Managing Mobile Market Users Based on the AARRR Model in the Age of Big Data
大数据时代基于 AARRR 模型的移动市场用户管理

${}^{[a]}$${}^{[a]}$^([a]){ }^{[a]} School of Business Administration, South China University of Technology, Guangzhou, China.
${}^{[a]}$${}^{[a]}$^([a]){ }^{[a]} 华南理工大学工商管理学院，中国广州。
*Corresponding author. *通讯作者。
Received 16 October 2015; accepted 28 February 2016
2015 年 10 月 16 日接收;接受日期 2016 年 2 月 28 日
Published online 16 March 2016
2016 年 3 月 16 日在线发布

Key words: Customer relationship management; User lifecycle; AARRR model; Data mining; Mobile internet application
关键词： 客户关系管理;用户生命周期;AARRR 模型;数据挖掘;移动互联网应用程序

Chen, Q. L., & Du, L. (2016). Managing Mobile Market Users Based on the AARRR Model in the Age of Big Data. Management Science and Engineering, 10(1), 58-66. Available from: URL: http://www.cscanada.net/index.php/mse/article/view/8208 DOI: http://dx.doi.org/10.3968/8208
陈倩骏和杜倪L. （2016）。在大数据时代基于 AARRR 模型管理移动市场用户。管理科学与工程， 10（1）， 58-66.可从： URL： http://www.cscanada.net/index.php/mse/article/view/8208 DOI： http://dx.doi.org/10.3968/8208

With the computer science and technology and the rapid development of Internet, information is also a way to rapid transmission of explosive. The information explosion has two meanings. One is explosive growth of data; the other is the explosion of the data value. The data contain great knowledge and value, the increasing
随着计算机科学技术的进步和互联网的飞速发展，信息也成为一种快速传播的爆炸性途径。信息爆炸有两个含义。一个是数据的爆炸式增长;另一个是 data 值的爆炸。数据包含丰富的知识和价值，不断增长的
amount of data that makes the information increasingly rich and valuable.
使信息越来越丰富和有价值的数据量。

The advent of the era of big data and Mobile Internet has brought the infinite opportunity to telecom operators. Telecom operators offer all channels for data transfer and exchange, which is the congenital advantage of the telecom operators and they would be the most potential competitors in the era of big data. The telecom operators have the largest data resources of the users, including the user identity data, telecom communication data, and the behavior data of searching the internet and so on. These resources will become the most important competitive factors of the telecom operators. The telecom operators should use the digital management to conduct service innovation so that they can meet the change and the opportunities of the era of big data.
大数据和移动互联网时代的到来，给电信运营商带来了无限的机遇。电信运营商提供数据传输和交换的所有渠道，这是电信运营商的先天优势，他们将是大数据时代最具潜力的竞争对手。电信运营商拥有最大的用户数据资源，包括用户身份数据、电信通信数据、互联网搜索行为数据等。这些资源将成为电信运营商最重要的竞争因素。电信运营商应该利用数字化管理进行业务创新，以应对大数据时代的变化和机遇。

Mobile application stores as the “habitat” of the mobile internet applications and itself is an application also. It is the windows for the developers to promote their mobile internet applications. In order to take advantage of the users scale, The Telecom Operator launched the Mobile Market application store (hereafter referred to as MM). The user can download the mobile application from the MM in their mobile phone, the developers can management their application through the developer community and the telecom operator can categorize and sell the applications through shelf management and the analysis of user personalized information. MM as The Telecom Operator Value-Added Service Platform, it carries out build the industrial chain of the mobile phone software of responsibility, so the user management of MM is crucial important for The Telecom Operator in the competition between The Telecom Operator and the OTT (Over the top) application server provider.
移动应用程序商店作为移动互联网应用程序的“栖息地”，它本身也是一个应用程序。它是开发人员推广其移动互联网应用程序的窗口。为了利用用户规模，电信运营商推出了 Mobile Market 应用商店（以下简称 MM）。用户可以在手机上从 MM 下载移动应用程序，开发人员可以通过开发者社区管理他们的应用程序，电信运营商可以通过货架管理和用户个性化信息分析对应用程序进行分类和销售。MM作为电信运营商增值服务平台，负责构建手机软件的产业链，因此MM的用户管理对于电信运营商在电信运营商与OTT（Over the top）应用服务器提供商的竞争中至关重要。

According to customer management, conducting comprehensive, effective and fine management around the customer life cycle and utilizing different management
根据客户管理，围绕客户生命周期进行全面、有效、精细的管理，利用不同的管理
strategies to manage the customer in different life cycle phases is essential to the customer relationship management (CRM) of MM. The concept of customer life cycle was first put forward by Ives and Learmonth (1984), they proposed that introduced the time variables into the customer relationship research (Ives & Learmonth, 1984). In the research of customer life-cycle management, buyerseller relationship in marketing channels introduced by Dwyer had important significance, and the customer life cycle was divided into five general phases: awareness, exploration, expansion, commitment, dissolution ( Dwyer & Schurr, 1987). To evaluate the customer life cycle management in the companies, the research in this field was mainly aim at calculating the value of customer life cycle as the indicator and utilizing different marketing strategies in different life cycle phases without specific operation model to conduct the customer management. Social network plays an important role in the Internet age and “go viral” which is utilizing the existing customer to spread the information about the potential new customer becomes a new way of customer acquisition. So we need an operation model which conforms to the mobile internet age to conduct the mobile internet applications operations.
在不同生命周期阶段管理客户的策略对于 MM 的客户关系管理 （CRM） 至关重要。客户生命周期的概念最早是由Ives和Learmonth（1984）提出的，他们提出将时间变量引入客户关系研究（Ives & Learmonth，1984）。在客户生命周期管理的研究中，Dwyer引入的营销渠道中的买方销售关系具有重要意义，客户生命周期分为五个一般阶段：认知、探索、扩展、承诺、消解（Dwyer & Schurr，1987）。为了评价企业的客户生命周期管理，该领域的研究主要以计算客户生命周期的价值为指标，在不同的生命周期阶段利用不同的营销策略，没有特定的运营模式进行客户管理。社交网络在互联网时代发挥着重要作用，利用现有客户传播有关潜在新客户的信息的“病毒式传播”成为一种新的客户获取方式。所以我们需要一种符合移动互联网时代的运营模式来进行移动互联网应用的运营。

In our research, we introduce the data- analysis platform which is developed by The Telecom Operator in the age of big data, and we utilize the AARRR model which depends on the feature of MM to evaluate the customer management of MM.
在我们的研究中，我们介绍了电信运营商在大数据时代开发的数据分析平台，并利用依赖于 MM 特性的 AARRR 模型来评估 MM 的客户管理。

CRM refers to a customer-focused business strategy. The concept of CRM is not new and it opens the mind and became an attitude to customers and to the company itself. To create and add value for the company and its customers, CRM dynamically integrates three elements: sales, marketing and the customer care service (Ricardo, 2006).
CRM 是指以客户为中心的业务策略。CRM 的概念并不新鲜，它打开了人们的思路，并成为对客户和公司本身的一种态度。为了为公司及其客户创造和增加价值，CRM 动态地集成了三个要素：销售、营销和客户服务（Ricardo，2006 年）。

The definitions of CRM are various in the literature. CRM was defined as “a set of business processes and overall policies designed to capture, retain and provide service to customers.” by Scott (Scott, 2001). “A coherent and complete set of processes and technologies for managing relationships with current and potential customers and associates of the company, using the marketing, sales and service departments, regardless of the channel of communication” was the definition of CRM defined by Injazz and Karen (Injazz & Karen, 2004). Generally, CRM is defined as the “management of mutually beneficial relationship(s) from the seller’s perspective” (LaPlaca, 2004, p.463). However, the definition of CRM requires a more specific definition. Current definitions of CRM can be classified into one of two categories: strategic or operational, and the strategic
CRM 的定义在文献中多种多样。CRM 被定义为“一组旨在捕获、保留和向客户提供服务的业务流程和整体策略”，作者 Scott （Scott， 2001）。“一套连贯且完整的流程和技术，用于管理与公司当前和潜在客户以及员工的关系，使用营销、销售和服务部门，无论沟通渠道如何”是Injazz和Karen定义的CRM（Injazz & Karen，2004）。通常，CRM 被定义为“从卖方的角度管理互惠关系”（LaPlaca，2004 年，第 463 页）。但是，CRM 的定义需要更具体的定义。CRM 的当前定义可分为两类之一：战略或运营，以及战略
definitions and operational definitions maybe closely related (Keith & Eli, 2008). The operational definitions of CRM are more closely related to the processes and technologies associated with enabling better customer relationships, for example, the operational explanation of CRM, “CRM allows companies to gather customer data swiftly, identify the most valuable customers over time, and increase customer loyalty by providing customized products and services” (Rigby et al., 2002, p.101). The technologies and processes associated with CRM in the operational definition also implies one of the most popular trends that CRM is a series of information technology (IT) products oriented towards automating some business processes such as marketing, sales or services (Kirby, 2001). However, CRM should be viewed more than implementation of IT (Light, 2001). As IT is one of the important roles in CRM, the role of IT tools in CRM impacts the different customer stages: initiation, maintenance, and termination (Reinartz, Krafft, & Hoyer, 2004). Moreover, a well-organized IT infrastructure and architecture can facilitate different customer processes, such as segmentation of customers based on their value or prediction of customer behaviours (Clark & Smith, 2003; Gummesson, 2002; Ryals & Knox, 2001). The objectives of CRM include acquiring new customers, enhancing the profitability of existing customers and retaining profitable customers (Tsai, Hu, & Lu, 2015). Customer retention is a strategic imperative for most firms (Anderson & Mittal, 2000). Firms want to understand how to retain customers (Bolton, Lemon, & Bramlett, 2006; Gounaris, 2005), and through better customer relationship and customer experience they can minimise the risk of customer loss (Granovetter, 1973; Haleblian, Devers, McNamara, Carpenter, & Davison, 2009; Heinonen et al., 2010). For the experience-based organizations, if they can identify who the most valuable customer will be, where they live, how they purchase, and other kinds of data, there will be a tremendous amount of readily available data and marketing models (Harvard Business Review, 2015). Life-cycle service offerings are based on long-term relationships and focused on understanding the activities that customers perform when using and/or operating a product (Cornet et al., 2000; Davies, 2004; Galbraith, 2002; Johnstone, Dainty, & Wilkinson, 2009; Prahalad & Ramaswamy, 2000; Windahl et al., 2004; Wise & Baumgartner, 1999). For mobile internet applications, their customer processes are digital processes, and automation is one of the key capabilities. Automation involves the digitization and streamlining of steps in the journey that were formerly done manually (Edelman & Singer, 2015). Proactive personalization in the mobile internet application is building on the automation capability which means companies should take information gleaned either from past interactions with a customer or from existing resources and use it to instantaneously customize the
定义和操作定义可能密切相关（Keith & Eli，2008）。CRM 的运营定义与实现更好的客户关系相关的流程和技术更密切相关，例如，CRM 的运营解释，“CRM 允许公司快速收集客户数据，随着时间的推移识别最有价值的客户，并通过提供定制的产品和服务来提高客户忠诚度”（Rigby 等人，2002 年， 第 101 页）。运营定义中与 CRM 相关的技术和流程也暗示了最流行的趋势之一，即 CRM 是一系列面向自动化某些业务流程（如营销、销售或服务）的信息技术 （IT） 产品（Kirby，2001）。然而，应该更多地看待 CRM 而不是 IT 的实施 （Light， 2001）。由于IT是CRM中的重要角色之一，IT工具在CRM中的作用会影响不同的客户阶段：启动、维护和终止（Reinartz， Krafft， & Hoyer， 2004）。此外，一个组织良好的IT基础设施和架构可以促进不同的客户流程，例如根据客户的价值或对客户行为的预测来细分客户（Clark & Smith，2003年;Gummesson， 2002;Ryals & Knox，2001 年）。CRM 的目标包括获取新客户、提高现有客户的盈利能力和留住盈利客户（Tsai、胡 和 Lu，2015 年）。客户保留是大多数公司的战略必要条件（Anderson & Mittal，2000年）。 公司希望了解如何留住客户（Bolton， Lemon， & Bramlett， 2006;Gounaris，2005 年），通过更好的客户关系和客户体验，他们可以最大限度地降低客户流失的风险（Granovetter，1973 年;Haleblian， Devers， McNamara， Carpenter， & Davison， 2009;Heinonen et al.， 2010）。对于基于体验的组织来说，如果他们能够确定谁是最有价值的客户、他们住在哪里、他们如何购买以及其他类型的数据，那么将有大量现成的数据和营销模型（哈佛商业评论，2015 年）。生命周期服务产品基于长期关系，侧重于了解客户在使用和/或操作产品时执行的活动（Cornet 等人，2000 年;Davies， 2004;Galbraith， 2002;Johnstone， Dainty， & Wilkinson， 2009;Prahalad & Ramaswamy， 2000;Windahl 等人，2004 年;Wise & Baumgartner，1999年）。对于移动互联网应用程序，他们的客户流程是数字流程，自动化是关键功能之一。自动化涉及到以前手动完成的旅程步骤的数字化和简化（Edelman & Singer，2015）。移动互联网应用程序中的主动个性化建立在自动化功能之上，这意味着公司应该从过去与客户的互动或现有资源中收集信息，并使用它来即时定制
customer’s experience (Edelman & Singer, 2015). The branded mobile phone applications (branded apps) will be widely used in the coming years (Zhao & Christine, 2015). The branded apps are marketing devices because their high level of user engagement (Calder, Malthouse, & Schaedel, 2009). There are five dimensions of branded apps: tool-, game-, social-, m-commerce-, or designcentric (Zhao & Christine, 2015). For tool-centric apps, the main goals are to identify the motivations and requirements of customers in using/buying products and to develop service to assist customers in these processes (Zhao & Christine, 2015). Customer life-cycle service is important for the CRM of mobile internet applications, and CRM requires operation model to customize the customer’s experience and retaining profitable customers.
客户体验（Edelman & Singer，2015 年）。品牌手机应用程序（品牌应用程序）将在未来几年得到广泛使用（Zhao & Christine，2015）。品牌应用程序是营销工具，因为它们的用户参与度高（Calder， Malthouse， & Schaedel， 2009）。品牌应用有五个维度：工具、游戏、社交、移动商务或以设计为中心（Zhao & Christine，2015）。对于以工具为中心的应用程序，主要目标是确定客户使用/购买产品的动机和要求，并开发服务以在这些过程中帮助客户（Zhao & Christine，2015）。客户生命周期服务对于移动互联网应用的 CRM 非常重要，CRM 需要运营模式来定制客户体验并留住盈利客户。

AARRR model which is widely used to build the customer-management evaluation system, was first introduced by Dave McClure in the Start Up Metrics for Pirates and it is a management framework of customer life-cycle and operations objectives. In figure 1, Acquisition, Activation, Retention, Revenue and Refer are the elements of AARRR model and each of them means the five most important stages of mobile internet applications (McClure, 2007). The Telecom Operator modified the AARRR operation model based on the features of MM.
AARRR 模型被广泛用于构建客户管理评估系统，由 Dave McClure 在 Pirates 的启动指标中首次引入，它是客户生命周期和运营目标的管理框架。在图 1 中，获取、激活、保留、收入和推荐是 AARRR 模型的要素，它们每个要素都意味着移动互联网应用程序的五个最重要的阶段（McClure，2007 年）。电信运营商根据 MM 的特点修改了 AARRR 运营模式。

Figure 1 图 1
The AARRR Model AARRR 模型
The first A in AARRR represents Acquisition, which means the objective of customer management at this stage is acquiring new customers. In this stage, the telecom operator needs to focus on the increasement of new customers and evaluate the performance of different distribution channels.
AARRR 中的第一个 A 代表 Acquisition，这意味着此阶段客户管理的目标是获取新客户。在这个阶段，电信运营商需要专注于增加新客户，并评估不同分销渠道的表现。

The second A in AARRR represents Activation, which means the objective of customer management at this stage
AARRR 中的第二个 A 代表激活，即此阶段客户管理的目标
is increasing the level of activity. In this stage, the telecom operator not only needs to focus on the current activity level of customers but also their sustained activity in given time.
正在提高活动水平。在这个阶段，电信运营商不仅需要关注客户的当前活动水平，还需要关注他们在给定时间内的持续活动。

The first R in AARRR represents Retention, which means the objective of customer management at this stage is increasing the rate of customer retention. Generally, the cost of retaining an existed customer is far less than acquiring a new customer.
AARRR 中的第一个 R 代表留存率，这意味着此阶段客户管理的目标是提高客户留存率。一般来说，留住现有客户的成本远低于获得新客户的成本。

The second R in AARRR represents Revenue, which means the objective of customer management at this stage is obtaining income. In this stage, “income” is generalized to “output” in The Telecom Operator. The Telecom Operator not only focuses on the direct revenue but also pay attention to the indirect revenue, such as downloads and customers’ data traffic.
AARRR 中的第二个 R 代表收入，这意味着这个阶段客户管理的目标是获得收入。在这个阶段，“收入”被推广到电信运营商中的“产出”。电信运营商不仅关注直接收入，还关注间接收入，例如下载和客户的数据流量。

The third $R$$R$RR in AARRR represents Refer, which means the objective of customer management at this stage is virus transmission. Thanks to the social network, virus transmission becomes the new method of acquiring new customers. Low cost and spread effectively are two advantages of this method. K index is the indicator of virus transmission performance, which means the spread of conversion (people of spreading successful divide total spread times).
AARRR 中的第三个 $R$$R$RR 代表 Refer，这意味着此阶段客户管理的目标是病毒传播。多亏了社交网络，病毒传播成为获取新客户的新方法。低成本和有效的传播是这种方法的两个优点。K 指数是病毒传播性能的指标，即传播转化率（传播成功人数除以总传播次数）。

The company can discover the problems in every stage of customer management performance in time through this AARRR operation model and identify unmet needs.
公司可以通过这种 AARRR 运营模式及时发现客户管理绩效各个阶段的问题，并识别未被满足的需求。

As shown in Figure 2, in order to meet with the era of big data, The Telecom Operator has established the digitized management system, set up the basic data platform.
如图 2 所示，为了迎接大数据时代，电信运营商建立了数字化管理系统，建立了基础数据平台。

The company can collect the customer information in time through the basic data platform, and the company also set up the data-label database, which as shown in Figure 3.
该公司可以通过基础数据平台及时收集客户信息，并且该公司还建立了数据标签数据库，如图 3 所示。

The data-label database consists of two parts: The data attribute unique to telecom operator and the common data labels of Internet firms. The common data labels of Internet firms includes social attribute, consumption attribute, socialization and the attribute of product using, which are the common labels in Internet firms. The social attribute contains the sex of customer, the age of customer, the education of customer, the career of customer and so on. The consumption attribute contains the shopping trips, the sum of money, the shopping channels and so on. Socialization contains the degree of social, the main social circle, channel preference, the frequency of social activities and so on. The attribute of product using contains product types, the usage scenario, the function
数据标签数据库由电信运营商特有的数据属性和互联网公司常用的数据标签两部分组成。互联网企业常见的数据标签包括社会属性、消费属性、社会化和产品使用属性，这些都是互联网企业常见的标签。social 属性包含客户的性别、客户的年龄、客户的教育程度、客户的职业生涯等。consumption 属性包含购物行程、金额总额、购物渠道等。社会化包含社交程度、主要社交圈、渠道偏好、社交活动频率等。product using 的属性包含产品类型、使用场景、功能
usage and the stickiness of product usage and so on. These five common data labels can depict the customers in real
使用情况和产品使用情况的粘性等。这五个常见的数据标签可以真实地描述客户
life situations, their preferences and their use of internet products.
生活状况、他们的偏好和他们对互联网产品的使用。

Figure 2 图 2

Figure 3 图 3
The Data-Label Database Data-Label 数据库

The marketers chose five million customers to conduct the SMS marketing in a marketing activity, and the goal of this marketing activity is acquiring ten thousand new customers. However, the effect of the marketing is not as expected and there are only five thousand new customers.
营销人员在营销活动中选择了 500 万客户进行 SMS 营销，此营销活动的目标是获取 1 万新客户。然而，营销的效果并不如预期，只有 5000 个新客户。

In order to solve this problem, the marketer utilized persona to conduct the potential customer analysis to rediscover the target customers. Persona means depict the features of the customer through the mathematical language when the customer volume is enough.
为了解决这个问题，营销人员利用角色进行潜在客户分析，以重新发现目标客户。Persona 是指当客户数量足够时，通过数学语言描述客户的特征。

In our research, we chose the active customer of MM to conduct the persona and find out the features of target
在我们的研究中，我们选择了 MM 的活跃客户来进行角色并找出目标的特点
customer. As shown in Figure 4, with traffic flow and traffic charge, for example, comparing them with the total customers PV (PV: Page View), we find that the customer whose traffic flow is over 30 M or traffic charge is more
客户。如图 4 所示，以流量和流量费用为例，将它们与总客户 PV（PV：页面浏览量）进行比较，我们发现流量超过 30 M 或流量费用更多的客户
than 10 yuan will be more active than other customers. So we can make a conclusion that the customer with the feature we mentioned above will be more likely to become the active customer of MM.
超过 10 元会比其他客户更活跃。所以我们可以得出一个结论，具有我们上面提到的功能的客户将更有可能成为 MM 的活跃客户。

Figure 4 图 4

The other attribute of MM in the database was analysed in the way described above and the analyst gave the features of the target customers. However, because of the company’s secrecy work, we just showed part of the conclusion information which as shown in Table 1. According to the features of target customers, which can enhance the marketing effect, the marketer could reselect five million potential customers to conduct the SMS marketing and got fourteen thousand new customers in the end.
以上述方式分析了数据库中 MM 的另一个属性，分析师给出了目标客户的特征。但是，由于公司的保密工作，我们只展示了部分结论信息，如表 1 所示。根据目标客户的特点，可以增强营销效果，营销人员可以重新选择 500 万潜在客户进行短信营销，最终获得 1.4 万新客户。

Although the customer down-load and installs the MM,
尽管客户下载并安装了 MM，
the frequency and using time of MM use is far less than that of the competitive applications. How to improve the activation of the customer becomes the most urgent problem. Utilizing the platform of The Telecom Operator user experience centre, the company invites different kinds of customers to conduct the product use experience and survey so that the company can find out the imperfection of MM.
MM 的使用频率和使用时间远少于竞争应用程序。如何提高客户的激活率成为最紧迫的问题。利用 The Telecom Operator 用户体验中心的平台，公司邀请不同类型的客户进行产品使用体验和调查，以便公司找出 MM 的不足之处。

The analysts arranged the information collected from the user experience centre and identified that MM had three main imperfections in Table 2.
分析师对从用户体验中心收集的信息进行了整理，并在表 2 中确定了 MM 有三个主要缺陷。

Table 1 表 1
The Features of Different Kinds of Customers
不同类型客户的特点

The retention in MM contains four parts: losing user structure analysis, losing user behavioural differences analysis, the reasons of user losing analysis and user losing warning and monitor system.
MM 中的留存包含四个部分：流失用户结构分析、流失用户行为差异分析、用户流失原因分析以及用户流失预警和监控系统。

The main task of losing user structure analysis is depicting the structure of the losing user and find out the features of the losing users and their telecom
流失用户结构分析的主要任务是描绘流失用户的结构，并找出流失用户及其电信的特点
communication information. As shown in Figure 5, we find that most of the losing users use domestic brand mobile phone such as ZTE, Lenovo and Coolpad. And the mobile phone brand used between the losing user and the retain user are quite different. We can make a preliminary conclusion that the customer who uses cheaper mobile phones may be more likely to become the losing users.
通信信息。如图 5 所示，我们发现大多数流失用户使用的是中兴、联想、酷派等国产品牌手机。而流失用户和留存用户之间使用的手机品牌差异很大。我们可以初步得出结论，使用更便宜的手机的客户可能更有可能成为流失的用户。

Figure 5 图 5

The main task of losing user behavioural differences analysis is analysing the essential differences between the losing user and loyal user in MM usage. In our research, we find that the loyal customer and the losing customer performed great different in the average MM visiting days
流失用户行为差异分析的主要任务是分析流失用户和忠诚用户在 MM 使用方面的本质差异。在我们的研究中，我们发现忠诚顾客和流失顾客在平均 MM 访问天数中的表现差异很大
per month and the average PV per month, which means that before the customer lose, the user agglutinant has been decreased. As shown in Figure 6 and Figure 7, we can use average MM visiting days per month and the average PV per month as the important monitoring indicators.
和每月的平均 PV，这意味着在客户亏损之前，用户凝集物已经减少。如图 6 和图 7 所示，我们可以将每月平均 MM 访问天数和每月平均 PV 作为重要的监测指标。

Figure 6 图 6

The main task for the reasons of user losing the analysis is to find out the crucial cause of the customer lose. Combining qualitative method with quantitative method to find out the main cause of customer losing, we can get the point cut to improve the product. In the
对用户流失原因的分析主要任务是找出客户流失的关键原因。将定性方法与定量方法相结合，找出客户流失的主要原因，我们可以获得积分以改进产品。在
qualitative research, we chose part of the customer who just leaves MM to conduct this research and find out these two main problems cause them losing: the interface and the content. The interface of MM is not user-friendly because of aesthetics deficiency, no guidance, imperfect
定性研究，我们选择了刚刚离开 MM 的部分客户进行这项研究，并找出导致他们失去的这两个主要问题：界面和内容。MM的界面因为美观不足，没有指导，不完美，所以对用户不友好
display effect, mixed sections and too many functions. Content in MM can not satisfy the needs of the customer,
显示效果、混合截面和太多功能。内容在MM不能满足客户的需求，
the resource is not sufficient, the quality of the content is not unified and so on.
资源不足、内容质量不统一等等。

Figure 7 图 7
Average PV Per Month 每月平均 PV

Customer losing warning and monitor system can help the company to monitor the customers’ behaviour any time and master the unexpected behaviours of the customers and carry out the marketing activities in time to retain the customers. Through what we mentioned above, the customers who were going to give up using MM would have some unexpected behaviours in the average MM visiting days per month and the average PV per month, we can dovetailed this two indicators into our customer losing warning and monitor system.
客户流失预警和监控系统可以帮助公司随时监控客户的行为，掌握客户的意外行为，及时开展营销活动以留住客户。通过我们上面提到的，打算放弃使用 MM 的客户在每月的平均 MM 访问天数和每月的平均 PV 中会出现一些意想不到的行为，我们可以将这两个指标与我们的客户流失警告和监控系统相吻合。

As a mobile application store, the main income of MM is the share revenue of application download from MM. Although the number of users has reached a certain scale, the users who would pay for the application is low, which
作为移动应用商店，MM 的主要收入是 MM 的应用下载份额收入。虽然用户数量已经达到了一定的规模，但愿意为应用程序付费的用户却很少，这
leads to the speed of revenue increase is not the same as the numbers of customers increase. How to improve the download-pay conversions is the main task in the revenue stage and the key is to how to prompt the customers to download the application with high download-pay conversions.
导致收入增长的速度与客户数量的增加不同。如何提高下载付费转化率是收入阶段的主要任务，关键是如何提示客户下载下载付费转化率高的应用程序。

Firstly, we need to analyze the structure of revenue. The revenue of MM from the game application is $96.17\mathrm{\%}$$96.17\mathrm{\%}$96.17%96.17 \% which makes game application become the primary income of MM.
首先，我们需要分析收入的结构。MM 来自游戏应用的收入使 $96.17\mathrm{\%}$$96.17\mathrm{\%}$96.17%96.17 \% 游戏应用成为 MM 的主要收入。

We analyzed the download-pay performance of all kinds of games, and used the download-pay conversions as the indicator of revenue.
我们分析了各类游戏的下载付费表现，并将下载付费转化作为收入指标。

Download-pay conversion $=\frac{\text{(payment customers)}}{(\text{download customers})}$$=\frac{\text{ (payment customers) }}{(\text{ download customers })}$=(" (payment customers) ")/((" download customers "))=\frac{\text { (payment customers) }}{(\text { download customers })}.
下载-付费转换 $=\frac{\text{(payment customers)}}{(\text{download customers})}$$=\frac{\text{ (payment customers) }}{(\text{ download customers })}$=(" (payment customers) ")/((" download customers "))=\frac{\text { (payment customers) }}{(\text { download customers })} .

Figure 8 图 8
The Download and Payment of Different Kinds of Games
各种游戏的下载和付款

As shown in Figure 8, we can find that the downloadpay conversions of the leisure games and racing games are higher than the others, so the marketer can focus on these two kinds of games to improve the download-pay conversions.
如图 8 所示，我们可以发现休闲游戏和赛车游戏的 downloadpay 转化率高于其他游戏，因此营销人员可以专注于这两类游戏来提高下载付费转化率。

The customer choose different kinds of games just like choose the daily necessities in the market, so we introduce the method of “Market Basket Analysis” to the MM games download-pay conversions analysis and then we can analyze the preference of the customers and find out the association rules among different kinds of games in order to recommend the associated games to the customers.
客户选择不同种类的游戏就像选择市场上的日用品一样，所以我们在MM游戏下载-付费转化分析中引入了 “市场篮分析” 的方法，然后我们可以分析客户的偏好，找出不同类型游戏之间的关联规则，以便向客户推荐相关的游戏。

Association analysis is widely used in the transaction database to extract the interesting associations (Berry & Linoff, 2004), so that the company can conduct service innovation and customization through the high quality and practical information. Here we utilize the association analysis to analyze the customer usage pattern of the games: when game A was played by the customer, how much the probability that the game B would be played by the customer also.
关联分析在交易数据库中被广泛使用，以提取有趣的关联（Berry & Linoff，2004），以便公司可以通过高质量和实用的信息进行服务创新和定制。在这里，我们利用关联分析来分析游戏的客户使用模式：当客户玩游戏 A 时，客户也玩游戏 B 的概率是多少。
$D=\{T_1,T_2,\dots .,T_n\}:$$D=\left\{T_1,T_2,\dots .,T_n\right\}:$D={T_(1),T_(2),dots.,T_(n)}:D=\left\{T_{1}, T_{2}, \ldots ., T_{n}\right\}: The set of a given game payment
$D=\{T_1,T_2,\dots .,T_n\}:$$D=\left\{T_1,T_2,\dots .,T_n\right\}:$D={T_(1),T_(2),dots.,T_(n)}:D=\left\{T_{1}, T_{2}, \ldots ., T_{n}\right\}: 给定游戏付款的集合
$I=\{i_1,i_2,\dots \dots ,i_m\}$$I=\left\{i_1,i_2,\dots \dots ,i_m\right\}$I={i_(1),i_(2),dots dots,i_(m)}I=\left\{i_{1}, i_{2}, \ldots \ldots, i_{m}\right\} : The set of games payment
$I=\{i_1,i_2,\dots \dots ,i_m\}$$I=\left\{i_1,i_2,\dots \dots ,i_m\right\}$I={i_(1),i_(2),dots dots,i_(m)}I=\left\{i_{1}, i_{2}, \ldots \ldots, i_{m}\right\} ：游戏支付套装
Every set from the set of a given game payment is the subset of the set of games payment that $T_i\in I$$T_i\in I$T_(i)in IT_{i} \in I.
给定游戏付款集中的每组都是 的游戏付款集的子集。 $T_i\in I$$T_i\in I$T_(i)in IT_{i} \in I

The association rules of the games payment are $X\to Y$$X\to Y$X rarr YX \rightarrow Y, $X,Y\in I$$X,Y\in I$X,Y in IX, Y \in I, and $X\cap Y=\varnothing .X$$X\cap Y=\varnothing .X$X nn Y=O/.XX \cap Y=\varnothing . X is the antecedent and $Y$$Y$YY is the consequence. For the association rules $X\to Y$$X\to Y$X rarr YX \rightarrow Y :
游戏支付的关联规则是 $X\to Y$$X\to Y$X rarr YX \rightarrow Y ， $X,Y\in I$$X,Y\in I$X,Y in IX, Y \in I ， 并且 $X\cap Y=\varnothing .X$$X\cap Y=\varnothing .X$X nn Y=O/.XX \cap Y=\varnothing . X 是前提， $Y$$Y$YY 是结果。对于关联规则 $X\to Y$$X\to Y$X rarr YX \rightarrow Y ：

The confidence of a rule $X\to Y$$X\to Y$X rarr YX \rightarrow Y is the proportion of transactions in D that contains X which also contains Y. Rules that have support/confidence greater than userspecified support/confidence are said to have minimum support/confidence.
规则 $X\to Y$$X\to Y$X rarr YX \rightarrow Y 的置信度是 D 中包含 X 且包含 Y 的事务的比例。支持/置信度大于用户指定的支持/置信度的规则称为最低支持/置信度。

Table 3 表 3
The Association Rules of the Games
比赛的协会规则

Through the association analysis of the games, as shown in Table 3, we find that when a customer has pay for the racing games, he/she is very likely to pay for the leisure games (a possibility of $30\mathrm{\%}$$30\mathrm{\%}$30%30 \% ). When a customer has pay for the leisure games, he/she is very likely to pay for the racing games (a possibility of $27\mathrm{\%}$$27\mathrm{\%}$27%27 \% ). Thus, we suggest that when these bundles of games are detected, the company should send marketing information related to
通过对游戏的关联分析，如表 3 所示，我们发现当客户为赛车游戏付费时，他/她很可能会为休闲游戏付费（一种可能性 $30\mathrm{\%}$$30\mathrm{\%}$30%30 \% ）。当客户为休闲游戏付费时，他/她很可能会为赛车游戏付费（一种可能性 $27\mathrm{\%}$$27\mathrm{\%}$27%27 \% ）。因此，我们建议在检测到这些游戏捆绑包时，公司应发送与
the leisure games to the racing games customer and send marketing information related to the racing games to the leisure games customer.
将 Leisure Games 发送给 Racing Games 客户，并将与赛车游戏相关的营销信息发送给 Leisure Games 客户。

As the wide spread of the social media, the information diffusing through the social media platforms cannot compare to current types of advertisement. Utilizing social media platform to conduct viruses-like spreading of MM is a new and effective ways to acquire new customers. In the new version of MM, it added a function about sharing information to their friends through the social media platforms such as wechat and weibo. Thanks to the information spreading through the social media platforms, the MM was known by more potential customers. The marketer can also conduct the marketing activities on the social media platform then when the customer join the marketing activities, the information of MM can also wide spread through the social media while the cost of these kinds of marketing activities is lower than the traditional marketing activities and they are effective.
随着社交媒体的广泛传播，通过社交媒体平台传播的信息无法与当前类型的广告相提并论。利用社交媒体平台进行类似病毒的 MM 传播是一种新的有效方法来获取新客户。在新版本的 MM 中，它增加了通过微信、微博等社交媒体平台与朋友分享信息的功能。由于信息通过社交媒体平台传播，MM 被更多潜在客户所认识。营销人员还可以在社交媒体平台上进行营销活动，然后当客户加入营销活动时，MM 的信息也可以通过社交媒体广泛传播，而这类营销活动的成本低于传统营销活动，而且效果很好。

Our research is based on the theory of Customer Relationship Management and utilizing the AARRR model based on the features of MM to conduct our analysis, set up some management indicators in different stage to monitor the performance of MM and realize the change of the customer in time. Using the data collected from the MM to do the data mining which shows that the AARRR model can guide the operations management of MM and has instructional significance and referenced value for the other mobile internet application operations management.
我们的研究基于客户关系管理理论，并利用基于 MM 特征的 AARRR 模型进行分析，在不同阶段设置一些管理指标，以监控 MM 的表现并及时实现客户的变化。利用从 MM 收集的数据进行数据挖掘，表明 AARRR 模型可以指导 MM 的运营管理，对其他移动互联网应用运营管理具有指导意义和参考价值。

The result of our research indicate that the AARRR model can give some useful introduction to the operations management of MM that the model based on the customer life-cycle management and combine the data analysis and data mining to conduct the operations management in order to do the precision marketing of MM.
研究结果表明，AARRR 模型基于客户生命周期管理，结合数据分析和数据挖掘进行运营管理，从而对 MM 的运营管理进行有益的介绍，从而做好 MM 的精准营销。

Even though this research makes significant contributions to the operations management of mobile internet applications, it nonetheless possesses limitations that open up avenues for future research. First, this study focuses on analyzing structured data. Future research could investigate how to combine unstructured data with structured data to understand customer behaviors and thereby conduct customization and innovation decisionmaking (Chan et al., 2016).
尽管这项研究对移动互联网应用程序的运营管理做出了重大贡献，但它仍然具有为未来研究开辟道路的局限性。首先，本研究侧重于分析结构化数据。未来的研究可以调查如何将非结构化数据与结构化数据相结合，以了解客户行为，从而进行定制和创新决策（Chan et al.， 2016）。

Anderson, E. W., & Mittal, V. (2000). Strengthening the satisfaction-profit chain. Journal of Service Research, 3(2), 107-120.
安德森，E. W.和米塔尔，V.（2000）。加强满意度-利润链。服务研究杂志，3（2），107-120。

Berry, M., & Linoff, G. S. (2004). Data mining techniques for marketing, sales, and customer relationship management. Indianapolis, Indiana: Wiley Publishing.
Berry， M. 和 Linoff， G. S. （2004）。用于营销、销售和客户关系管理的数据挖掘技术。印第安纳州印第安纳波利斯：Wiley Publishing。
Bolton, R. N., Lemon, K. N., & Bramlett, M. D. (2006). The effect of service experiences over time on a supplier’s retention of business customers. Management Science, 52(12), 1811-1823.
博尔顿，R. N.，柠檬，K. N.和布拉姆莱特，M. D. （2006）。服务体验随时间推移对供应商留存业务客户的影响。管理科学， 52（12）， 1811-1823。
Calder, B. J., Malthouse, E. C., & Schaedel, U. (2009). An experimental study of the relationship between online engagement and advertising effectiveness. Journal of Interactive Marketing, 23(4), 321-331.
考尔德，BJ，马尔特豪斯，EC，和沙德尔，美国（2009）。在线参与与广告效果之间关系的实验研究。互动营销杂志，23（4），321-331。
Chan, H. K., Wang, X., Lacka, E., & Zhang, M. (2016). A mixedmethod approach to extracting the value of social media data. Production and Operations Management Advance online publication. DOI: 10.1111/poms. 12390
Chan， H. K.， Wang， X.， Lacka， E.， & Zhang， M. （2016）.一种提取社交媒体数据价值的混合方法。Production and Operations Management Advance 在线出版物。DOI：10.1111/poms。12390
Clark, M., & Smith, B. (2003). Building the foundations for effective CRM. Management Focus, 20, 15-17.
克拉克，M.和史密斯，B.（2003）。为有效的 CRM 奠定基础。管理焦点， 20， 15-17.
Cornet, E., Katz, R., Molloy, R., Schädler, J., Sharma, D., & Tipping, A. (2000). Customer solutions: From pilots to profits (pp.1-15). New York.
Cornet， E.， Katz， R.， Molloy， R.， Schädler， J.， Sharma， D.， & Tipping， A. （2000）.客户解决方案：从飞行员到利润（第 1-15 页）。纽约。
Davies, A. (2004). Moving base into high-value integrated solutions: A value stream approach. Industrial and Corporate Change, 13(5), 727-756.
戴维斯，A.（2004 年）。将基地转移到高价值的集成解决方案中：一种价值流方法。工业与企业变革，13（5），727-756。
Dwyer, F. R., Schurr, P. H., & Sejo, O. (1987). Developing Buyer-Seller Relationships. The Journal of Marketing: 51(2), 11-27.
Dwyer， F. R.， Schurr， P. H.， & Sejo， O. （1987年）。发展买方-卖方关系。市场营销杂志：51（2），11-27。
Edelman, D. C., & Singer, M. (2015). Competing on Customer Journeys. Harvard Business Review, 88-100.
Edelman， D. C.， & Singer， M. （2015）.在 Customer Journeys 上竞争。哈佛商业评论，88-100。
Galbraith, J. R. (2002). Organizing to deliver solutions. Organizational Dynamics, 31(2), 194-207.
加尔布雷斯，JR（2002 年）。组织起来提供解决方案。组织动力学，31（2），194-207。
Gounaris, S. P. (2005). Trust and commitment influences on customer retention: Insights from business-to-business services. Journal of Business Research, 58(2), 126-140.
Gounaris， SP （2005 年）。信任和承诺对客户保留的影响：来自企业对企业服务的洞察。商业研究杂志，58（2），126-140。
Granovetter, M. S. (1973). The strength of weak ties. American Journal of Sociology, 78(6), 1360-1380.
Granovetter， MS （1973 年）。弱关系的力量。美国社会学杂志，78（6），1360-1380。
Gummesson, E. (2002). Relationship marketing and a new economy: It’s timefor de-programming. Journal of Services Marketing, 16(7), 585-589.
Gummesson， E. （2002 年）。关系营销和新经济：是时候去编程了。服务营销杂志，16（7），585-589。
Haleblian, J., Devers, C. E., McNamara, G., Carpenter, M. A., & Davison, R. B. (2009). Taking stock of what we know about mergers and acquisitions: A review and research agenda. Journal of Management, 35, 469-502.
Haleblian， J.， Devers， C. E.， McNamara， G.， Carpenter， M. A.， & Davison， R. B. （2009）.盘点我们对并购的了解：回顾和研究议程。管理杂志， 35， 469-502.
Harvard Business Review. (2015) Boosting demand in the “experience economy”. Retrieved 2015, January 20 from http://hbr.org/2015/01/boosting-demand-in-the-experienceeconomy
哈佛商业评论。（2015） 促进“体验经济”的需求。2015 年 1 月 20 日从 http://hbr.org/2015/01/boosting-demand-in-the-experienceeconomy 检索
Heinonen, K., Strandvik, T., Mickelsson, J., Edvardsson, B., Sundström, E., & Andersson, P. (2010). A customerdominant logic of service. Journal of Service Management, 21(4), 531-548.
Heinonen， K.， Strandvik， T.， Mickelsson， J.， Edvardsson， B.， Sundström， E.， & Andersson， P. （2010）.以客户为导向的服务逻辑。服务管理杂志，21（4），531-548。
Injazz, D., & Karen, P. (2004). Understanding customer relationship management (CRM). People, process and technology. Retrieved from http://www.emeraldinsight. com/1463-7154.htm
Injazz， D. 和 Karen， P. （2004）。了解客户关系管理 （CRM）。人员、流程和技术。取自 http://www.emeraldinsight。com/1463-7154.htm

Ives, B., & Leamonth, G. P. (1984). The Information system as a competitive weapon. Communications of the ACM, 27(12), 1193-1201.
Ives， B.， & Leamonth， GP. P. （1984年）。信息系统作为竞争武器。ACM 通讯，27（12），1193-1201。
Johnstone, S., Dainty, A., & Wilkinson, A. (2009). Integrating products and services through life: An aerospace experience. International Journal of Operations & Production Management, 29(5), 520-538.
Johnstone， S.， Dainty， A.， & Wilkinson， A. （2009年）。将产品和服务贯穿一生：航空航天体验。国际运营与生产管理杂志，29（5），520-538。
Kirby, J. (2001). CRM program management: the art of change. In: Presentations of the Conference on Making the Vision a Reality, Gartner Group, Paris, France.
柯比，J.（2001 年）。CRM 项目管理：变革的艺术。收录于：关于实现愿景的会议的演讲，Gartner Group，法国巴黎。
LaPlaca, P. J. (2004). Letter from the editor: Special issue on customer relationship management. Industrial Marketing Management, 33(6), 463-464.
拉普拉卡，PJ（2004 年）。编辑来信：客户关系管理特刊。工业营销管理，33（6），463-464。
Light, B. (2001). A review of the issues associated with customer relationship management systems (pp.1232-1241). In: Proc. Ninth European Conference on Information Systems - ECIS 2001, Association for Information Systems, Bled, Slovenia.
莱特，B.（2001 年）。与客户关系管理系统相关的问题回顾（第 1232-1241 页）。在：第九届欧洲信息系统会议 - ECIS 2001，信息系统协会，斯洛文尼亚布莱德。
McClure, D. (2007). Startup metrics for pirates. Retrieved from http://500hats.typepad.com/500blogs/2007/09/startupmetrics.html
麦克卢尔，D.（2007 年）。盗版者的启动指标。取自 http://500hats.typepad.com/500blogs/2007/09/startupmetrics.html
Prahalad, C. K., & Ramaswamy, V. (2000). Co-opting customer competence. Harvard Business Review, 79-87.
Prahalad， C. K.， & Ramaswamy， V. （2000年）。利用客户能力。哈佛商业评论，79-87。
Reinartz, W., Krafft, M., & Hoyer, W. D. (2004). The customer relationship management process: Its measurement and impact on performance. Journal of Marketing Research, XLI, 293-305.
Reinartz， W.， Krafft， M.， & Hoyer， W. D. （2004年）。客户关系管理流程：其衡量和对绩效的影响。营销研究杂志，XLI，293-305。
Ricardo, C. (2006). Methodology for customer relationship management. The Journal of Systems and Software, 79, 1015-1024.
里卡多，C.（2006 年）。客户关系管理方法。系统与软件杂志，79,1015-1024。
Richards, K. A., & Jones, E. (2008). Customer relationship management: Finding value drivers. Industrial Marketing Management, 37, 120-130.
理查兹，K. A.和琼斯，E.（2008年）。客户关系管理：寻找价值驱动因素。工业营销管理， 37， 120-130.
Rigby, D. K., & Ledingham, D. (2004). CRM done right. Harvard Business Review, 82(11), 118-129.
Rigby， D. K.， & Ledingham， D. （2004年）。正确完成 CRM。哈佛商业评论，82（11），118-129。
Ryals, L., & Knox, S. (2001). Cross-functional issues in the implementation of relationship marketing through customer relationship management. European Management Journal, 19(5), 534-542.
Ryals， L. 和 Knox， S. （2001）。通过客户关系管理实施关系营销中的跨职能问题。欧洲管理杂志，19（5），534-542。
Scott, D. (2001). Understanding Organizational Evolution: Its impact on management and performance. Quorum Books.
斯科特，D.（2001 年）。了解组织演变：它对管理和绩效的影响。Quorum Books。
Tsai, C.-F., Hu, Y.-H., & Lu, Y.-H. (2015). Customer segmentation issues and strategies for an automobile dealership with two clustering techniques. Expert Systems, $32(1)$$32(1)$32(1)32(1).
蔡 C.-F.， 胡， Y.-H.， & 卢， Y.-H.（2015）. 具有两种聚类技术的汽车经销商的客户细分问题和策略。Expert Systems、 $32(1)$$32(1)$32(1)32(1) .
Windah1, C., Andersson, P., Berggren, C., & Nehler, C. (2004). Manufacturing firms and integrated solutions: Characteristics and implications. European Journal of Innovation Management, 7(3), 218-228.
Windah1， C.， Andersson， P.， Berggren， C.， & Nehler， C. （2004年）。制造企业和综合解决方案：特征和影响。欧洲创新管理杂志，7（3），218-228。
Wise, R., & Baumgartner, P. (1999). Go downstream the new profit imperative in manufacturing. Harvard Business Review, 133-142.
Wise， R.， & Baumgartner， P. （1999年）。顺流而下，推动制造业新的利润需求。哈佛商业评论，133-142。
Zhao, Z. Z., & Balague’, C. (2015). Designing branded mobile apps: Fundamentals and recommendations. Business Horizons, 58, 305-315.
赵，Z. Z.和巴拉格'，C.（2015）。设计品牌移动应用程序：基础知识和建议。商业视野，58,305-315。