A/b test helps the growth of game business

More technical exchanges 、 Job opportunities 、 Probationary benefits , Welcome to your attention Byte bounce data platform wechat official account , reply 【1】 Enter the official communication group

 

Development status and challenges of China's game industry

Weak growth in the domestic market

2021 Annual game sales revenue 2965.13 One hundred million yuan , Year-on-year growth 6.4%, The growth rate is higher than 2020 Annual reduction 15%. The scale of game users is up to 6.66 Hundreds of millions of people , Year on year growth was just 0.22%.2021 The growth of game sales revenue in , The demographic dividend of the game tends to be saturated , The domestic game market as a whole is changing from an incremental market to a stock market . Affected by the law on the protection of minors and the restrictions on the distribution of domestic game editions , Domestic development will be limited for some time in the future .

The self-developed game has a strong momentum at sea

2021 year , The actual sales revenue of China's independently developed games in overseas markets reached 180.13 Billion dollars （ near 1200 RMB ）, Than 2020 Years increased 25.63 Billion dollars , Year-on-year growth 16.59%. From the average growth rate in the past five years , The share of Chinese games going to sea shows a steady upward trend . The income distribution of overseas key regions of China's independently developed mobile games is , From the U.S. 、 Japan 、 Korean revenue accounts for 32.58%,18.54% and 7.19%, Contributed to 58.31% The income of the .

 

 

Private traffic disturbs the industry pattern

In recent years , With the rapid development of we media platform , The accumulation of private domain traffic provides a new launch foundation for the release of games . With the help of precision marketing technology , The game launch on the we media platform has achieved remarkable results . Play with celadon 《 The strongest snail 》 For example ,《 The strongest snail 》 It will be launched on two platforms, Tiktok and watermelon video , Warm up through suspense 、 Super challenge 、 New interactive ways such as the task of the whole people , The target users have been accurately reached , After the game goes online 10 God iOS Flowing water super 1.04 Billion , For a month in a row iOS Top ten best sellers of the game .

As the number of domestic game users tends to be saturated , China's game industry has also gradually transformed from a high-speed growth period , The market maturity is improved , Competition tends to be refined . The cost of game companies is high , It's hard for users to retain , The requirements for fine operation are becoming higher and higher . Superimpose the challenges of game sailing and private domain traffic operation , The demand and dependence of game enterprises on data analysis are further improved .

 

Quick trial and error ,A/B Testing helps create popular games

The game industry has developed so far , Has experienced the era of console games 、 The age of web games 、 Mobile game era and other stages , Today, mobile game revenue accounts for% of domestic game revenue 7 As above , Many traditional game companies are facing the pressure of transformation . As the game technology matures , A large number of emerging companies are pouring into the industry , Further exacerbated the competitive pressure .

The game R & D process is roughly divided into project approval 、 verification 、 Development 、 test 、 There are six stages of online preparation and operation , Each stage is often accompanied by many modifications . How effective is the modification ？ Whether there will be a negative impact on the core indicators ？ Traditional thinking , It is to verify the effect of new features after they are put online , However, the online cycle is often long , Over reliance on experience , Problems with a wide range of influence .

 

 

A/B The experiment takes out a small part of the online traffic , Completely randomly assigned to the original strategy A（ The control group ） And new strategies B（ test group ）, Combined with certain statistical methods , Get an accurate estimate of the relative effects of the two strategies . adopt A/B Experiments can be made on new features of the game / Modify to make scientific evaluation , Use small online traffic to verify the effect , And be able to effectively measure the benefits .

A/B experiment , The results of scientific evaluation modification

 

Before we do the experiment , The experimenter needs to agree in advance on an expectation of the level of significance , That is, limit " The original assumption was correct but rejected " The upper limit of probability is α, The generally accepted probability range for making such errors is 0%～5%. In the statistical calculation of the experiment , We use it p-value To represent the actual probability of making such a mistake in an experimental scenario . as long as p-value Less than what we set in advance α, The experiment reached the standard at the significance level . Calculating the significance level is a method used to combat sampling errors , When our experiment reaches the standard at the significance level , We think there are 95% The above assurance shows that the experimental results are reliable . This greatly reduces , The possibility of the negative impact of the game modification on the indicators .

 

《 The whole people bomb 》 Is a simple operation 、 Easy to use ball jumping tower leisure games , The player can make the marble penetrate the colored bricks layer by layer , Enjoy the pure pleasure of ejection , You can also unlock more unknown balls , Experience the forming fun of synthesis . Game operation is to verify the impact of the game home page on the player's retention and advertising data , Compare the two home page schemes ： Tower jumping page （ The control group ）、 Composite page （ test group ）

after 2 Zhou's experimental test , Results show , The first page is a composite page in 3 Japan 、7 The core indicators such as daily retention and the effect of advertising are better than those of the control group . Finally, choose the scheme of full volume composition page , Online and achieved excellent results .

 

In addition to the above scenes ,A/B Experiments are also often used to verify the following changes. Novice guided changes , Novice task difficulty adjustment , Difficulty of initial resource adjustment and other checkpoints 、 Equipment explosion rate 、 Changes to the mall interface that affect the game experience, such as the adjustment of the winning rate of the treasure chest , Integral system , Bulletin board 、banner、 Button 、 The change of functional areas such as icons matches the system 、 Friend system 、 Chat system and other social functions change the game IP、NPC、 Game scenario 、 Plot change experiment stratification , Expand online traffic

Experimental stratification , Expand online traffic

Small A To validate new weapon systems , need 50% Flow verification for ; Small B Modify the mall interface , Also needed 50% Of traffic ; Small C Improved the novice task , Also need to 50% Flow verification for . At the same time 3 An experiment , need 150% Of traffic . The online traffic is not enough , Only experiments can be prioritized , Let several experiments be carried out successively , But this leads to inefficiency .

 

adopt A/B Experimental traffic layering technique , Copy the total traffic “ Countless times ”, formation ” Countless ” Flow layer . The flow layers are orthogonal to each other , In addition, the experimental traffic running on the same layer is mutually exclusive , So as to ensure that multiple experiments run at the same time will not interfere with each other .

 

Experimental layering technology can effectively solve the problem of insufficient online traffic , Especially in the game beta stage , It is often difficult to obtain users at this stage . You need to release the internal test task , Get gamers by small-scale buying , The cost of traffic acquisition is high . At this stage, you need to verify the performance of the game 、 Compatibility , But also on the way the game is played 、 Level settings 、 Story plot 、 Adjust the revenue model . If the internal test time is too long , Not only does it require sustained buying , It will also slow down the online rhythm of the game , You may even miss the hot spots in the market .

adopt A/B Experimental traffic layering technique , Break the flow limit , Greatly shorten the game verification time , Fast online iteration .

Scene capabilities , Make promotion more efficient

 

In addition to the traditional client 、 Other than the server experiment , We also provide for special scenarios A/B Experimental ability , Make the experiment more convenient , The effect is more remarkable .

 

• Visualization experiments ： Through WYSIWYG online editing , Reduce in Web/H5 Page optimization scenario , Products and operators use A/B The cost of experimental tools . Promotion page for game activities , The effect comparison of the lottery page , Go online quickly without programming .

 

 

• Push experiment ： Through the title of the push notification 、 Content 、 Click action to test , Find the best solution . Support multiple push channels 、 Device type ; Can be applied to game pull new 、 Recall and campaign promotion , Timely response results , Avoid negative effects .

• Advertising experiment ： Improve advertising ROI, Optimize advertising strategy , You can compare different experimental texts 、 Advertising effects of different delivery channels . Control by transforming efficiency measures “ Natural transformation ” Influence on the experimental effect , Ensure the scientificity and preciseness of advertising experiments .

The game goes out to sea ,A/B Escort

 

Nowadays, global operation has become a new trend , Compared with domestic game revenue, overseas game revenue has higher growth . Domestic self-developed games have gone to sea , Quickly seize the share of key regions . At the same time, it also brings differentiated operation 、 Global publishing 、 New challenges such as information security .

 

• Regional differentiated operation

Because of the culture of each country 、 Religious beliefs 、 Differences in trends , Same game activities and modifications , In different regions overseas, the benefits are often different . For the company , Differentiated operations in different overseas regions , It is often necessary to accumulate local operation experience , The cost is usually high . Blindly promote , It is easy to lead to some negative effects .

A/B The experiment supports the selection of game users in different regions , Targeted tests can be carried out for users in different regions , Can quickly compare experimental results , Determine the best plan for the area .

 

 

• Global publishing

Global release is also a challenge that the company has to face in the process of going to sea . Global release means an increase in operation and maintenance costs , At the same time, multi version management also increases the complexity of distribution . Running A/B experiment , Support direct conversion to features for publishing . It also supports progressive publishing 、 Scheduled release , And rollback operations . By using the feature publishing function , Make the publishing process more concise , Reduce release risk .

 

 

Volcano engine A/B test , Time limited free , Apply now ！

A/B test , Get rid of speculation , Use scientific experiments to measure the benefits of decision-making , Build better products , Let every step of the business lead to growth . Volcanic engine releases growth boost for the first time 「 Kindling plan 」, Volcano engine A/B Test as 「 Kindling plan 」 One of the products , We will provide you with... Free of charge 2 Billion events and 5 ten thousand MAU, And up to 12 A month's right to use . Click on try now .