明日dmp一站式营销数据

明日DMP基于隐私安全计算技术，配合中国数安港的安全管理措施，严格做到了数据可用不可见，解决了各方参与数据流通和应用的后顾之忧，打造了一个安全应用底座；同时搭建了基于各方数据联合，孵化新应用的平台产品。 明日DMP管理近10亿月活设备的行为数据，日处理超过50亿广告数据，日交互请求超过100亿。基于以上海量数据，明日DMP 利用数据挖掘和机器学习技术，形成对消费者的洞察、分类及应用，建立了9大类超过1500个标签，包括人口属性、人生阶段、兴趣爱好等。 线上数据和交易转化数据打通是数字营销的一大难题， 明日DMP以安全为基础，使得转化数据得以流通，开创了以下应用场景： 1.基于转化效果的高潜人群挖掘，分析已有转化人群的特征，利用机器学习算法在海量设备中找到与之有类似行为的人群，进行广告投放，进一步优化广告的转化效果，在某快消类广告投放上转化率提升近一倍； 2.基于已有的转化效果对媒体效果进行归因分析，帮助品牌规划广告预算投入，在某母婴客户上节约预算投入约20%；同时明日DMP也完成了跟市场主流的广告投放系统、广告监测系统的对接，服务客户覆盖各个行业，帮助客户管理投放的营销预算达数十亿。
我司在收集用户数据的过程中，将第一时间对收集后的数据进行去标识化或匿名化处理，用户数据第一次上报，通过系统的注册服务，结合设备标识与APP标识，根据算法加工生成"个推(母公司)匿名设备级标识符",通过个推(母公司)匿名设备级标识符与业务数据关联，在后续的业务使用过程中完全使用该标识符作为数据完整生命周期的标识。 这些数据经 筛选、清洗、整理 并经深度挖掘后进入公司“个推大数据平台”，根据隐 私政策所声明的目的、方式、范围等的约束，进行标签的加工。加工过程会根据标签类型有所区分，如人口属性、消费水平、美妆个护等常规标签通过机器学习模型构建，其中“人口属性-性别”标签，通过逻辑回归模型对用户使用设备、用户浏览广告类型、用户偏好APP等要素机器学习训练后得出评分（Score），分值大于0.6则判断为男性，小于0.4则判断为女性）

Tomorrow DMP, based on privacy-preserving secure computing technologies and in conjunction with the security management measures of China Data Security Port, strictly implements the principle of 'data available but not visible', eliminating the worries of all parties involved in data circulation and application, and building a secure application foundation. Meanwhile, it has developed a platform product that jointly utilizes data from all parties to incubate new applications.
Tomorrow DMP manages behavioral data of nearly 1 billion monthly active devices, processes over 5 billion pieces of advertising data per day, and handles more than 10 billion interactive requests daily. Leveraging data mining and machine learning technologies based on the above massive data, it forms consumer insights, classification and application scenarios, and has established over 1,500 tags across 9 major categories, including demographic attributes, life stages, interests and hobbies, etc.
The integration of online data and transaction conversion data has long been a major challenge in digital marketing. Based on security, Tomorrow DMP enables the circulation of conversion data, and has created the following application scenarios:
1. High-potential customer mining based on conversion effects: analyze the characteristics of existing converted customers, use machine learning algorithms to find groups with similar behaviors among massive devices for advertising delivery, further optimize the advertising conversion effect, and the conversion rate has nearly doubled in a certain fast-moving consumer goods advertising campaign;
2. Media effect attribution analysis based on existing conversion effects: help brands plan advertising budget allocation, and saved approximately 20% of budget for a certain maternal and infant customer. Meanwhile, Tomorrow DMP has completed docking with mainstream advertising delivery systems and advertising monitoring systems in the market, serving customers across various industries, and helping clients manage billions of dollars in marketing advertising budgets.
During the process of collecting user data, our company will immediately perform de-identification or anonymization processing on the collected data. When user data is reported for the first time, through the system's registration service, combined with device identifiers and APP identifiers, the "GeTui (Parent Company) Anonymous Device-level Identifier" is generated via algorithmic processing. This identifier is used to associate with business data, and will be fully utilized as the identifier for the entire data lifecycle in subsequent business operations.
These data are screened, cleaned, organized and deeply mined before entering the company's "GeTui Big Data Platform", where tag processing is carried out in accordance with the purposes, methods and scope stated in the privacy policy. The processing process varies according to the tag type: conventional tags such as demographic attributes, consumption levels, beauty and personal care are built through machine learning models. Take the "Demographic Attribute - Gender" tag as an example: a score is obtained through machine learning training on factors such as user device usage, advertising browsing types, and preferred APPs using a logistic regression model. A score greater than 0.6 is judged as male, and a score less than 0.4 is judged as female.

搜集汇总

数据集介绍

特点

该数据集由浙江明日数据智能有限公司提供，包含501条营销相关数据，每周更新。数据通过协议获得，应用于数字营销场景，如高潜人群挖掘和广告预算优化。数据处理采用去标识化或匿名化技术，并通过机器学习模型构建多种标签。