that is particularly common in Confucian cultural circles and rarely seen in Europe and the United States.
However, in the eyes of European and American players, the "fairy" in their eyes may not be our kind of game, but a game that looks more sci-fi, similar to "Star Trek", "Star Wars" or SLG and RPG products such as "Lord of the Rings" and "Lord of the Rings" in Middle-earth. Therefore, a certain genre may be popular in different markets, but it is presented in different ways. Will there be convergence points in the future? It's possible, it's just that time will be long. If Disney hadn't worked so hard to launch Marvel, we probably wouldn't have known spider-man, Iron Man and other characters.
We've seen Chinese companies working hard now, but when will it be as successful as the export of Japanese manga in the past 90s? It may take a long time. Fortunately, the Internet age has the opportunity to accelerate, because it presents more diversified ways, and we can present our culture in various ways such as comics, long videos, short videos, etc. I think we have a chance to achieve the same level as Japanese cultural exports in a shorter period of time.
36Kr: The going out of any cultural product to the sea depends on the prosperity of its entire cultural system. Looking back at the most successful cases of Chinese games going to sea, you will find that they are either two-dimensional or American comics, because for most overseas players, their familiarity with this type of culture is still stronger than traditional Chinese culture.
Morden Chen: I agree very much, and you will find that successful Chinese game companies have a global presence. There are many foreigners in Mihayou's team, and there are also community managers everywhere who can make suggestions for various cultures, so that a game really has elements of national cultures.
I think the advertisers and developers of China's overseas games have gotten better and better. In addition to the American comic style games such as "Awakening of All Nations" and the two-dimensional games such as "Original God", there is also a category called match-3+. We have seen a lot of teams that make match-3 games carry out a lot of innovation, such as match-3 + business elements, simulation elements, doomsday elements, home improvement elements, etc., there are male and female directions, they have been successful in Japan, South Korea, Europe and the United States.
How can Chinese game companies buy players at low prices in overseas markets?
36Kr: MOBA (Multiplayer Online Tactical Competitive Games) and SLG (Strategy Games) are much more expensive to acquire customers than ultra-casual games, right?
Morden Chen: There's actually a big difference in multiples. Ultra-casual games generally take advertising as the main source of income, and its logic is to obtain potentially retained users at a low price as much as possible, as long as the user is willing to play, willing to spend time, and watch the advertisement. Heavy gaming, on the other hand, will place more emphasis on ROAS (return on Advertising spend).
For heavy game advertisers, they want to buy as many "big R's" as possible. The so-called big R is the person who can spend money in the game. But they also don't want the players to be free of ordinary people. We divide the kryptonite level into: heavy krypton, medium krypton, microkr, and non-krypton, just like the pyramid. Why does the game also need to have people who don't have kryptonite? Quite simply, if everyone is kryptonite, you spend 10,000 yuan, I also spent 10,000 yuan, both of us are well equipped, we can't score the score, krypton gold has no sense of superiority.
Then we, as professional buyers, need to use professional means to let advertisers buy the amount they want - both kryptonite players and non-kryptonite players.
Moloco's system can differentiate the value of the target user. We have a CPI model when you need to push volume, a CPA (cost per action) model when you need to push a paid event, and a ROAS model when you need a big R user who can pay repeatedly. There are also advertisers who want to recall old users who abandon pits, and we also have a Retargeting model to meet the demand. Advertisers can use us at every stage.
36Kr: How do these models meet the customer's requirements for indicators?
Morden Chen: Moloco has made five models of CPI, Retention, CPA, ROAS and Retargeting according to the needs of advertisers.
CPI, as the name suggests, is the cheapest installation, regardless of whether it is retained or not. Giving the machine a "I'm going to install the cheapest one and nothing else" instruction is the result.
Retention is retention, which adds the learning of "retention events" to the scope of the first model, and the machine tries to find users who are not only cheap but also can stay. The model behind what the industry calls look likeLook-Alike is look likeLook-Alike technology, which is to find similar people. It is based on the advertiser's first-party user data (First Party Data), and with the three-party data source (Contextual Data) let the system learn and find similar users.
The CPA model requires users to generate a special behavior, such as purchasing, recharging, completing registration, etc., and to depress the unit price of a single event as much as possible.
The ROAS model is a CPA model multiplied by a ROAS factor. When users already have installed and paid behavior, ROAS hopes that users can spend more money to buy more props, so that advertisers can earn. Using the ROAS model to find a big R is to consider CPA and behavioral factors, then consider the factors of repeated purchases, and then find a look like for repeated purchases.
The Retargeting model is a recall, and the old player who retrieves the pit through ID echo.
Moloco does its job through massive machine learning automation, using as little manpower as possible with minimal use. Help advertisers drive output per capita. Our account manager's job is to look at the data and analyze it and provide specific recommendations to the customer. In the future, this part of the work may also be more automated and intelligent.
36Kr: As far as I know, Moloco has a product called DSP, which can guess whether the user will install after seeing the advertisement before the user installs it, and whether it will produce paid behavior after installation. How accurate is it?
Morden Chen: I want to show you a diagram, and I'll use the most simple and easy-to-understand way to explain what exactly is behind machine learning. In fact, the so-called "guessing" refers to the use of correlation to make judgments, which is of statistical significance behind it. The so-called machine learning is the application of statistics after having advanced computing power.
Courtesy of Moloco
There are five consumers on the left, and you can see what they bought separately. Machine learning found a correlation between them — buying milk was related to buying Coke, buying diapers was related to buying milk, and people who bought milk and diapers were also inclined to buy beer. So by logic, if I know you've put milk and diapers in your shopping cart, I'll push you an ad for beer at this point.
Machine learning algorithms have no way of telling you what the reason is, only that you have the phenomenon and relevance. We use data that does not invade user privacy for machine learning. We know that when a user buys milk or coke, it's not a privacy invasion, because I don't know who is buying it, and what we do is push you a beer when you buy a milk diaper.
With that in place, our machine learning will guess how much an ad placement should spend on which customer. There are tons of calculations running every day, every hour, every second, bidding constantly. This is called machine learning.
36Kr: So is the correlation between milk, cola, and beer similar in the case of the game industry?
Morden Chen: Based on the game industry, we actually have a diagram.
Courtesy of Moloco
Machine learning algorithm prediction application in the game, the dimensions of first-party data (as shown above) include: In-app purchase, play time, levels played, etc., we will use these data performance to predict the user's performance in the game, as the basis for bidding.
36Kr: After the release of the version number was stopped, did you feel that the competitive pressure of Chinese games going to sea became greater? If Tencent and NetEase both set the game going to sea as an important goal this year, will it be more difficult for small and medium-sized game companies to do it?
Morden Chen: Data.ai and Sensor Tower and other industry data analysis companies will have a Global List of China overseas every year, ranking based on the total number of installs they have collected, as well as a list for in-app purchase (in-app payment), by looking at these two lists, you will find that although in the case of super competition, almost every year the top 30 of the list will appear a few new faces, there will also be a player in the back to rise, the front to fall.
Under such competitive conditions, why can anyone still win? There may be many reasons behind this, it is possible that they were originally big game practitioners to start a business, it may also be that the company has opened up new perspectives, new ideas, and may have changed in the original category and opened up a new world.
The game industry is actually very innovative. If you have a good idea and put it into practice, new opportunities are always available. Looking back at "Piano Block 2" made by Cheetah 5 or 6 years ago, it was also a big innovation at that time, so it got a large share in the world. Some people think that Tencent and NetEase are so big, we must lose, no. They're big because they have their main battlefield and the main track, and you don't have to do exactly the same thing with him.
However, from the perspective of buying volume, things are getting more and more expensive, and it is more and more difficult for users to buy, what to do? The key is whether you can have a strong team and a strong financing ability to get you to get ammunition. Spend money after financing, spend money after spending money, and make investors willing to give you more money to do the second and third products.
From the perspective of the whole environment, entrepreneurs really have to grasp a deep understanding of a certain market and a certain type of product before going to sea in order to win the opportunity they deserve. If you don't understand anything at all and want to do it, it will be very difficult. I think it will be more and more difficult to go to sea, and we must have a full understanding of a certain market before starting a business.
36Kr: Do you feel that there has been a significant increase in the level of CPI in overseas markets from the middle of last year to the present?
Morden Chen: According to our own observation, although the CPI is really rising every year, it has not particularly soared, and it is difficult to say whether this situation will be in the future.
But in 20 and 21 years, there is indeed a situation of rapid growth, and the important reason behind it is that everyone can't go out at home during the epidemic, so this time is indeed whether it is e-commerce or games, the purchase volume has become more expensive, because many advertisers want to obtain online users, and everyone's competition has become fierce. EdTech companies, in particular, need to increase advertising to get customers because students can't go to school or cram school. So in the epidemic, although it is impossible to quantify the increase in advertising, we can see that its increase is even higher. But now we're past that period of the highest gain.
36Kr: That is to say, compared with the epidemic period, the impact of the suspension of the distribution of the version number is not very obvious?
Morden Chen: I think it will have an impact when domestic companies are ready to go to sea. Before that, the focus of each company was still in China, overseas is only a game company that does it at the same time. In the future, if they want to focus half of their attention abroad now, it will definitely change the entire ecology.
But I don't think it's that time yet. Look carefully at several head games Tencent or NetEase and several other big manufacturers, they originally had to make a sea, and there is no particularly obvious strategic upgrade at present. From last year to this year, I have seen small and medium-sized companies begin to study the phenomenon of releasing overseas games, and there are also game companies that originally only do domestic distribution have begun to try overseas markets, and have come to us to carry out overseas user growth advertising campaigns to buy overseas users, but for the time being, there has been no excessive competition in traffic.