Preparation before access

In this section, you need to achieve the following objectives:

• Understanding what user behavior analysis is
• Understanding concepts such as events, users, and attributes in the field of user behavior analysis

What is user behavior analysis?

In your daily work, you may encounter the following problems:

• Are the newly launched product features being used by users daily? Has the conversion rate improved after the redesigned order page?
• How is user engagement with the newly launched campaign? At what stage do users churn?
• How many users clicked on the ads placed through these channels? Did any of these users subsequently register on the landing page?

To answer the above questions, we need to pay attention to various user behaviors on our product and conduct statistical analysis of these behaviors. Taking the first question as an example, the behaviors we need to focus on include: clicking on function entry points, entering new function pages, and clicking target buttons. If we statistically analyze these behaviors, we will obtain the following metrics.

If you were to analyze behavior, you might ask the following questions:

• What is the conversion rate between each step of the three actions?
• Do users who access a new feature through different entry points have different conversion rates?

The above is a small case study on user behavior analysis, demonstrating the three steps we need to take to conduct behavior analysis:

1. Raise business issues
2. Define the objects of analysis for the problem, specifically which behaviors.
3. Statistical analysis of behavior

In fact, we intentionally omitted the collection of behavioral data between steps 2 and 3. This task involves a very wide range of aspects and is highly complex. We will break down this task and introduce it step by step later.

Before starting data collection, it is necessary to understand how Sensors Data uses data to describe user behavior, which will facilitate subsequent work.

How to describe user behavior

In Sensors Analytics, we use the "Event Model" to describe various user behaviors. The Event Model includes two core entities: Event and User.

Why can combining these two entities clearly describe user behavior?

In fact, when describing user behavior, we often only need to clearly describe a few key points to describe the entire behavior. These key points include: who, when, where, how, and what they did.

This purpose can be achieved by combining the entities of Event and User. These two entities will be introduced separately below.

Event Entity

A complete event includes the following key elements:

Who: This refers to the user who participated in this event.

When: refers to the actual time when the event occurred.

Where: the location where the event occurred.

How: This refers to the way the user engages in this event. This concept is quite broad, including the device the user uses, the browser used, the app version used, the operating system version, the access channel, and the referrer when redirected. Currently, Sensors Analytics has the following pre-defined fields to describe this type of information, and users can also add corresponding custom fields according to their needs.

 $app_version：应用版本
$city：城市
$manufacturer：设备制造商，字符串类型，如"Apple"
$model：设备型号，字符串类型，如"iphone6"
$os：操作系统，字符串类型，如"iOS"
$os_version：操作系统版本，字符串类型，如"8.1.1"
$screen_height：屏幕高度，数字类型，如 1920
$screen_width：屏幕宽度，数字类型，如 1080
$wifi：是否 WIFI，BOOL 类型，如 true

What: Records the specific details of user-performed events using fields. Different events require different information; some typical examples are given below:

For a "purchase" type event, the fields that may need to be recorded include: product name, product type, quantity purchased, purchase amount, payment method, etc.

For a "search" type event, the fields that may need to be recorded include: search keywords, search type, etc.

For a "click" type event, the fields that may need to be recorded include: click URL, click title, click location, etc.

For an event of type "user registration", the fields that may need to be recorded include: registration channel, registration invitation code, etc.

For an event of the type "user complaint", the fields that may need to be recorded include: complaint content, complaint recipient, complaint channel, complaint method, etc.

For an event of type "return request", the fields that may need to be recorded include: return amount, reason for return, and return method.

Any field describing an event is an event attribute. Which events to collect, and which event attributes to collect for each event, depends entirely on the product type and analysis requirements.

User entity

Each User entity corresponds to a real user, and each user has various attributes. Common attributes include age and gender, while business-related attributes may include membership level, current points, number of friends, etc. These fields describing the user are called user attributes.

Summarize

This section uses an example to fully illustrate the value of user behavior analysis and explains "how Sensors Analytics uses data to record user behavior".

Subsequent chapters will answer the question, "How does Sensors Data collect user behavior data?"

We break this problem down into three sub-problems. First, we need to identify who the "parties" involved in the behavior are, i.e., Who, the first key factor in the Event entity. See step 2: How to accurately identify users .

Secondly, business personnel need to clearly define what data to collect and figure out where to embed what kind of data points. For related information, please refer to step 3: Data Collection Scheme Design .

Then, the developers implement code tracking based on the "Event Design" document provided by the business personnel. For details, please refer to Step 4: Basic Data Validation .