Understand the collection, analysis and application of "user behavior data" in one article

More and more companies have begun to explore the commercial value of user behavior data and use behavioral data for precise and effective digital marketing . Taking the technology finance industry as an example, data from a well-known company shows that user behavior data is four times more effective than financial data.

1. Data sources of enterprises

The purpose of enterprises collecting, storing and analyzing data is to solve business needs, optimize business operation processes, improve their operating efficiency and reduce costs. Through data mining, in-depth analysis and visualization, enterprise business data can fully discover problems in business operations, and then formulate more scientific and reasonable operation strategies to realize the value of data.

Enterprises have three types of data:

Internal transaction data

Interaction data between enterprises and users

Third-party data, or external data

In the past, most of a company’s data assets were built on transaction data, using user attributes, sales data, logistics data, internal processes and other data to build data assets and carry out commercial applications. With the advent of the "user era", the scale and flexibility of data, as well as the ability to collect and use data, will determine the core competitiveness of an enterprise.

By mining and collecting business information, companies can predict market demand and conduct intelligent decision-making analysis to develop more effective strategies. In addition, they can optimize operations with user data and business data as the core, improve product performance, optimize operational efficiency, and conduct precision marketing through user portraits , market and channel analysis, and sales data analysis.

2. Collection and Analysis of Behavioral Data

User behavior data: mainly includes users' browsing/clicking/posting behaviors on websites and mobile apps. Behavioral data actually has great commercial value, but many companies do not know how to apply it. User behavior data is basically collected using SDK, which collects the user's click behavior on the page and can also return parameters. The SDK is just a few lines of lightweight code, and the type of data collected depends on the tracking point. SDK has no technical barriers in data collection. The main technical barrier to behavioral data application lies in the processing and analysis of massive behavioral data.

1. What is the privacy of data collected by SDK?

Many companies always believe that data collected by SDK will involve personal privacy, which is mainly because they do not understand the technical principles of SDK data collection.

SDK, Software Development Kit , literally means software development kit, which uses N lines of software code to collect data. Any data collected by the SDK comes from the user's subjective behavior. The personal privacy data obtained by enterprises in normal business activities does not violate regulations. It is illegal for personal privacy data to be used by enterprises and third parties without the user's authorization.

2. How difficult is data processing and analysis?

The processing and analysis of user behavior data has a high technical threshold:

The SDK will collect a large amount of "dirty data", including some blank areas and special symbols, and even data types that have never been seen before. The processing and analysis of these dirty data pose great technical challenges, especially the real-time collection and processing of data. Usually, technical personnel can form a mature technical architecture only after they have experienced massive data collection and processing and filled a large number of "technical pits".

Data collection and processing is a dirty and tiring job that needs to be carried out in a real data environment and has high technical barriers and thresholds.

3. Are you just giving up like this ?

In order to lower the technical barriers to data collection and processing and help enterprises collect data accurately and efficiently, Zhuge.io has summarized a series of successful experiences from a large number of accumulated business practices for you to directly "borrow" and cross this technological gap. During the data collection phase, Zhuge.io will escort you throughout the entire process, from sorting out data analysis needs, organizing point documentation, to the final technical execution, and guide you step by step to take the first step in data analysis.

3. Commercial Value of User Behavior Data

In order to ensure that the user's product usage process is smooth, product design from the user's perspective requires close attention to user feedback and needs. By observing user behavior data or directly communicating with users to obtain this feedback, we can find out where users are stuck or make mistakes. Only in this way can we polish out the best user experience path. This is the value of user behavior data. Before behavioral data can be valuable, it needs to be structured and labeled:

Structuring refers to converting the presentation of behavioral data from unstructured data to structured data, and classifying and counting them;

Labeling refers to labeling behavioral data with business tags based on business scenarios, focusing on devices and deeply integrating with business scenarios.

When labeling behavioral data, there are usually three data dimensions: time, frequency, and results.

1. Time

The time dimension of behavioral data focuses on the time period and duration of the behavior. The time period data is used to select the time range of the target device, and is used for marketing campaign analysis and marketing promotion plan setting. Time periods can also be used in risk control and anti-fraud scenarios. The app usage behaviors of special groups have high similarity in time periods. Duration focuses on the process of the behavior and records the start and end time of the behavior.

All session records of a user

Description: A full-view user portrait, including the user's access time period, access duration, and even the time when the user initiates and ends a session.

Duration is of great significance for analyzing user behavior. Different lengths of time represent different characteristics of users. Based on the user's life cycle, we can gain insight into the interaction status between users and products. It has high commercial value in some data model analyses and can be used for purchasing population analysis, product experience analysis, and even anti-fraud analysis.

2. Frequency

The frequency of behavioral data mainly focuses on the number of times and trends of certain specific behaviors. The number of times has a large positive correlation with the user's interest. Within a certain period of time, the number of clicks and views is proportional to the user's purchasing demand. After being labeled, the times can be used for marketing and identifying potential users.

In addition, through page click analysis, we can understand product experience and user needs, thereby optimizing product layout and selling related products. The number of times is weakly correlated with product transactions and user purchasing demand, but combined with trend data such as the number of clicks and views, these data can reflect product conversion and user purchasing behavior.

For example, if a user suddenly logs in to automotive products frequently within a certain period of time, trend analysis can predict the user's purchasing needs. In some scenarios, trend data has higher commercial value than frequency data and can directly predict customer purchasing needs.

3. Results

The results of behavioral data mainly focus on whether the transaction is completed and are used to determine the results of user clicks and browsing. The result data is divided into transaction and non-transaction. Based on business needs, the filled values ​​can also be collected for further application.

Transaction data can be used for product experience analysis, user experience analysis, and channel ROI analysis;

The data on non-transactions can be used for secondary marketing to re-market to potential users. A comprehensive analysis can be conducted based on the time period, duration, and frequency data to screen out the target customer group. In addition, by combining transaction data and time data, after locking in product problems, we can more accurately optimize the product experience and analyze the conversion funnel.

The resulting data can be used for direct marketing and can be added to the data model as reference data for an important dimension.

Example scenario:

Screen out users who prefer long-term financial products

Note: By setting conditions, users who have searched and viewed the categories and details pages of long-term financial products more than 3 times in the past 30 days are screened out and defined as users who have a preference for long-term financial products. Secondary marketing for potential users can be carried out for this customer group, such as pushing interest rate coupons for long-term financial products to them to encourage them to complete their investment .

4. Scenario-based Application of Behavioral Data

Starting from business needs (business scenarios), finding behavioral data that is highly relevant to them is one of the ideas for establishing scenario-based behavioral data labels, and analyzing the transaction path (transaction steps) of a certain business in the product. In the first few steps approaching the transaction path, establish scenario-based labels based on time, frequency, and results.

Marketing based on behavioral data needs to focus on measuring marketing effectiveness and iterative optimization of marketing plans. Through multiple marketing attempts, a more appropriate way to establish behavioral labels can be found, and the frequency, time period, results and other values ​​can be determined. A stable operation plan and operation plan can be gradually established. Some fixed operation plans can be solidified on a certain day of the week or even a certain time period to form a fixed operation plan.

The key to marketing success lies in continuous experimentation, optimizing the various data dimensions and values ​​in scenario-based tags, and solidifying the plan when the expected effect is achieved to form a standard operation plan.

Example scenario:

1. Promote the conversion of first orders: continuously reduce the time from new addition to first investment

Taking the first-order conversion scenario as an example, a condition is set for new users: if the account opening behavior is not completed within one day after successful registration, a text message/mobile PUSH will be sent. After completing the operation, the effect can be automatically measured, and the number of users who have achieved "first-order investment" within 3 days after the operation is executed can be counted, and the absolute number/ conversion rate /transaction amount can be analyzed.

2. Promote additional investment: guide additional assets to promote the migration of user value levels

For active users in the novice period, operations can be completed through manual settings: filter out users with assets less than 50,000 and no investment behavior in the past 30 days, accurately push cash coupons to them, and check the effect through automatic measurement: the number of users who have used cash coupons to invest within 3 days, and automatically analyze the absolute number/conversion rate/transaction amount.

Based on user behavior data and with users at the center, all functional experiences should revolve around user needs and user perceptions, so as to improve user satisfaction, and then the improvement of conversion rate will be a natural result.

The author of this article @朱葛io compiled and published by (Qinggua Media). Please indicate the author information and source when reprinting!

Product promotion services: APP promotion services, information flow advertising, advertising platform