视频直播是当下电子商务领域最流行的营销方式之一。随着视频直播营销的不断普及，企业积极尝试将此种营销方式用于房产营销，以达到扩大业务边界和提升经营效益的目的。虽然业内针对房产营销已存在具备一定成熟度的客户管理系统，但由于视频直播模式重塑了房产营销的业务流程，传统房产营销客户管理系统无法满足视频直播房产营销的个性化需求，这一现状造成了视频直播平台房产营销业务效率低下、客户满意度低的困境，客户管理系统的建设和集成仍然是研究的难点。此外，由于视频直播房产营销的服务者不能预知客户的购房意愿，从而不能提供针对性服务，导致服务无法与客户需求匹配、客户购房转化低等问题，而业内针对视频直播房产营销场景下，客户购房意愿的预测仍缺乏聚焦性研究。因此，本文以提升视频直播房产营销业务运营效率和服务质量、扩大企业经营范围为目标，聚焦客户管理系统建设和客户购房意愿预测，提出针对性解决方案。论文的主要内容包括：（1）针对视频直播房产营销业务缺乏专门的客户管理方案和工具的问题，设计了一个客户管理系统，并对系统应用效果进行评估。本文针对系统设计过程中面临的业务流程不完善、管理工具不足的问题，构建了基于软件即服务模式的客户管理系统，设计了核心业务流程、系统服务架构和主要功能模块，实现了多角色登录、路由及权限校验、客户建联及分配等重点功能。通过对比系统上线前后的运行数据和利用模糊层次分析法，综合定量与定性的方法，验证了该系统对提升视频直播房产营销业务运营效率、服务质量以及扩大企业经营范围都有积极作用。（2）针对视频直播房产营销过程中服务者无法预知客户购房意愿来提供针对性服务的问题，构建了一个客户购房意愿预测模型。首先，基于消费者行为分析模型，分析了视频直播场景下客户的关键购房行为及影响客户购房意愿的因素，建立模型指标体系。然后，基于极端梯度提升算法构建了一个客户购房意愿预测模型，通过对客户管理系统收集的原始数据进行清洗、处理，进行模型训练和效果评价。最后，将模型应用于实际业务场景并给出应用效果和运营调整策略。实验结果证明，该预测模型有较高的准确性，表明基于该模型提出的房产营销运营策略对提升客户转化率、企业营销效率以及客户服务效率均产生了积极作用。

Live streaming is one of the most popular marketing methods in the current e-commerce industry. With the continuous popularity of live streaming marketing, enterprises actively attempt to apply this marketing method to real estate marketing in order to expand business boundaries and improve operational efficiency. Although there are mature customer management systems in the real estate marketing industry, the reshaping of business processes by live streaming marketing renders traditional customer management systems inadequate to meet the personalized needs of live streaming real estate marketing. This situation has resulted in low efficiency and customer satisfaction in real estate marketing on live streaming platforms, and the construction and integration of customer management systems remain challenging research areas. Additionally, service providers in live streaming real estate marketing cannot predict customers‘ purchasing intentions, thus unable to provide targeted services, leading to mismatches between services and customer demands, and low conversion rates. However, focused research on predicting customer purchasing intentions in the context of live streaming real estate marketing is still lacking. Therefore, this paper aims to improve the operational efficiency and service quality of live streaming real estate marketing businesses, as well as to expand enterprise operations. It focuses on the construction of customer management systems and the prediction of customer purchasing intention, proposing targeted solutions. The main contents of the paper include:(1) Addressing the lack of specialized customer management solutions and tools for live streaming real estate marketing, a customer management system was designed, and the effectiveness of its application was evaluated. In response to the imperfect business processes and inadequate management tools encountered during the system design process, this paper constructed a software-as-a-service-based customer management system. It designed core business processes, system service architecture, and major functional modules, implementing key functions such as multi-role login, routing and permission verification, and customer connection and distribution. By comparing pre-and post-launch operational data of the system and utilizing the fuzzy analytic hierarchy process, a comprehensive quantitative and qualitative method, it was verified that the system has a positive impact on improving the operational efficiency and service quality of live streaming real estate marketing businesses, as well as expanding enterprise operations.(2) To address the issue of service providers in live streaming real estate marketing being unable to predict customer purchasing intentions for tailored services, a customer purchasing intention prediction model was developed. Firstly, based on a consumer behavior analysis model, the key purchasing behaviors of customers in the live streaming scenario and the factors influencing their purchasing intentions were analyzed, establishing a model indicator system. Then, using the Extreme Gradient Boosting algorithm, a customer purchasing intention prediction model was constructed. The model was trained and evaluated using raw data collected from the customer management system, after cleaning and processing. Finally, the model was applied to real business scenarios, and the application effects and operational adjustment strategies were provided. Experimental results demonstrate that the prediction model exhibits high accuracy, indicating that the real estate marketing operational strategies proposed based on this model have a positive effect on improving customer conversion rates, enterprise marketing efficiency, and customer service efficiency.